View original document

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

June 2016

Are you being taken for a phool?
Phishing for Phools: The Economics of Manipulation and
Deception. By George A. Akerlof and Robert J. Schiller.
Princeton, NJ: Princeton University Press, 2015, 288 pp.,
$24.95 hardback.
Take a stroll with me down memory lane, and let’s
remember our economics 101 class about decisionmaking
and budget quandaries. If we were to go to the supermarket
with a fixed budget to buy strawberries and blackberries,
our goal would be to obtain a combination that makes us
happiest (maximizing our utility function). This strategy is
part of rational choice theory, according to which a person
is believed to always be making logical choices that provide
him or her the most satisfaction. However, there is a
tendency to overlook irrational behavior in mainstream
economic theory. What if I do not stick to that budget?
Rather than doing what standard economics teaches us,
suppose I pay with a credit card and exceed my fixed
budget. In this scenario, the credit card industry exists as
long as there is a profit to be made from providing credit
because enough people demand to go beyond their budget.
In the book Phishing for Phools: The Economics of
Manipulation and Deception, George Akerlof and Robert
Schiller dive deeper into the realm of behavioral economics.
They make a case for why free markets that provide people
what they want (as long as there are incentives in place) at
an equilibrium level can be manipulated or distorted,
thereby creating a new equilibrium that they call a “phishing
equilibrium.”

Richard Hernandez
hernandez.richard@bls.gov
Richard Hernandez is an economist in the Office
of Publications, U.S. Bureau of Labor Statistics.

Let’s start by defining the relevant terms found throughout
the book. A phish is defined as the means by which a
phisherman (the agent performing the phish) gets his or her
target to do what the phisherman wants. (The phishing discussed in this book is not to be confused with that in the
field of information and computer technology, whereby individuals attempt to acquire sensitive information, such as
Social Security numbers and passwords, or even money by masquerading as a trustworthy source in an electronic

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

communication environment.) A phool is someone who has been successfully phished. There are two types of
phools. The psychological phool can be phished by one of two methods: either by having a cognitive bias that is
exploited or by giving in to emotions despite an awareness of the situation at hand. The information phool acts on
facts that are purposely intended to be misleading. The authors also delve into four areas in which they believe
that “NOBODY-COULD-POSSIBLY-WANT” to be phooled—areas where it makes no sense not to have optimal
outcomes: our health, the quality of our government, market stability, and personal financial security. With all these
terms, the authors relate stories in many settings, ranging from consumer and financial markets to congressional
elections, to prove their theory of phishing equilibrium.
Taking us on a discovery ride into advertising, the authors provide ample examples from the minds of advertising
“gurus” such as Albert Lasker and David Ogilvy. Crafting the right story and creating an accurate message that
leads to customer engagement with a product is something marketers are able to master. So, how do they come to
sell you that product? Simply by finding out what works and what doesn’t, using trial-and-error statistical tests. Did
you ever wonder why advertisements provide different redemption codes? The answer is that there is no better
way to target an advertiser’s audience than by testing which codes work and which do not. If the redemption code
for a product touted in advertisement A was redeemed more often than the code for a product extolled in
advertisement B, then a logical conclusion would be to run only advertisement A in the future. This consideration is
one of many aspects of phishing equilibrium, meaning that, if there is a way to make a profit from our tastes, then
the phisherman will keep trying until he or she finds it. In our era of big data, marketers have become increasingly
knowledgeable about our preferences and are better capable of exploiting them. They are getting better and better
at playing to our human nature of wanting a product rather than needing it. At its core, the book is trying to flesh
out the idea that there is a narrative in our minds which leads us to make irrational decisions—an idea that
standard economics misses. Akerlof and Schiller tie it back to the strawberries-and-blackberries example: Say the
blackberry marketers crafted the narrative that blackberries are superior in taste to strawberries. Then, even
though we wouldn’t be maximizing our utility by purchasing only blackberries, we end up doing so because the
monkey-on-the-shoulder tastes created by advertisers establish a new market equilibrium. In this regard, the
authors conclude that free markets allow people to choose between their “real tastes” and “monkey-on-theshoulder-tastes,” and then people are freely available to be phished.
Certainly, Akerlof and Schiller do not attack the free market; rather, they argue that free markets have systemic
flaws. They maintain that free markets are still the best economic means of raising living standards for all, but
there are some unwanted externalities that the phishermen take advantage of in plying their “trade.” For example,
deregulation of the banking industry led to the savings-and-loan crisis of 1986–95, and that crisis in turn brought
about the recession of 1990–91. In that scenario, the externality was the inflation-adjusted cost of $230 billion on
the backs of taxpayers that was caused by the failure of the savings and loans, which became “tools for the
phishermen.”
The book offers various examples of phools being phished in many settings in a very easy-to-read way. With
regard to phishing equilibrium, one may ask what the authors consider to be fair or unfair in the marketplace. The
book does not define or explore what constitutes fair or wanted outcomes. Instead, it leaves the reader wanting
resolutions of some issues. For example, is merely having an awareness that there are deficiencies in the market
enough for people to modify their behavior so that they don’t get taken for a phool? How do we go about solving
the problem of companies getting around certain legislation? Moreover, does it really matter if I buy more
blackberries than strawberries? I am the only person affected by that decision. However, in a different scenario,

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

one in which the player learns to phish in a way that affects us all (e.g., by crashing the financial markets),
minimizing market inefficiencies does matter. And that is where the real lesson of this book comes into play: we
should always be aware that we can get phished, and we must find ways to minimize that possibility.

3

June 2016

Consumer Expenditure Survey Methods
Symposium and Microdata Users’ Workshop, 2015
This report describes the fourth annual Consumer
Expenditure Survey (CE) Survey Methods Symposium,
which took place on July 14, 2015, and the 10th annual CE
Microdata Users’ Workshop, which took place on July 15–
17, 2015.
The CE is the most detailed source of expenditures,
demographics, and income collected by the federal
government. Every year, the Bureau of Labor Statistics
(BLS) CE program releases microdata on the CE website
from its two component surveys (the Quarterly Interview
Survey and the Diary Survey), which are used by
researchers in a variety of fields, including academia,
government, market research, and other private industry
Geoffrey D. Paulin
paulin.geoffrey@bls.gov

areas.1
In July 2006, the CE program office conducted the first in a
series of annual workshops to (1) help users better
understand the structure of the CE microdata; (2) provide

Geoffrey D. Paulin is a senior economist in the
Consumer Expenditure Survey Program, U.S.
Bureau of Labor Statistics.

training in the uses of the survey data; and (3) promote
awareness of the different ways in which the data are used
and explore possibilities for collaboration through
presentations by current users and interactive forums.

Nhien To
to.nhien@bls.gov
Nhien To is an economist in the Consumer
Expenditure Survey Program, U.S. Bureau of
Labor Statistics.

Starting in 2012, the program office added an additional
day to the event for a symposium to explore topics in
survey methods research in support of the Gemini Project,
a major effort to redesign the CE survey (more information
here: https://www.bls.gov/cex/geminiproject.htm).

In addition to the CE program staff, workshop speakers have included economists from BLS regional offices and
researchers not affiliated with the BLS; similarly, symposium speakers have included CE program staff, other BLS
National Office staff, and speakers from outside the BLS.

Survey methods symposium

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

As in previous years, the goals of the 2015 CE Survey Methods Symposium were (1) to provide an update on the
status of the Gemini Project, including results from recent projects; and (2) to feature research related to the
redesign, including imputation of assets and liabilities, testing the feasibility of collecting outlet data with
expenditures, and building a respondent burden index. There were two sessions, one focusing on each topic.

Gemini Project to redesign the CE
Overview of the Gemini Project. As a continuation of the work presented at the 2013 and 2014 CE Survey Methods
Symposium, Laura Erhard (CE) provided an overview of the multiyear Gemini Project, which was launched in 2009
to research and develop a redesign for the CE. Much research has been conducted, including recommendations
from the National Academies’ Committee on National Statistics (CNSTAT). Erhard shared an illustration of the
2013 survey redesign plan and explained the details of the new protocol that were being assessed in the Proof of
Concept (POC) Test from July 2015–October 2015. The objectives of the POC Test were to address
methodological issues and evaluate operational and experiential factors.
Following the POC Test, CE will be conducting two additional major field tests: an Incentives and Outlets Test in
2016 and a Large-Scale Feasibility Test in 2019. The Incentives and Outlets Test will be conducted in the CE
production sample, follow the proposed Gemini Redesign structure of incentives, and look at both operational
issues and effects on data quality and response rates. The Large-Scale Feasibility Test will include all the
components of the redesign and incorporate lessons learned. It will be conducted much like the POC Test, but with
a larger sample, and the instrument will closely resemble the final design.
Results from the Individual Diaries Feasibility Test. Ian Elkin (CE) shared preliminary results from the Individual
Diaries Feasibility Test (IDFT), which was designed to inform the operational and data quality aspects of collecting
expenditures from household members using personal electronic diaries. The IDFT tested two separate web
instruments: a mobile version (via smartphone) and a desktop version. The test was fielded from August 2014–
December 2014, and the sample targeted area mobile usage, Internet penetration, multiperson households, and
English-speaking households (as a Spanish diary was not developed to reduce costs, although one will be
developed for production and included in the Large-Scale Feasibility Test).
The preliminary results reinforced the benefits of electronic diaries from a performance and data quality standpoint.
For example, diary keeping information that is automatically generated allows for protocols to be updated more
rapidly, electronic diaries allow for the collection of data from paper diaries that are not successfully picked up, and
access to up-to-date paradata allows for on-demand interviewer intervention. In addition, an examination of the
characteristics of respondents showed that electronic diaries, specifically mobile diaries, do an effective job of
targeting groups generally underrepresented in CE data.
Of note, development of CE’s electronic diaries is an ongoing, iterative process, along which the IDFT was one
milestone. In addition to substantive development milestones (e.g., instrument specifications), protocol milestones
also exist, a number of which were omitted from the IDFT fielding. These include burden-appropriate login
procedures, as well as specified monitoring and follow-up activities for field staff. Continued testing, into the LargeScale Feasibility Test, will bring together the cumulative efforts of substantive development and protocol
implementation.

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Progress on the CE Electronic Diary. In 2013, CE began the process of building a mobile survey instrument to
collect diary data. This work was done to build the instrument used in the IDFT, but extended past the IDFT into
developments of later designs. The process began with rough drawings of instrument screenshots that were
turned into formal written requirements and handed off to programmers. During the process, a series of usability
tests were conducted. Brandon Kopp (Office of Survey Methods Research, OSMR) oversaw and conducted the
usability tests and shared an overview of the process along with results and recommendations from the tests at the
symposium. Most of the tests focused on the mobile version of the diary, but a subset of tests also looked at the
desktop version.
The test results uncovered issues ranging from the login process and password change requirements to
respondent difficulty in entering data that met data requirements. Many changes were made to the instruments, but
because of limited time and resources, not all issues were addressed. Work to improve the electronic diary is
ongoing. There is a plan to propose alternative designs for the desktop version of the diary and conduct usability
testing. In addition, CE will be exploring alternate ways of simplifying the login process. Incremental changes will
be made up until (and after) it is being used in production.

Gemini Redesign related research
Overview of CE Research. Branch of Research and Program Development (BRPD) Chief Adam Safir began the
second session of the symposium with an overview of ongoing CE survey improvements since 2003 and
summarized the current research agenda. The agenda, which BRPD updates annually, focuses on research
issues within the context of the CE’s long-term goals. It also communicates CE’s research plans and priorities with
respect to the redesign and reflects discussions with internal and external stakeholders. The agenda isn’t set in
stone and can change over time as new research findings and questions emerge.
Investigating the Imputation of Assets and Liabilities in the CE Interview Survey. Geoffrey Paulin (CE) spoke about
the problem of nonresponse and how it affects the data collected on expenditures, income, taxes, and assets and
liabilities. While there are methods in place for handling this problem with most of these data, nonresponse to
questions that collect data on assets and liabilities is currently under investigation. The purpose of the project is to
design a method to impute missing Interview asset and liability data, leveraging models from income imputation
and other relevant procedures. The goal of the project is to implement this method into production with 2018
Quarter 2 data.
The team considered several methods, including: one used by the Survey of Consumer Finances, which utilizes an
iterative, multiple imputation process; regression trees; and hot deck, but none were feasible. Ultimately, the team
decided to investigate a system based on income imputation processing. Paulin described what that system entails
and discussed the challenges involved in adapting it to assets and liabilities. The project is a work in progress.
Testing the Feasibility of Integrating Outlets into the CE Diary. Currently, CE and TPOPS (Telephone Point-ofPurchase Survey) collect complementary and potentially redundant information for the CPI (Consumer Price
Index). As a result, the CE and CPI programs are interested in determining the impact on data quality and
respondent burden of collecting outlet data in the CE Diary. Erica Yu (OSMR) conducted a small study looking at
two ways of integrating outlets in the CE Diary and found that (1) the collection of outlet information did not
substantively affect CE data quality, (2) participant ratings of burden showed no large effects due to the addition of

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

outlets, and (3) there was a possible increase in time taken to enter items in the diary. This was a small study, and
further research is needed.
Developments in Building a Respondent Burden Index. Danny Yang (OSMR) discussed the work being done to
develop a composite burden index for CE that would track perceived respondent burden over time. This would
allow CE to detect and understand changes in burden following modifications to the survey, evaluate the
association between the survey burden index and other survey measures of interest, and develop interventions
that would reduce respondents’ perception of burden. Burden scores could be integrated into the overall
assessment of survey performance for CE management.
Conclusions: With many research projects underway involving the overall CE Gemini Project, the 2015 CE Survey
Methods Symposium was a successful event focused on sharing recently completed and current work with data
users and others interested in CE’s survey research. These research projects help the program move toward
achieving its overall redesign goals, and the symposium serves as a channel for discussion and the exchange of
ideas.
The symposium drew a little over 50 attendees from areas such as universities, academic programs in survey
methodology, nonprofit organizations, private companies, medical-related establishments, and other federal
agencies.

Microdata users’ workshop
Day one. The first day of the 2015 workshop opened with presenters from the CE program. Bill Passero provided
an overview of the CE, featuring topics such as data collection and publication. Brett Creech then presented an
introduction to the microdata, including an explanation of their features, including data file structure and variable
naming conventions.
The morning concluded with presentations by researchers not affiliated with the CE program who have used the
microdata for a variety of purposes. The first speaker, Stephen Brumbaugh, discussed automobile loans made to
low-income consumers. The second speaker, Taylor Smith, related spending patterns to changes in housing
wealth in recent years.
After the lunch break, CE economist Aaron Cobet described forthcoming changes in the Public Use Microdata
(PUMD) website, and solicited comments from the attendees. The rest of the afternoon was dedicated to practical
training, in which attendees had the opportunity to perform programming exercises using the microdata.
The day concluded with an information-sharing group session among workshop participants and CE program staff.
This was an open forum in which attendees met informally to discuss their research and offered suggestions for
improving the microdata. One recommendation was that the CE make information more readily available to users.
Specifically, the CE needed to find a better way of presenting the documentation, highlighting key topics, and
making online help tools more dynamic.
Day two. The second day opened with more advanced topics, with Brian Nix of the BLS Division of Price Statistical
Methods presenting technical details about sampling methods and construction of sample weights. Meaghan
Smith (CE) followed with a presentation on imputation and allocation of expenditure data in the CE.

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The remainder of the morning was dedicated to research presentations by non-BLS attendees. The first of these,
entitled “The 2011 Payroll Tax Cut and Household Spending: Evidence from a Quasi-Natural Experiment” (Naveen
Singhal), examined how expenditures made by consumers changed in different states in response to a cut in
payroll taxes. The experiment became possible because one state, Illinois, raised income taxes by about the same
percentage as the reduction in payroll taxes, yielding no net cut for residents of that state. The presenter of the
second work, entitled “Household Consumption Smoothing between Monthly Housing Payments” (Li Zhang), had
returned to the workshop for a second consecutive year, having presented a different paper (“The Effect of
Casinos on Household Consumption”) in 2014. The third presentation, entitled “Income-Expenditure Elasticities of
Less Healthy Consumption Goods” (Adam Hoffer), used data from the Diary Survey to analyze expenditures on
foods like cola and donuts to estimate how tax increases might affect expenditures on these goods.
After a break for lunch, Carol Boyd Leon and Charlotte Irby, technical writer-editors of the Monthly Labor Review
(MLR), described the publication process, from submission to printing, for attendees interested in having their work
appear in the MLR.
After this description of the MLR process, the technical instruction resumed with a presentation of a topic of
perennial interest to CE microdata users: how to apply longitudinal weights to the interview data. Following this
presentation, Evan Hubener (CE) led a discussion highlighting some of the limitations of the CE survey data and
provided best practices for dealing with weights under these circumstances. Hubener detailed how the Interview
Survey collects data from respondents for 4 consecutive calendar quarters. During each interview, the respondent
is asked to provide information on expenditures for various items during the past 3 months. However, not all
participants remain in the sample for all four of these interviews. Those who do remain have different
characteristics (e.g., higher rates of homeownership and average age) than those who do not. Therefore,
attempting to analyze average annual expenditures by only examining respondents who participate for all four
interviews yields biased results.
Following the Hubener presentation, the workshop pivoted to a session explaining an important feature of certain
variables in the microdata: topcoding. In a presentation entitled “Balancing Respondent Confidentiality and Data
User Needs,” Arcenis Rojas (CE) explained that, in order to preserve the confidentiality of the data, values for
some variables, such as income sources and certain expenditures (e.g., rent, among others), are topcoded. In this
process, values that exceed a predetermined critical value are replaced with a new value. In each case, changed
values are flagged for user identification.2 At the conclusion of this presentation, practical training resumed for the
rest of the afternoon.
Day three. On the final day, CE staff featured advanced topics, starting with Barbara Johnson-Cox explaining how
sales taxes are applied to expenditure reports during the data production process. Next, Geoffrey Paulin described
the correct use of imputed income data and sample weights in computing population estimates. The latter session
noted that the proper use of weights requires a special technique to account for sample design effects that, if not
employed, result in estimates of variances and regression parameters that are incorrect.3 Researcher Walter Lake
(Pew Charitable Trusts) followed, describing a user-friendly tool he was developing to allow researchers to obtain
time-series estimates from microdata both for demographic groups and detailed expenditures not available in
online formats through the CE website.4 After a break, Aaron Cobet described the new methods in CE for
estimating income taxes paid by consumer units, the amounts for which replace those reported by consumers
during their interviews, as these data have been found to be extremely unreliable.5 The CE uses the National

5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Bureau of Economic Research TAXSIM program to estimate federal and state income taxes. These new estimates
were introduced with the publication of the 2013 annual tables. They represent a major improvement to the quality
of the CE after-tax income data. The session concluded with a “sneak peek” of developments for CE microdata by
Steve Henderson. In 2015, there were many changes made to the Interview Survey. These included the
introduction of new health care questions, the dropping of the first interview or “bounding interview,”6 and the
implementation of a redesigned sample.7 Regarding publications, Henderson noted that detailed data tables,
which had been available only on request, would be published online, starting with one at the all-consumer-unit
level. In addition, a new higher income table and a new table looking at spending by birth year of the reference
person, divided into generations, would be released.8
After a lunch break, practical training continued, including a presentation of a computer program available with the
microdata for use in computing correct standard errors for means and regression results when using (1)
unweighted nonimputed data, (2) population-weighted nonimputed data, and (3) multiply imputed income data,
both unweighted and population weighted (Paulin). Finally, attendees were debriefed in a feedback session
designed to solicit opinion on how to improve future workshops, CE program outreach, and other topics of interest
to attendees. Most of the suggestions were related to methods for raising awareness about future workshops.
Some users provided additional outlets for us to post information about the 2016 workshop. Users also mentioned
the need for additional sample programming codes.

2016 Symposium and workshop
The next Survey Methods Symposium will be held July 12, 2016, once again concomitant with the 11th annual
Microdata Users’ Workshop (July 13–15, 2016). While the symposium and workshop will remain free of charge to
all participants, advance registration is required. For more information about these and previous events, visit the
CE website (www.bls.gov/cex) and look for “Annual Workshop” under the left navigation bar titled “CE PUBLICUSE MICRODATA.” For direct access to this information, the link is www.bls.gov/cex/csxannualworkshop.htm.
Additional details about previous symposia are available at https://www.bls.gov/cex/ce_workshop_archive.htm.

Highlights of workshop presentations
Following are highlights of the papers presented during the workshop, listed in the order of presentation. They are
based on summaries written by the respective authors.
Stephen Brumbaugh, Ph.D. candidate, UCLA Department of Urban Planning, “Driven to Poverty? An Analysis of
Automobile Expenditures in Low-Income Households” (Interview Survey), day one.
Transportation is the second-largest expense category for American households after housing, and the
financial burdens of transportation for low-income households—in particular, the costs of buying, operating,
and maintaining a vehicle—are a prominent concern among policymakers and antipoverty advocates.
Nonetheless, few researchers have directly examined vehicle expenditures in low-income households. In my
dissertation, I attempt to fill this research gap by analyzing Consumer Expenditure Survey microdata. My
research is guided by three major questions: whether consumer characteristics like race and education
explain differences in vehicle expenditures among low-income households; how the nature of vehicle repair

6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

expenditures for these households has changed as automotive technology improves; and whether transit
expenditures explain differences in automobile expenditures.
Taylor Smith, Ph.D., Georgia Gwinnett College, “How Do Changes in Housing Wealth Affect Consumption
Behavior?” (Interview Survey), day one.
Between 1997 and 2006, the price of the typical American house increased 124 percent. This housing boom
and its resulting 2008 bust have been cited as major determinants of changes in household consumption over
this period. Using more than 12 years of consumer data merged with several macroeconomic time series, we
estimate the impacts of housing wealth on 13 specific expenditure categories and the overall budget formation
of Americans. We find that housing market fluctuations during this period were indeed a determinant of
consumption change, but only in certain sectors, and that the effects were smaller than some news media and
previous literature have suggested. Additionally, we show that effect magnitudes vary greatly across young
and old homeowners, and across the housing boom and bust periods.
Naveen Singhal, Ph.D. candidate, University of Illinois at Chicago, “The 2011 Payroll Tax Cut and Household
Spending: Evidence from a Quasi-Natural Experiment” (Interview Survey), day two.
In 2011, the federal government reduced the payroll tax rate from 6.2 to 4.2 percent, while at the same time
Illinois increased its state income tax rate from 3 to 5 percent. Consequently, Illinois workers were largely
unaffected by these tax changes, but workers elsewhere experienced an increase in their take-home income.
Using this variation in tax liability, I estimate that for every dollar of tax decrease, household spending
increased by about 89 cents, especially on recreation, dining, vacations, clothing, and personal care.
Additional analysis indicates that the estimates are unlikely to be biased from Illinois-specific shocks and may
therefore be interpreted causally.
Li Zhang, Ph.D. candidate, University of Virginia, “Household Consumption Smoothing between Monthly Housing
Payments” (Diary Survey), day two.
This paper studies consumption smoothing of households between monthly payments of mortgage or rent.
The paper’s focus on regular payments contrasts with most of the literature, which finds excess sensitivity to
regular receipt of income. Using the Consumer Expenditure Survey (CE) Diary Survey from 1998 to 2011, I
find that spending on nondurable goods is $3.34, or 9.0 percent higher per day during the two weeks following
the day when a housing payment occurs, compared with the two weeks prior to that day, which is inconsistent
with the consumption smoothing predicted by the life cycle/permanent income hypothesis. This finding is
robust to the coincident timing of households’ regular housing payments and their regular income arrivals, and
suggests that findings in the previous literature of excess sensitivity of consumption to regular income arrivals
may in part reflect excess sensitivity to the timing of making regular payments. The increase in biweekly
average spending following a housing payment day is larger for households in which the household head has
lower educational attainment, larger for households with lower income, and has a U-shaped profile in age of
household head. My finding is not fully consistent with existing theories that explain departures from
consumption smoothing between regular payments, including liquidity constraints and uncertainty about bank
account balances.

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Adam Hoffer, Ph.D., Assistant Professor, University of Wisconsin-La Crosse, “Income-Expenditure Elasticities of
Less Healthy Consumption Goods” (Diary Survey), day two.
There is a long-running policy debate regarding the use of tax policy to modify consumption choices and
health outcomes. Specifically, should taxes be imposed on “unhealthful” foods to discourage their
consumption and thereby reduce unhealthy outcomes? Objections to this policy include the positing that such
goods are price inelastic (i.e., purchases are not sensitive to changes in prices), so the imposition of taxes
(essentially equivalent to increasing prices) would be ineffective. This work examines expenditures for cola
and donuts, and finds that the expenditures are income inelastic. Therefore, to the extent that taxes reduce
income for purchasers of these goods (that is, if the goods cost more, purchasers have less income to allocate
to other goods and services), they do little to discourage consumption of these goods.
Walter Lake, Senior Associate, Research Financial Security and Mobility, Pew Charitable Trusts, “Introducing KIWI:
A Stata Package to Explore BLS Consumer Expenditure Data” (Interview Survey), day three.
The BLS Consumer Expenditure Survey Public Use Microdata (PUMD) are a very rich, multifaceted set of
data with a wealth of information surpassed only by the complexity of the procedures necessary to extract that
information. The technical knowledge required to assemble the data prior to analysis creates a barrier for all
but the most advanced users of statistical software packages. Lowering the barriers to entry will increase the
number of researchers from a variety of fields that can access and utilize the data. To facilitate this, I have
created an add-on package for STATA statistical software that streamlines the process for data aggregation
and variable creation. Through the use of a graphical user interface (GUI) with drop-down menus and
selection buttons, the user can assemble and analyze PUMD with just a few mouse clicks. The GUI allows the
user to weight the variables, run crosstabs, and output basic graphs. Two versions of the algorithm that
powers the GUI are available to accommodate different levels of statistical programming prowess. The STATA
is very functional but still a work in progress and should be ready for public release within the next year.

BLS Speakers
Staff of the CE Program
Cobet, Aaron. Senior Economist, Branch of Information and Analysis (BIA), days one, two, and three
Creech, Brett. Economist, BIA, day one
Curtin, Scott. Supervisory Economist, Chief, Microdata Section, BIA, day one
Henderson, Steve. Supervisory Economist, Chief, BIA, days one and three
Hubener, Evan. Economist, BIA, day two
Johnson-Cox, Barbara. Economist, Branch of Production and Control (P&C), day three
Passero, Bill. Supervisory Economist, Chief, Processing and Analysis Section, BIA, days one and two
Paulin, Geoffrey. Senior Economist, BIA, day three

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Rojas, Arcenis. Economist, BIA, days one and two
Smith, Meaghan. Supervisory Economist, Chief, Phase 3 Section, P&C, day two
Other BLS speakers
Boyd Leon, Carol. Technical Writer-Editor, Monthly Labor Review Branch, day two
Irby, Charlotte. Technical Writer-Editor, Monthly Labor Review Branch, day two
Nix, Brian. Mathematical Statistician, Division of Price Statistical Methods, day two
Speakers from outside BLS
Brumbaugh, Stephen, “Driven to Poverty? An Analysis of Automobile Expenditures in Low-Income
Households” (Interview Survey), day one
Hoffer, Adam, “Income-Expenditure Elasticities of Less Healthy Consumption Goods” (Diary Survey), day
two
Lake, Walter, “Introducing KIWI: A Stata Package to Explore BLS Consumer Expenditure
Data” (Interview Survey), day three
Singhal, Naveen, “The 2011 Payroll Tax Cut and Household Spending: Evidence from a Quasi-Natural
Experiment” (Interview Survey), day two
Smith, Taylor, “How Do Changes in Housing Wealth Affect Consumption Behavior?” (Interview Survey),
day one
Zhang, Li, “Household Consumption Smoothing between Monthly Housing Payments” (Diary Survey),
day two

SUGGESTED CITATION

Geoffrey D. Paulin and Nhien To, "Consumer Expenditure Survey Methods Symposium and Microdata Users’
Workshop, 2015," Monthly Labor Review, U.S. Bureau of Labor Statistics, June 2016, https://doi.org/10.21916/mlr.
2016.24.
NOTES
1 The Quarterly Interview Survey is designed to collect data on expenditures for big-ticket items (e.g., major appliances, cars, and
trucks) and recurring items (e.g., payments for rent, mortgage, or insurance). In the Interview Survey, participants are visited once
every 3 months for 4 consecutive quarters.
In the Diary Survey, participants record expenditures daily for 2 consecutive weeks. The survey is designed to collect expenditures for
small-ticket and frequently purchased items, such as detailed types of food (e.g., white bread, ground beef, butter, or lettuce).
The CE microdata may be downloaded on the CE website (https://www.bls.gov/cex/pumd.htm).

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

2 Details about topcoding are provided in the public-use microdata documentation for the year of interest. (See, for example,
Consumer Expenditure Interview Survey, Public Use Microdata, 2013 User’s Documentation, September 10, 2014, https://
www.bls.gov/cex/.)
3 The CE sample design is pseudorandom. The proper use of weights requires the use of the method of balanced repeated
replication.
4 Using the link to a BLS-maintained online tool (https://data.bls.gov/cgi-bin/dsrv?cx), users can obtain time-series data for published
expenditure categories by predetermined demographic series (e.g., age of reference people under 25, 25 to 34, etc.). The new tool
will allow users to select data both at detailed levels (e.g., floor coverings) for different groups (e.g., income quintile, age of reference
person, or a cross-tabulation of these items) in nominal or real (i.e., inflation-adjusted) dollars. The new tool also allows users to
choose whether to display means by calendar year (consistent with CE publications) or collection year (i.e., the year in which the
expenditure information was collected, but not necessarily when the expenditures were made). For example, note that with its 3month recall, Interview Survey respondents who are visited in January are reporting expenditures that took place in the prior year.
5 For details, see Geoffrey D. Paulin and William Hawk, “Improving data quality in Consumer Expenditure Survey with TAXSIM,”
Monthly Labor Review, March 2015, https://www.bls.gov/opub/mlr/2015/article/pdf/improving-data-quality-in-ce-with-taxsim.pdf.
6 The purpose of the bounding interview is to ensure that consumers interviewed more than once do not report expenditures in
subsequent interviews for which data have already been collected. As an example, if a respondent in the first interview reports
purchase of a refrigerator for $500 and does so once again in the second interview, the interviewer can make sure that the secondinterview report is indeed a new refrigerator, different from the one reported 3 months earlier in the bounding survey.
7 The sample redesign occurs decennially, when certain cities or other areas enter the sample and others leave, based on changes in
population or other factors.
8 These new tables were introduced in September 2015 and can be found at: https://www.bls.gov/cex/csxresearchtables.htm.

RELATED CONTENT

Related Articles
Consumer Expenditure Survey Microdata Users’ Workshop and Survey Methods Symposium, 2014, Monthly Labor Review, July
2015.
Consumer Expenditure Survey Microdata Users’ Workshop and Survey Methods Symposium, 2013, Monthly Labor Review, April
2014.

Related Subjects
Seasonal adjustment

Consumer expenditures

10

June 2016

The first 50 years of the Producer Price Index:
setting inflation expectations for today
To help mark the Monthly Labor Review’s centennial, the
editors invited several producers and users of BLS data to
take a look back at the last 100 years. This article tells the
story of the first 50 years of the Producer Price Index (PPI).
During this period, the PPI established its reputation as a
key economic indicator and set the precedents for future
inflation expectations.
The Producer Price Index (PPI) is our nation’s primary
measure of price changes in the domestic supply chain,
allowing us to monitor how price increases or decreases
are passed through from producers to consumers. This
Principal Federal Economic Indicator marked its 125th
anniversary in March 2016. Over its history, the PPI has set
precedents for timely and accurate price index data. Its
methodology and content evolved early on, as our economy
was changing from one based on agriculture to one based
on manufacturing. Today, the PPI continues to evolve as

Lana Conforti
conforti.lana@bls.gov
Lana Conforti is an economist in the Office of
Prices and Living Conditions, U.S. Bureau of
Labor Statistics.

the service-providing sectors now contribute more to gross
domestic product than the goods-producing sectors.
The PPI has also set precedents for helping inform the inflation expectations of economists and government
leaders. It was the first economic indicator to show the macroeconomic effects of trade and war. It was the first
measure to reveal the negative consequences that can arise from deflation and the impacts of fiscal and monetary
policy on prices. Capturing a long history of price changes, the PPI continues to provide insights into bellwethers of
economic turmoil and growth.
The PPI has been well documented since World War II, but its early history is less well known.1 This article tells
the story of the first 50 years of the PPI and how it established its reputation as a key economic indicator.

PPI origin

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

March 3, 2016, marked the 125th anniversary of the PPI—one of the oldest economic time series compiled by the
federal government. The index, known as the Wholesale Price Index (WPI) until 1978, was established as part of a
U.S. Senate resolution on March 3, 1891, the last day of the last session of the 51st U.S. Congress.2 This
Congress was famously known as the “Billion-Dollar Congress,” because of its expensive initiatives, such as

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

expanding the Navy and creating pensions for families of military members who served in the Civil War. It operated
in an era of industrialization, immigration, and economic growth.3 Two of its most well-known bills were the
Sherman Antitrust Act, which sought to protect consumers from certain anticompetitive business practices that
tended to raise prices (e.g., monopolies and cartels),4 and the McKinley Tariff Act of 1890, which raised duties on
imports with the goal of protecting domestic industries from foreign competition.5 Born out of the necessity to
measure the impact of such economic policies, the resolution marking the origin of the PPI read thus:
Resolved, The Committee on Finance be, and they are hereby, authorized and directed, by subcommittee or
otherwise, to ascertain in every practicable way, and to report from time to time to the Senate, the effect of the
tariff laws upon the imports and exports, the growth, development, production, and prices of agricultural and
manufactured articles, at home and abroad….6
In response to this resolution, Senator Nelson W. Aldrich, who later played a role in the establishment of the
Federal Reserve System, authored a report on Retail Prices and Wages in July 1892.7 According to this report, the
demand for price and wage data arose because the lack of reliable data had caused persistent disputes over
economic facts. In addition, legislators realized it would be impossible to judge the relative economic progress of
the United States and its people without measures of prices and wages. For these reasons, the Senate Committee
on Finance made sure to establish a legacy of objective and accurate data:
There was no expectation that the members of the committee would agree about the political or even the
economic bearings of the facts ascertained; but all were desirous that hereafter there should be no reason to
question the integrity of the facts.8
A Bureau of Labor Statistics (BLS) committee headed by Dr. Roland Falkner, a statistics professor from the
University of Pennsylvania, was tasked with collecting prices and producing the original index data.9 At the Senate
committee’s request, prices were collected from seven main distribution centers across the country:
• Baltimore
• Boston
• Chicago
• Cincinnati
• New Orleans
• New York
• San Francisco
Over a 28-month period beginning in mid-1889, BLS obtained 52,393 price quotations for 218 items purchased by
wholesalers (commonly referred to as “jobbers” at the time).10 In a rather informal collection process, experts in
the field (today known as field economists) received the following instructions:
As soon as you have completed the collection of wages and prices in [your city], please collect the quotations
for wholesale prices….You can change the word “retail” to “wholesale” and make the blank conform. By
“wholesale prices” I mean…the prices to jobbers.11
Once collected, these data were compiled by BLS into the first WPI, which was made up of eight equally weighted
groupings of products:

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

• Food
• Cloths and clothing
• Fuel and lighting
• Metals and implements
• Lumber and building materials
• Drugs and chemicals
• House-furnishing goods
• Miscellaneous
The data produced by the BLS committee were published in Senator Aldrich’s 1892 report, which showed that
prices fell 0.3 percent from June 1889 to September 1891.12 After this first publication, in 1893, the Senate
committee and BLS completed their initial mission by publishing a report on historical prices for the years 1840–
90.13 This massive compilation was the first of its kind in U.S. history and was made possible only by the
dedicated efforts of field economists and with cooperation from the business community.
After several years of planning, in 1900, BLS published Wholesale Prices, 1890 to 1899, the first publication
produced without congressional oversight.14 But it wasn’t until March 1902 that the regular annual publication of
Course of Wholesale Prices (hereafter referred to as Wholesale Prices) began.15 Each annual publication
contained monthly data for the previous calendar year and analysis of the data both for that year and from the
base period, 1890.

Early PPI publications
Since the initial annual publication of Wholesale Prices, which contained data for 1890–1901, BLS made routine
methodological improvements. For example, in that publication, items were sorted into groupings based on origin,
rather than end use, which was how they were organized in the original reports prepared in coordination with
Congress. In addition, farm products were separated from the category of food, forming a ninth major grouping.16
Each annual WPI publication—typically published in March—included a comprehensive analysis of price
movements in the U.S. economy from the base period of 1890 through the most recently completed calendar
year.17 Monthly prices were usually from the first day of each month, and data sources included trade journals,
manufacturers, and, to a lesser extent, boards of trade and other government bureaus. Average nominal prices for
all items were published alongside the index data. Later, the publication also featured price data for up to 26
foreign countries.
The first principal indicator of wholesale price changes, or headline index, was the WPI for All Commodities, which
combined indexes for about 250 select items from all nine major groupings. Although weighting was not yet used
to combine indexes, the choice of index for inclusion in the aggregate WPI was based on the “relative importance”
of that index’s product in the U.S. economy. The term “relative importance” was carried over to describe weights
once they were implemented in 1914. To this day, PPI tables of weights continue to be referred to as relative
importance tables.
Most of the original products remain in the current survey in some variant, although a few, such as wool
broadcloth, granite plates and teacups, and wooden pails and tubs, have become obsolete.18 In contrast, a

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

number of manufactured products covered in the PPI today—products such as processed foods, consumer goods,
plastic products, most chemicals, and machinery—are noticeably absent from the original index.
BLS also published composite indexes, to compare changes in prices for raw materials in the WPI with those for
manufactured goods. Such indexes, which grouped goods by level of processing, would later become the focus of
PPI reports. Since the initial publication of Wholesale Prices, the history of the PPI—just like that of the U.S.
economy—has been one of constant improvement and growth.

Early industrialization

5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The stock market crash and the bank failures that set off an economic downturn in 189319 were reflected in WPI
declines in the first 7 years for which data exist. In 1898, the Spanish–American War shifted political and economic
power to the United States, propping up its future trade and industrialization.20 The WPI reflected this upturn in
1898 and increased by more than 8.0 percent in 1899 and 1900. This was the first recorded pickup in inflation
during wartime, an occurrence that would become a trend. From 1901 to 1913, the index rose moderately, at an
annual average rate of 1.6 percent.21 (See figure 1.)
At a time marked by advances in assembly line manufacturing and the U.S. acquisition of the Panama Canal,
which further opened the nation’s economy to trade, BLS in 1908 added new products to the WPI.22 Among the
1908 additions were horses, dressed poultry, and canned food. Concerns about accurate techniques for
substituting old with new products into aggregate WPIs led to the introduction of two indexes: one with the list of
products included before additions or substitutions were made and one with the new list of goods.23 This laborintensive practice continued as new products were added again in 1914, 1915, and 1921, so customers could
compare price changes between any two years with WPIs made of identical goods.
In 1914, war was declared in Europe, the U.S. Federal Reserve banks opened for business, and the WPI
underwent its first major revision. Most significantly, weights were applied to WPIs for the first time on the basis of
1909 U.S. Census of Manufacturers value-of-shipments data. To conform to the new weighting system, all indexes
were retroactively calculated back to 1890. The base dates for all WPIs were also updated to 1914, to use the
latest and most trustworthy data as a base and to accommodate the addition of new goods. The number of series
published increased to 340, and BLS began publishing WPI data for all commodities and major commodity
groupings in the Monthly Labor Review in 1915. WPI data were included in the first issue of the Review alongside
data for the Consumer Price Index, which had just appeared in 1914.24

World War I to the Great Depression

6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The annual publication of Wholesale Prices was suspended in 1918 because of the U.S. entry into World War I.
Resuming publication in 1921, BLS issued a report with WPI data for 1917–19, to highlight price changes during
the war.25 Although a major revision had taken place as recently as 1914, the data were revised once again, this
time with a prewar base year of 1913. As in the previous rebasing, index numbers were revised all the way back to
1890. This recalculation accommodated the addition of indexes for goods introduced in the 1919 data and the
revision of previous data with more pricing information. Further, weights for several home furnishings were
obtained for the first time, and a new aggregate index for furniture was added. The number of series published
now totaled 371. A useful highlight of this bulletin were charts printed on perforated translucent paper, which
allowed data users to tear out and overlay charted data for easy comparison.

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Data during World War I revealed unprecedented price increases between 1917 and 1919, with the allcommodities index rising 56 percent. Historically, to that point, the headline index had moved most closely with the
WPI for farm products, which rose 65 percent during the 1917–19 period. Other notable increases were seen in
the indexes for cloth and clothing, house furnishings, and building materials, which more than doubled during the
war. Exceptions to the typical wartime inflation trend occurred in the indexes for fuels, metals, and chemicals,
because of the introduction of government price controls. In addition to the usual analysis, data for 1919 were
supplemented with extensive commentary on how the war affected price changes for certain important food
commodities.26 Before resuming publication of Wholesale Prices, BLS also assisted with a special report, History
of Prices During the War, published by the U.S. War Industries Board in 1919.27
In 1922, BLS published a number of bulletins to bring WPI data completely up to date. Beer, whiskey, and liquors
were dropped from the index because of the 18th amendment prohibition on the sale of alcoholic beverages.
Overall, the total number of WPI series grew to 450.28 For the first time, BLS had enough data to create new
subgroupings under the existing nine major groupings. For example, under the farm products grouping,
subgroupings were created for grains and for livestock and poultry. Subgroupings for drugs and pharmaceuticals,
paper and pulp, and iron and steel were also among the more notable additions.
Along with the addition of new subgroupings, significant changes in methodology were implemented. BLS returned
to grouping products by end use (as in the original indexes produced in coordination with Congress), as opposed
to by origin. This resulted in some products falling into multiple categories; for example, steel and nails fell under
both building materials and metal products, and potatoes fell under both food and farm products. However, each
product was still weighted only once in the WPI for All Commodities.29 The relative-importance weights were also
updated with data from the 1919 Census of Manufacturers.30
The last WPI revision before the Great Depression occurred with the release of 1927 data. Data were rebased
from 1913 to 1926 and revised with weights updated from the 1923 and 1925 biennial censuses.31 With more than
100 newly introduced indexes, the total number of series went up to 550. Notable additions to the WPI survey were
automobiles, tires, and sewing machines. New aggregate WPIs were also introduced for nonagricultural goods,
raw materials, semi-manufactured articles, and finished products.
Because of the high volume of indexes added by 1927, the revision was completed in stages, and the final report
with revised data back to 1890 was published in January 1929.32 Table 1 presents a list of the major product
groupings and subgroupings in this report.
Table 1. WPI product groupings and subgroupings, 1927
Major grouping
Farm products
Foods

Subgroupings
Grains, livestock and poultry, other farm products
Meats; butter, cheese, and milk; other foods

Hides and leather products(1) Hides and skins(1), leather, boots and shoes, other leather products(1)
Textile products(1)

Cotton goods, silk and rayon(1), woolen and worsted goods(1), other textile products(1)

Fuel and lighting

Anthracite coal, bituminous coal, coke(1), manufactured gas(1), petroleum products(1)

Metals and metal products

Iron and steel, nonferrous metals, agricultural implements(1), automobiles(1), other metal products(1)

Building materials

Lumber, brick, Portland cement(1), structural steel, paint materials(1), other building materials

See footnotes at end of table.

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 1. WPI product groupings and subgroupings, 1927
Major grouping

Subgroupings

House-furnishing goods

Chemicals, drugs and pharmaceuticals, fertilizer materials, mixed fertilizers(1)
Furniture, furnishings

Miscellaneous

Cattle feed, paper and pulp, rubber(1), automobile tires(1), other miscellaneous

Chemicals and drugs

Notes:
(1) New category introduced in 1927.

Source: U.S. Bureau of Labor Statistics.

After the end of World War I in 1918, the WPI continued its upward trend through mid-1920, but, in early 1921, it
abruptly returned to its pre-1918 level. From 1922 to 1925, the index displayed volatility, although in a tight range.
(See figure 2.)

The Great Depression to World War II
The WPI demonstrated its potential as a leading economic indicator with broad-based declines beginning in 1926.
On October 24, 1929, a day known as “Black Thursday,” the stock market crashed, precipitating the Great
Depression.33 By October 1930, the WPIs for all major commodity groupings had dropped below their 1926 base,
and declines continued through 1932. According to WPI data, it took until 1943 for prices to recover to their
predepression levels, partly because of a midrecovery recession in 1937. In comparison, during the 2007–09
Great Recession, the PPI for All Commodities turned down in July 2008 but recovered within 6 years.

10

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

11

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Before 1925, shorter periods of deflation had occurred periodically, but the 1925–33 deflationary period was, and
remains, the longest streak of year-over-year declines reflected in the all-commodities index. In the 1930s, it
became widely accepted that long-run price declines had intensified the Great Depression,34 and deflation attained
its unsavory status. Seeking to get deflation under control, the federal government passed two new pieces of
legislation in 1933, both of which contributed to an upturn in the WPI. The first, the Agricultural Adjustment Act,
levied new taxes on certain farm products. This measure, combined with a major drought that led to the 1934
“Dust Bowl,” put upward pressure on farm and food prices.35 The second legislation, the National Industrial
Recovery Act, created inflation pressure in nonfood prices by deregulating industries, allowing them to fix prices
and create production quotas.36
With the beginning of the Great Depression, the need for economic data became more urgent than ever. BLS
responded to this need in 1930, when it began publishing monthly bulletins with WPI data to supplement the
comprehensive annual issue of Wholesale Prices. The new bulletins presented analyses on price movements for
the most recent year, for the most recent month, and for selected weeks. These short-term analyses contrasted
with those included in the annual publication, which emphasized historical price movements in the WPI from the
base date.
The economic volatility of the 1930s also encouraged experimentation with the content of the WPI publication. In
1932, the number of WPI price series expanded from 550 to 784. The new series were almost entirely for fully
manufactured goods. Some commodities were rearranged, and new subgroupings were added. Weights were
updated in 1932 with the most recent Census of Manufacturers data, but revised figures were only calculated back
to 1926, rather than 1890, as had been done in past revisions. Notices touting improvements in sample coverage,
transaction descriptions, and weights for specific WPI groupings (such as farm machinery and cement) began
appearing in the reports as well.37 These notices resemble today’s PPI notices of resampling of industries. The
focus of the written analysis in the monthly WPI reports also began to shift, from the all-commodities index to both
the aggregates excluding food and the indexes for goods by level of processing. In addition, the emphasis
frequently shifted between short- and long-term price changes. For a time in 1935 and 1936, weekly analysis led
the headline paragraph. During the same period, the WPI program switched from publishing international prices
and average nominal prices on a monthly basis to publishing them only a few times a year.
In mid-1936, the monthly WPI report announced methodological improvements that would ensure increased WPI
coverage, more detailed descriptions for sampled transactions, and better accuracy of prices and item
classifications.38 In January 1937, the program moved from calculating WPIs as chain-type indexes to calculating
them with a fixed-base formula. No immediate difference between the two types of indexes was noticeable for
overlapping data in 1936; however, in the long run, it was expected that the new methodology would be crucial to
maintaining accurate inflation data as product substitutions were made. With nearly 800 series being published in
1937, products were constantly growing obsolete or being replaced by newer alternatives.39

12

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The WPI for All Commodities finally broke its depression slump in mid-1933, finishing the year with a double-digit
advance. Bolstered by the impacts of the Agricultural Adjustment Act and the Dust Bowl, farm and food price
increases led the broad-based rise in WPIs through 1936. In the midst of economic recovery, however, came the
recession of 1937–38. This downturn has recently gained the attention of researchers, because of its parallels with
the slow recovery from the 2007–09 Great Recession. Several factors have been cited as causes for the 1930s
“recession within depression,” including higher taxes, rollbacks in government stimulus spending, and an increase
in reserve requirements for banks by the Federal Reserve. These policies were implemented when the economy
was not quite ready to absorb them.40 Since this time, measures of inflation, such as the PPI, have gained
recognition as crucial indicators for policymakers.
After falling to a 5-year low in mid-1939, the WPI turned upward again, mainly because of the outbreak of World
War II. The subsequent production increases to support the war effort boosted the economy through the duration
of the war. (See figure 3.)

PPI today
Since World War II, the PPI has grown to cover nearly all output in the goods-producing sectors and much of the
output in the construction, utilities, and service sectors. In total, the PPI now tracks prices for over three-quarters of
private domestic production, from raw materials to final-demand goods and services. In addition, indexes
measuring price changes are available by industry, end use, and stage of production. Each month, the PPI
program publishes nearly 10,000 individual indexes, which are used commonly by private businesses and the
government as indicators of inflation. As evidence of its ongoing leadership in price-index development, the
program has issued more than a dozen notices of expansion or methodology improvement in the last 5 years
alone. The most prominent of these notices has been on the shift in focus of the PPI news release—a shift from
the index for finished goods to the index for final demand.41
SUGGESTED CITATION

13

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Lana Conforti, "The first 50 years of the Producer Price Index: setting inflation expectations for today," Monthly
Labor Review, U.S. Bureau of Labor Statistics, June 2016, https://doi.org/10.21916/mlr.2016.25.
NOTES

14

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

1 See various articles on PPI methodology, https://www.bls.gov/ppi/methodology.htm; Edgar I. Eaton, “A description of the revised
Wholesale Price Index,” Monthly Labor Review, vol. 74, no. 2, February 1952, pp. 180–187; Pearl C. Ravner, “Price trends and the
business cycle in postwar years,” Monthly Labor Review, vol. 85, no. 3, March 1962, pp. 241–248; Bennett R. Moss, “Industry and
sector price indexes,” Monthly Labor Review, vol. 88, no. 8, August 1965, pp. 974–982; John F. Early, “Improving the measurement of
producer price change,” Monthly Labor Review, vol. 101, no. 4, April 1978, pp. 7–15; John F. Early, Mary Lynn Schmidt, and Thomas
J. Mosimann, “Inflation and the business cycle during the postwar period,” Monthly Labor Review, vol. 107, no. 11, November 1984,
pp. 3–7, https://www.bls.gov/opub/mlr/1984/11/art1full.pdf; and Andrew G. Clem, “Milestones in Producer Price Index methodology
and presentation,” Monthly Labor Review, vol. 112, no. 8, August 1989, pp. 41–42, https://www.bls.gov/mlr/1989/08/rpt1full.pdf.
2 “50th to 59th Congresses (1887–1907),” History, Art & Archives (U.S. House of Representatives), http://history.house.gov/Institution/
Session-Dates/50-59/.
3 Cynthia C. Northrup, The American economy: a historical encyclopedia, volume 1 (Santa Barbara, CA: ABC-CLIO, 2011), p. 1257;
and Sean D. Cashman, America in the Gilded Age: from the death of Lincoln to the rise of Theodore Roosevelt (New York, NY: New
York University Press, 1993).
4 “The antitrust laws,” Guide to Antitrust Laws (Federal Trade Commission), https://www.ftc.gov/tips-advice/competition-guidance/
guide-antitrust-laws/antitrust-laws.
5 “The McKinley Tariff of 1890,” History, Art & Archives (U.S. House of Representatives), http://history.house.gov/HistoricalHighlight/
Detail.
6 Journal of the Senate, 51st Congress, 2d session (1891), p. 218.
7 Nelson W. Aldrich, Retail prices and wages, Report 986, Committee on Finance (Government Printing Office, 1892), https://
books.google.com/books?id=d5g0AQAAMAAJ. See also “Jekyll Island and the creation of the Fed,” Classroom Economist (Federal
Reserve Bank of Atlanta), https://www.frbatlanta.org/education/classroom-economist/jekyll-island.aspx.
8 Aldrich, Retail prices and wages, p. II.
9 Publications on wholesale prices from 1900 to 1913 are labeled as reports of the Department of Labor (DOL), which actually
referred to the Bureau of Labor, which was established in 1884. DOL, in its current form, was established in 1913. Also in 1913, the
word “statistics” was added to the name Bureau of Labor, and BLS was made a branch under DOL. Going forward, this article will
refer to BLS inclusive of the Bureau of Labor, as it was called before 1913. See “The Organic Act of the Department of Labor” (U.S.
Department of Labor), https://www.dol.gov/general/aboutdol/history/dolhistoxford.
10 Aldrich, Retail prices and wages, p. CV.
11 Ibid., p. CXXIII.
12 Ibid., p. VI.
13 Aldrich, Retail prices, wages, and transportation, Report 1394, Committee on Finance (Government Printing Office, 1893), https://
books.google.com/books?id=cziCjgEACAAJ.
14 Wholesale prices, 1890 to 1899, Bulletin 27 (U.S. Bureau of Labor Statistics, 1900) https://books.google.com/books?
id=qF7GAAAAMAAJ.
15 Course of wholesale prices, 1890 to 1901 (U.S. Bureau of Labor Statistics, 1902), https://books.google.com/books?
id=nokuAAAAYAAJ.
16 Ibid.
17 Unless otherwise specified, references to years of WPI publications will refer to the time period for which data are measuring
prices, rather than the year the publication was released.

15

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

18 Course of wholesale prices, 1890–1902, Bulletin 45 (U.S. Bureau of Labor Statistics, March 1903), https://fraser.stlouisfed.org/
docs/publications/bls/bls_v08_0045_1903.pdf.
19 Mark Carlson, “Causes of bank suspensions in the Panic of 1893” (Federal Reserve Board, 2002), http://www.federalreserve.gov/
pubs/feds/2002/200211/200211pap.pdf.
20 “The Spanish–American War, 1898,” Milestones, 1866–1898 (U.S. Department of State, Office of the Historian), https://
history.state.gov/milestones/1866-1898/spanish-american-war.
21 Wholesale prices, 1890 to 1913, Bulletin 149 (U.S. Bureau of Labor Statistics, May 1914), https://fraser.stlouisfed.org/scribd/?
item_id=497569&filepath=/docs/publications/bls/bls_0149_1914.pdf.
22 Wholesale prices, 1890 to 1908, Bulletin 81 (U.S. Bureau of Labor Statistics, March 1909), p. 219.
23 Wholesale prices, 1890 to 1914, Bulletin 181 (U.S. Bureau of Labor Statistics, October 1915), p. 255, https://fraser.stlouisfed.org/
scribd/?item_id=476826&filepath=/docs/publications/bls/bls_0181_1915.pdf.
24 William J. Wiatrowski, “Reimagining the Monthly Review, July 1915,” Monthly Labor Review, February 2016, https://www.bls.gov/
opub/mlr/2016/article/reimagining-the-monthly-review-july-1915.htm; and Stephen B. Reed, “One hundred years of price change: the
Consumer Price Index and the American inflation experience,” Monthly Labor Review, April 2014, https://www.bls.gov/opub/mlr/2014/
article/one-hundred-years-of-price-change-the-consumer-price-index-and-the-american-inflation-experience.htm.
25 Wholesale prices, 1890 to 1921, Bulletin 320 (U.S. Bureau of Labor Statistics, December 1922), p. 8, https://fraser.stlouisfed.org/
scribd/?item_id=492994&filepath=/docs/publications/bls/bls_0320_1922.pdf.
26 Wholesale prices, 1890 to 1919, Bulletin 269 (U.S. Bureau of Labor Statistics, July 1920), https://fraser.stlouisfed.org/scribd/?
item_id=476829&filepath=/docs/publications/bls/bls_0269_1920.pdf.
27 Wesley C. Mitchell, History of prices during the war (U.S. War Industries Board, 1919), https://books.google.com/books?
id=lOEZAAAAYAAJ.
28 Wholesale prices, 1890 to 1921.
29 Although the use of weights in aggregate WPIs was a major improvement, by the 1960s the distortion in aggregate WPIs caused
by multiple counting of price changes had become noticeable, leading BLS in 1978 to change focus from all-commodities WPIs to
stage-of-processing PPIs. Multiple counting is present in all-commodities indexes because they include every given product at all
levels of processing through the supply chain. For example, to the extent a price change is passed on from suppliers, a change in the
price of raw cotton—which is converted to yarn, then to fabric, and finally to apparel—would be included five times in the allcommodities index. See John Early, “Improving the measurement of producer price change,” Monthly Labor Review, vol. 101, no. 4,
April 1978, pp. 7–15.
30 Wholesale prices, 1890 to 1921.
31 Wholesale prices, 1913 to 1927, Bulletin 473 (U.S. Bureau of Labor Statistics, January 1929), https://fraser.stlouisfed.org/docs/
publications/bls/bls_0473_1929.pdf.
32 Revised index numbers of wholesale prices, 1923 to July 1927, Bulletin 453 (U.S. Bureau of Labor Statistics, September 1927),
https://fraser.stlouisfed.org/scribd/?title_id=4046&filepath=/docs/publications/bls/bls_0453_1927.pdf.
33 For historical recession dates, see “U.S. business cycle contractions and expansions” (Cambridge, MA: National Bureau of
Economic Research), http://www.nber.org/cycles.html.
34 W. L. Crum, “The course of commodity prices,” The Review of Economics and Statistics, vol. 16, no. 10, October 1934, pp. 207–
212, especially p. 207.
35 “Wholesale prices,” Monthly Labor Review, vol. 38, no. 2, February 1934, pp. 464–475; and Crum, “The course of commodity
prices.”

16

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

36 The act’s purpose was “to encourage national industrial recovery, to foster fair competition, and to provide for the construction of
certain useful public works, and for other purposes.” See “Transcript of National Industrial Recovery Act (1933),” https://
www.ourdocuments.gov/doc.php?doc=66&page=transcript.
37 Wholesale prices, June 1936, Serial No. R. 419 (U.S. Bureau of Labor Statistics, 1936); and Wholesale Prices, March 1939, Serial
No. R. 929 (U.S. Bureau of Labor Statistics, 1939).
38 “Wholesale prices,” Monthly Labor Review, vol. 43, no. 3, September 1936, pp. 772–782.
39 Jesse M. Cutts and Samuel J. Dennis, “Revised method of calculation of the Wholesale Price Index of the United States Bureau of
Labor Statistics,” Journal of the American Statistical Association, vol. 32, no. 200, 1937, pp. 663–674.
40 Patricia Waiwood, “Recession of 1937–38,” Federal Reserve History (Federal Reserve Bank of Richmond, November 2013), http://
www.federalreservehistory.org/Events/DetailView/27.
41 For special notices and more information on the PPI, visit https://www.bls.gov/ppi/. To subscribe for program updates, visit https://
www.bls.gov/ppi/update.htm.

RELATED CONTENT

Related Articles
New PPI net inputs to industry indexes, Monthly Labor Review, October 2015.
The Monthly Labor Review turns 100, Monthly Labor Review, July 2015.
Comparing new final-demand producer price indexes with other government price indexes, Monthly Labor Review, January 2014.
A new, experimental system of indexes from the PPI program, Monthly Labor Review, February 2011.

Related Subjects
Producer price index

Statistical programs and methods

Bureau of labor statistics

17

History

Prices

Inflation

June 2016

Does the place you grew up in shape your future
as an entrepreneur? Evidence from Italy
Yavor Ivanchev
Entrepreneurs, especially successful ones, have always been a mystery, if not an envy, for many of us bystanders.
We often surrender, perhaps conveniently so, to the argument that these individuals possess talents that we don’t
—innate abilities unattainable through formal or informal learning. That this view has taken hold is not surprising.
The business empires of Microsoft and Apple—two obvious examples—were built by college dropouts, and many
other enterprises have spawned and thrived under the watch of people with little managerial training.
Notwithstanding anecdotal support for the innate nature of entrepreneurship, economists have started to afford
greater importance to nurture. In a recent article titled “Learning entrepreneurship from other entrepreneurs?
” (National Bureau of Economic Research, Working Paper 21775, December 2015), Luigi Guiso, Luigi Pistaferri,
and Fabiano Schivardi report that adolescents growing up in areas with high concentration of firms are more likely
to become entrepreneurs and to be successful at their jobs. The authors see geographical firm density as a
learning opportunity: “for a young individual growing up in Silicon Valley,” they contend, “it should be easier than
elsewhere to learn how to set up and run a firm.”
To test this hypothesis, Guiso et al. rely on two complementary data sources from Italy—the Bank of Italy Survey of
Households Income and Wealth, which provides demographic and income data for a representative sample of
Italian households, and a sample of entrepreneurs and their firms from a survey conducted by the Italian National
Association of Insurance Companies (ANIA). Data from ANIA are supplemented with measures that capture
various managerial skills and help isolate learned from innate abilities. Firm concentration is recorded at the
provincial level, and variation in outcomes—likelihood of becoming an entrepreneur and entrepreneurial success—
is examined both across Italian provinces and over time within a province.
Consistent with theory, variation in outcomes does exist, and it is sizable. The authors’ regression analysis
indicates that an increase of one standard deviation in firm density in one’s location at “learning age” (age 18,
according to the study) is associated with an 8-percent increase both in the likelihood of sorting into
entrepreneurial occupation and in personal income. The businesses of individuals exposed to this environmental
channel also stand out in terms of performance, boasting significantly higher total and per-worker productivity.
These results remain robust in the presence of various statistical controls, including local availability of capital
(which could affect firm concentration) and coming from a household with entrepreneurial parents.
Guiso et al. are careful not to overstate their case, however, and while they highlight the importance of learning in
one’s formative years, they do not attempt to underplay the role of innate abilities. Indeed, marrying measures of
skill with ANIA data shows that the effect of firm density during adolescence is largely limited to the cultivation of
better managerial skills and practices. The authors surmise that other personal traits normally viewed as

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

preconditions for successful entrepreneurship—traits such as greater risk tolerance and hunch for business and
product innovation—likely remain in the domain of the innate.

2

June 2016

Labor market will shape U.S. economy in years to
come
Editor’s note: This essay is part of a series being published
to help commemorate the Monthly Labor Review’s
centennial (July 1915–July 2015). The essays―written by
eminent authorities and distinguished experts in a broad
range of fields―cover a variety of topics pertinent to the
Review and the work of the Bureau of Labor Statistics.
Each essay is unique and comprises the words and opinion
of the author. We’ve found these essays to be enlightening
and inspirational. We hope you do as well.
In its 100 years in publication, the Monthly Labor Review
has remained relevant and insightful through many labor
markets. This includes the boom times of the 1920s, 1960s,
and 1990s, and the dark times of the 1930s, 1970s, and the
past decade. It includes the World Wars, the Korean War,

Mark M. Zandi

Vietnam, Iraq, and Afghanistan. It includes the surge in
Mark M. Zandi is chief economist of Moody’s
Analytics, where he directs economic research.

female labor force participation in the second half of the
20th century, and the rapidly shifting fortunes of industries,
occupations, and regions of the country.
The next 25 years will surely be no different. There will be

good and bad times, wars, and large demographic changes. Some industries, occupations, and regions will rise in
importance, and others will decline. Through it all, the Monthly Labor Review will continue to provide the
information vital to understanding every development.
Among the most important coming changes to the labor market, and arguably among the most certain, is that the
nonwhite population will become the majority. Twenty-five years from now, rapidly growing Latino, Black, Asian,
and other minority groups will together account for more than half the population.
The economy’s success thus critically depends on raising the educational and skill levels of these groups so that
they can fill the jobs of the future. This poses significant challenges, as many are from lower income households
with fewer educational opportunities. Our property tax system does a poor job financing K-12 education in many
poor communities, and the current strategy of using student loans to finance the higher education of these
financially pressed households is failing.

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Heightening the importance of raising the skill levels of minority groups is the prospect that labor will be in
perennial short supply over the next quarter-century. This is in stark contrast to the past quarter-century, when
there was more or less a surfeit of labor. Unemployment was generally higher than that consistent with full
employment, and not surprisingly, wage growth and the share of national income going to labor declined.
Behind this dismal performance is the concerted effort by the Federal Reserve to wring out inflation, which was the
economy’s overwhelming problem a quarter-century ago. The Fed managed interest rates so that the economy
more often than not operated below full employment, weighing on wages.
The overriding problem today is disinflation, and the specter of deflation hangs over some parts of the world.
Global central banks, including the Fed, are working hard to support stronger job growth and reduce
unemployment in order to lift wage growth and inflation. The Fed will likely manage interest rates so that the
economy generally operates above full employment.
Adding to the coming tight labor market is the inexorable aging of the large baby boom generation into retirement.
Labor force participation rates are set to steadily decline over the next quarter-century as the boomers leave the
workforce, and labor force growth will come to a virtual standstill at times. Businesses will need to raise wages
more aggressively to attract workers on the fringes of the labor force. The labor share of income should rise, and
the long-running skewing of income distribution may even unwind somewhat.
Given the tight labor market and the growing minority share of the population, the politics around immigration
reform are likely to shift. A path to citizenship for the undocumented seems likely, with more legal immigrants
allowed into the country, particularly the highly skilled and educated. The U.S. labor force will grow increasingly
more diverse, not just in global gateway cities, but in farther-flung places across the nation.
American businesses will also look to cultivate a more global workforce. They will be selling more of their goods
and services more broadly across the world, and they will want workers in those places. They will also tap the
talents of those workers and forge truly global workforces, bringing the world closer together economically,
politically, and socially.
Businesses will also need to invest more aggressively in new technologies. Without that investment, productivity
growth will languish as it has in recent years, and the tight labor market will translate into slower overall economic
growth. Middle and lower income households will struggle with this the most, and the nation’s long-term fiscal
challenges will become even more daunting.
Faster innovation and productivity growth will pose other issues for the labor market. High-skilled workers will be
enabled by the new technologies, and low-skilled workers will be relatively untouched, as their tasks are more
idiosyncratic and thus less susceptible to technology’s effects. However, many middle-skilled workers are
vulnerable, and unless they can upgrade their skills, new technologies will push them down the income ladder.
The U.S.’s comparative economic advantage is embodied in its workers. Our economic success depends on our
ability to attract the best and brightest from across the globe and to empower all those who are here to become the
best they can be. This won’t be easy, but the Monthly Labor Review will be there to guide us.
SUGGESTED CITATION

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Mark M. Zandi, "Labor market will shape U.S. economy in years to come," Monthly Labor Review, U.S. Bureau of
Labor Statistics, June 2016, https://doi.org/10.21916/mlr.2016.26.
RELATED CONTENT

Related Articles
The Monthly Labor Review turns 100, Monthly Labor Review, July 2015.

Related Subjects
Bureau of labor statistics

History

3

June 2016

The Monthly Labor Review at 100—part III:
inflation, employment, and the labor force since
1980
To help mark the Monthly Labor Review’s centennial, the
editors invited several producers and users of BLS data to
take a look back at the last 100 years. This third article in a
series of four recounts the Review’s history since 1980,
focusing on its coverage of inflation, employment, and the
labor force.
Throughout its history, the Review has striven to maintain
the high quality of the articles it publishes and to keep the
public abreast of both the kinds of economic issues that
perennially affect the nation and new issues that arise as
the nation itself changes economically, demographically,
and culturally. This installment and the next present some
of the pressing issues, the timely issues, and the enduring
issues that have occupied the pages of the Review over the
last 35 years. Both contain what is hoped to be a
representative sample of the articles that have filled the

Brian I. Baker
baker.brian@bls.gov
Brian I. Baker is a senior technical writer–editor in
the Office of Publications and Special Studies,
U.S. Bureau of Labor Statistics.

pages of the journal during that period. This installment
examines three of the most important topics the Review
has brought before its readers since 1980: inflation,
employment, and the composition and dynamics of the
labor force. Both BLS and non-BLS authors contributed

amply to these “bread-and-butter” issues making up the Review’s content over the decades.

Inflation
The double-digit inflation of the late 1970s persisted into the early 1980s. The years 1979, 1980, and 1981 were
bleak economically, with inflation averaging 11.3 percent, 13.5 percent, and 10.3 percent, respectively.1 The
Review continued to track the rising inflation numbers coming in month after month, reporting them to concerned
readers in “Current labor statistics.” Throughout this period of high inflation, articles, reports, and summaries
appeared periodically, analyzing and explaining the numbers and offering reasons for their persistence. John F.
Early, Craig Howell, and Andrew Clem began things in May 1980 with a penetrating article titled “Double-digit
inflation today and in 1973–74: a comparison.” In it, they pointed to energy and housing prices, as well as the cost

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

of materials, as leading causes of the acceleration of inflation to double digits in both periods. Then, invoking this
similarity, they observed that one of the major factors—if not the major factor—that ended the double-digit inflation
of the earlier period was the recession from November 1973 to March 1975. Noting that that recession picked up
steam only in late 1974, the authors suggested that economists watch for another severe recession as the chief
(though not the only) way the high inflation of 1979–81 would end. That they were correct in their assessment,
radical though it was, is attested to by the arrival of a severe recession that brought inflation down from a peak of
10.8 percent in July 1981 to a trough of 3.8 percent in November 1982.

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Other articles appearing in 1980 that addressed the inflation scenario of 1979 were “Wage gains in 1979 offset by
inflation,” by Joan D. Borum (July), a repeat of the concern the author expressed in her 1979 article over the same
situation in 1978; “Slowdown in energy prices eases second-quarter inflation,” by William Thomas, Clem, and
Eddie Lamb (September), an entry in the “Anatomy of price change” series discussing an all-too-temporary
slackening in the pace of inflation during the quarter in question; “Record white-collar pay increase closes decade
but trails inflation,” by Felice Porter (November), another lamentation about too much inflation in the midst of
unusually high pay increases; and “Inflation slows in third quarter, although food prices soar,” by Howell, Thomas,
and Lamb (December), a mixed account of the inflation situation. Although high inflation continued into 1980, the
year 1981 saw a diminution in the number of articles devoted to the subject. Just four articles were written about
the pace of inflation in 1980, and by 1981, inflation, though still high, began to slow noticeably, as recorded in the
lone 1982 Review article on the subject, Craig Howell and Jesse Thomas’ “Price changes in 1981: widespread
slowing of inflation” (April). By the end of 1982, inflation had returned to oft-seen, lesser levels, and while the
Review continued, of course, to record the path of inflation up to the present, the era of double-digit inflation was
over.
Interestingly, in the midst of the rampant inflation of 1979–81 came a debate about how accurately the CPI
represented the phenomenon. The May 1980 issue of the Review—the same issue in which Early, Howell, and
Clem compared the double-digit inflation occurring at the time with that of 1973–74—also was the first of several
issues that featured a number of economists engaging in a spirited discussion over the merits of the CPI. The
proceedings began in that issue with a pair of articles, one by Daniel J. B. Mitchell and the other by Jack E. Triplett,
that addressed critics of the CPI. Mitchell opened the discussion by proposing that critics who maintain that the
CPI exaggerates inflation because some of its components are upward biased ignore the very real possibility that
other of its components have just the opposite effect. Also, maintained Mitchell, although there is no guarantee that
the two sets of components offset each other, neither is there a guarantee that they don’t—and anyway, there is at
least some effect in the opposite direction. In support of Mitchell’s argument, Triplett’s piece pointed to empirical
studies showing that non-BLS indexes of owner-occupied housing, rental costs, and new-car prices—components
ignored by critics—were, in general, higher than corresponding BLS indexes. Thus, it was by no means farfetched,
and was even quite plausible, said Triplett, that these components could dampen whatever upward bias might
exist in the other CPI components.
The March 1981 issue of the Review featured two articles on the accuracy of the CPI. In the first, BLS
Commissioner Janet L. Norwood built on Mitchell’s and Triplett’s exposition, examining “Two Consumer Price
Index issues: weighting and homeownership”—issues that those authors had cited as being ignored by critics.
Bringing to bear past BLS experience showing that weighting changes necessitated by revisions to the CPI had
little effect on the index and pointing out that estimates of inflation based on the U.S. Commerce Department’s
Deflator for Personal Consumption Expenditures (PCE) with different weights were little different from those of the
CPI, Norwood deflected the critics’ complaints as inadequate. Regarding homeownership, she offered an
experimental, more accurate “rental equivalence” CPI as a possible way of addressing the critics’ arguments. As
things turned out, BLS changed its treatment of homeownership and incorporated the change into the CPI in 1983.
The second article, “Indexing federal programs: the CPI and other indexes,” excerpted from a publication jointly
authored by the President’s Council of Economic Advisers and the Office of Management and Budget,
complemented Norwood’s article in that it, too, addressed the complaints leveled at the CPI regarding the issues of

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

weighting and homeownership—and, in line with Norwood’s conclusions, suggested the adoption of the new
experimental CPI as the best way of overcoming the critics’ objections.
Capping off the CPI controversy running through the high-inflation scenario of 1979–81 was a September 1981 trio
of articles under the title “Measuring prices.” Triplett led off the discussion with a highly technical piece titled
“Reconciling the CPI and the PCE Deflator.” Scrutinizing seven versions of the CPI published by BLS and three
versions of the PCE deflator published by the Bureau of Economic Analysis (BEA),2 he concluded that whatever
differences there were among them could be reconciled if one distinguished between “longer-term inflation
measurements and…period-to-period inflation rates.” In the second article of the trio, “Defining the rate of
underlying inflation,” David W. Callahan observed that “Overall measures of price increase reflect both a core rate
of inflation residing in the economy and the short-run effects of uncontrollable external shocks.” Citing the kinds of
shocks that need to be accounted for in the short term, he then reflected on how these prima facie very different
measures might be reconciled, concluding that reconciling them is necessary only in the short run, because, in the
long run, “All shocks are absorbed, all adjustments have been made, and the underlying rate of inflation coincides
with the long-term trend in the measure of overall inflation.” The final article, “Some proposals to improve the
Consumer Price Index,” by Phillip Cagan and former BLS Commissioner Geoffrey H. Moore, suggested fixing the
CPI by changing the way homeownership was measured (the measurement of homeownership was universally
acknowledged as problematic in the official BLS CPI) and experimenting with averaging current- and baseweighted indexes. Together, the articles in the three series from the Reviews of May 1980, March 1981, and
September 1981 on the ongoing controversy over the CPI heralded the change in the index described in Robert
Gillingham and Walter Lane’s June 1982 Review article “Changing the treatment of shelter costs for homeowners
in the CPI”—a change subsequently begun in 1983 and completed in 1987, as related by John L. Marcoot in
“Revision of Consumer Price Index is now under way” (April 1985).
After inflation eased in 1982, returning to “normal” levels toward the end of the year, not only was the era of
double-digit inflation over, but low—sometimes even negative—inflation prevailed from then until the present.3 Still,
although inflation did not possess the urgency it had earlier, the Review reported on it regularly: articles on the
year’s inflation appeared annually well into the 1980s, and features on various aspects of inflation and the CPI
arose from time to time, discussing theoretical and methodological issues (e.g., Kenneth J. Stewart and Stephen
B. Reed, “Consumer Price Index research series using current methods, 1978–98,” June 1999; Janice Lent,
“Estimating an energy consumer price index from establishment survey data,” December 2011), making
comparisons across countries (Walter Lane and Mary Lynn Schmidt, “Comparing U.S. and European inflation: the
CPI and the HICP,” May 2006), and, in a grand tour de force, chronicling the entire history of inflation and the CPI
—the latter’s changing methodology, the political forces shaping the index, and the American experience with
inflation for the past 100 years (Darren Rippy, “The first 100 years of the Consumer Price Index: a methodological
and political history,” April 2014; Stephen B. Reed, “One hundred years of price change: the Consumer Price Index
and the American inflation experience,” April 2014).

Employment
Employment is, by far, the bread and butter of the Review, with more than double the number of articles written on
it than on any other topic from 1980 to the present. That should come as no surprise, because employment is,
after all, the backbone of the U.S.—or, for that matter, any other developed country’s—economy. The Review’s
assessment of the employment situation in the 1980s is perhaps best captured by Lois Plunkert’s September 1990
5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

article, “The 1980’s: a decade of job growth and industry shifts.” Beginning with the interesting juxtaposition of two
recessions in 3 years at the beginning of the decade with “the longest peacetime expansion on record” throughout
the rest of the period, Plunkert went on to describe a scenario in which, indeed, the sheer magnitude of the growth
in employment was impressive, but the reach of the growth was uneven. Led by the large number of new jobs
posted month after month in the service sector, the economy nonetheless steadily lost jobs in the manufacturing
and mining industries, which never recovered from the aforementioned recessions. Even the service sector record
was spotty, with one-quarter of the new jobs created in business or health services and some industries, such as
communications, railroads, and water transportation, losing jobs. On net, the 1980s saw a continuation of the shift
from the goods-producing sector to the service-providing sector, with another 6 percent of employment shifting
over.
In addition to Plunkert’s comprehensive article, a number of articles of lesser scope, but no less substantive, were
written about various aspects of employment during the 1980s. Among the more salient ones were a special labor
force report by Allyson Sherman Grossman on “Working mothers and their children” (May 1981) that cited
changing marital patterns, high inflation, and smaller families as factors contributing to the labor force participation
of more mothers with children under 18 years old; BLS Commissioner Norwood’s August 1983 piece, “Labor
market contrasts: United States and Europe,” which was drawn from her commencement address at Tufts
University’s Fletcher School of Law and Diplomacy and looked beyond the unemployment rates of eight countries
in North America and Europe to find a more favorable employment outlook for the United States than for Europe—
an insight that turned out to be correct as time went on; an article titled “The declining middle class: a further
analysis” (September 1986), by Australian economist Patrick J. McMahon and BLS’s own John H. Tschetter, which
found that, between 1973 and 1982, despite an increase in the proportion of workers in higher paying occupations,
the earnings distribution in the top, middle, and bottom occupation groups had shifted such that each group
included more lower paying positions; and Bruce W. Klein and Philip L. Rones’ “A profile of the working
poor” (October 1989), in which the authors found that 6.4 million workers, or 5.6 percent of the labor force, were
members of poor families. Among these “working poor,” unmarried women with children were at the greatest risk
for living in poverty, with the chief reason being low wages likely due to low levels of education.
Employment in the 1990s is perhaps best summarized in Julie Hatch and Angela Clinton’s December 2000 article,
“Job growth in the 1990s: a retrospect.” After a brief recession that lasted from July 1990 to March 1991, the U.S.
economy “rebounded with the longest running expansion in the Nation’s history.” Employment grew by nearly 21
million over the decade, the most ever recorded between censuses. As in the 1980s, jobs continued to shift from
the goods-producing sector to the service-providing sector of the economy, with mining and manufacturing again
the biggest losers. Employment in business services grew the most, accounting for a third of all job growth in
services over the decade. Computer and data-processing services added more than a million jobs, and colleges
and universities could not produce graduates with computer degrees fast enough for the growing demand. Health
services, though growing at a slower pace than in the 1980s, still contributed more than 2.5 million jobs to the
economy. Government had mixed results: federal government employment fell to levels not seen since 1965 as
the Department of Defense lost 333,000 jobs, but state government and local government added jobs, although
state government did so at a level less than that of the 1980s. The stock market surged, and investors poured
money into Internet-based companies, or “dot-coms.” Amongst all the good news, however, the stage was set for
the bursting of the dot-com bubble in the opening years of the 21st century.

6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Foremost among articles addressing specific employment issues or subperiods of the 1990s was the “annual
story” series of articles that tracked employment every year. The title of each article revealed the progression of
the economy, from recession to record expansion, over the decade: “U.S. labor market weakened in 1990,” by
Steven E. Haugen and Joseph R. Meisenheimer II (February 1991); “Job market slid in early 1991, then struggled
to find footing,” by Meisenheimer, Earl F. Mellor, and Leo G. Rydzewski (February 1992); “1992: job market in the
doldrums,” by Thomas Nardone, Diane E. Herz, Mellor, and Steven Hipple (February 1993); “The labor market
improves in 1993,” by Jennifer Gardner, Hipple, and Nardone (February 1994); “Strong employment gains continue
in 1994,” by Lois M. Plunkert and Howard V. Hayghe (February 1995); “Slower economic growth affects the 1995
labor market,” by Gardner and Hayghe (March 1996); “Employment in 1996: jobs up, unemployment down,” by
William C. Goodman and Randy E. Ilg (February 1997); “Strong job growth continues, unemployment declines in
1997,” by Ilg and Clinton (February 1998); “Job growth slows during crises overseas,” by Goodman and Timothy D.
Consedine (February 1999), referring to economic problems in Asia that reduced foreign demand for U.S. goods;
and, despite the previous article, “The job market remains strong in 1999,” by Jennifer L. Martel and Laura A.
Kelter (February 2000). Taken together, the articles in the series clearly proclaimed the overall strength of the
economy from 1990 to 2000.
The “annual story” was not the only employment story that the Review covered in the 1990s. Another “story” was
nonstandard employment, also known as contingent employment. The October 1996 issue was given over entirely
to examining this topic. Anne E. Polivka began the discussion with two articles: “Contingent and alternative work
arrangements, defined” and “A profile of contingent workers.” The first article identified a contingent worker as a
worker without a contract, either explicit or implicit, for a long-term work arrangement and found that, under one set
of criteria, there were 6 million contingent workers in the United States, making up about 5 percent of the
workforce. The second article found that contingent workers were more likely than noncontingent workers to be
female, Black, young, enrolled in school, and employed in either services or construction; more than 10 percent
were teachers. Then came an article by Hipple and Jay Stewart on “Earnings and benefits of contingent and
noncontingent workers” which found, perhaps to no one’s surprise, that contingent workers generally earned less,
and were less likely to receive health insurance and pension benefits through their employers, than were
noncontingent workers—although many had access to health insurance from other sources. Next, Sharon R.
Cohany’s “Workers in alternative employment arrangements” reported that people working as independent
contractors, temporary help agency workers, contract company workers, or on-call workers differed not only from
workers in traditional arrangements, but also from one another in level of education and job security. Then,
returning with another article, “Earnings and benefits of workers in alternative work arrangements,” Hipple and
Stewart found that, although temporary help agency workers and on-call workers earned less than workers in
traditional arrangements, contract company workers and independent contractors earned more.
Finally, Polivka and Donna S. Rothstein ended the conversation that took place in the October 1996 issue with a
pair of articles that delved into the reasons for, and consequences emanating from, entering into an alternative
work arrangement. In Polivka’s “Into contingent and alternative employment: by choice?,” the author sought to
adjudicate between two opposing views: that being in a contingent arrangement consigns a worker to the bottom
of the economic ladder, with frequent job changes, little economic security, and no prospect of advancement; and
that contingent arrangements offer workers both pathways into the labor market and the flexibility to balance work
with other obligations. What she found was that neither picture told the whole story: workers enter into
nonstandard arrangements for many reasons, and although, indeed, some workers find themselves in such

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

arrangements involuntarily, they are relatively few in number; for many others, a contingent work arrangement
presents an opportunity to work that might otherwise be unavailable. Rothstein’s findings, set forth in her article
“Entry into and consequences of nonstandard work arrangements,” complement and add to Polivka’s. Rothstein,
too, found that workers enter into nonstandard arrangements for a host of reasons, two important ones for women
being the birth of a child and a change in marital status. Rothstein also found that women are more likely than men
to be in a contingent arrangement, especially if they had given birth during the previous 2 years—a finding implying
that such arrangements “provide more flexibility than full-time regular work arrangements.”
The employment situation of 2000–15 was in stark contrast to that of the 1990s. From the turn of the century to the
present, events, many unprecedented, unfolded at a rapid pace, and their economic effects were profound. The
bursting of the dot-com bubble in March 2000 and the subsequent stock market downturn; the terrorist attacks of
September 11, 2001, and another stock market tumble; the Iraq War; Hurricane Katrina in August 2005; the
subprime-mortgage crisis, the financial crisis, and yet a third stock market plunge, beginning in mid-2007; the
Great Recession of 2007–09; and later events—all had an economic impact, and the Review covered those
impacts in one or another article over the 15 years since 2000. Three articles by Michael Dolfman and colleagues
were particularly noteworthy. In the June 2004 issue, Dolfman and Solidelle F. Wasser’s article, “9/11 and the New
York City economy: a borough-by-borough analysis,” gave a detailed account of the economic aftermath of the
attack on employment in each of the city’s five boroughs. Focusing on Manhattan, they found that the export sector
—the most internationally oriented part of the city’s economy—was hardest hit, with the finance and insurance
industry losing more than 35,000 jobs since 2000; professional, technical, and scientific occupations about 34,000;
the information industry more than 23,000; and manufacturing more than 11,000. The other boroughs fared better
than Manhattan, but still did not necessarily go unscathed: Queens lost more than 7,000 jobs in the scheduled air
transportation industry and about 17,500 in export-related industries; Brooklyn about 4,500 in total; the Bronx
1,300 manufacturing jobs; and Staten Island less than 1,000 (with the caveat that not all of the jobs lost in
Brooklyn, the Bronx, and Staten Island could be definitively attributed to the attack itself).
Two years later, following up on the earlier article, Dolfman, Wasser, and Kevin Skelly evaluated “Structural
changes in Manhattan’s post-9/11 economy” in the October 2006 issue of the Review. Concluding that
employment bottomed out in Manhattan in 2005, they found that the borough nonetheless increased its role as a
wage generator, with high wages in the global, or export, sector driving demand in the local sector—those
industries supporting the global sector. In effect, although many jobs had been lost to the terrorist attack, wages in
certain industries and occupations—primarily financial and insurance; information; professional, scientific, and
technical services; management of companies; and real estate and leasing—had risen to such an extent that the
former influence of employment as a driver of the local economy was eclipsed by the new role of wages. As a
result, the global sector of Manhattan’s economy has diminished in importance and the local sector has advanced.
“In other words,” said the authors, “the rising income in the global sector is what is spurring demand for more labor
intensive local-sector jobs.”

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Less than 1 year later, in the June 2007 issue, Dolfman, Wasser, and Bruce Bergman assessed “The effects of
Hurricane Katrina on the New Orleans economy.” What they found was far worse than what they had concluded
about New York: the New Orleans economy was devastated, with 105,000 jobs, or roughly 42 percent of all jobs,
lost a year after the hurricane. In industry after industry, occupation after occupation, the figures were eye opening:
10 months after the hurricane, tourism was down by almost 23,000 jobs; healthcare by about 13,500; port
operations by 3,500; educational services by nearly 2,000; and professional, scientific, and technical services by
just under 1,700. Construction alone registered large employment gains—almost 5,000 jobs—and that was due
solely to recovery efforts that got under way after the industry lost jobs immediately after the hurricane. Using
location quotients, the authors nevertheless judged New Orleans’ prospects for recovery as favorable: tourism,
port operations, and educational services—despite suffering large employment losses—had survived as a
foundation for the future, and that “triumvirate source of economic strength…bodes well for the future.”
Dolfman and colleagues were not alone in examining the employment effects of Hurricane Katrina: a year after the
cataclysm, the Review’s entire August 2006 issue was devoted to the tragedy. An overview preceded six articles in
which a number of authors assessed various aspects of the employment impact of the hurricane, not just on New
Orleans, but also on other jurisdictions along the U.S. gulf coast. “The labor market impact of Hurricane Katrina: an
overview” introduced the articles that followed, presenting essentially a visual essay of many salient facts
surrounding the catastrophe. Then, Richard L. Clayton and James R. Spletzer discussed “Worker mobility before
and after Katrina,” finding that a number of the many workers displaced from New Orleans by the storm quickly
found jobs in Texas but still suffered a substantial decline in their short-term earnings. Next, Molly Garber, Linda
Unger, James White, and Linda Wohlford analyzed “Hurricane Katrina’s effects on industry employment and
wages” in 11 affected areas of Louisiana and Mississippi, and not only found that jobs were still down 1 year after

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

the hurricane, but also reported that BLS efforts to continue normal data collection and publication schedules and
to adjust normal estimation and imputation procedures were relatively successful in getting accurate information on
the employment effects of the storm. Following Garber and colleagues’ article was “The Current Population Survey
response to Hurricane Katrina,” by Lawrence S. Cahoon, Diane E. Herz, Richard C. Ning, Anne E. Polivka, Maria
E. Reed, Edwin L. Robison, and Gregory D. Weyland, which presented the results of a BLS–Census Bureau
collaboration showing that “jobless rates were sharply lower for those evacuees who returned home than for those
who did not.” Then, Sharon P. Brown, Sandra L. Mason, and Richard B. Tiller found that “The effect of Hurricane
Katrina on employment and unemployment” was severely depressed employment levels, and temporarily higher
unemployment levels, in Louisiana and Mississippi. Next, in the fifth article of the set, “Conducting the Mass Layoff
Statistics program: response and findings,” Brown and Patrick Carey touted the “careful collaboration between
BLS and State agencies” in overcoming data collection challenges and identifying the accommodation and food
services sector in Louisiana and Mississippi as the sector with the most mass layoffs due to Katrina. Finally,
rounding out the discussion of the hurricane, Charles S. Colgan and Jefferey Adkins examined the extent of
“Hurricane damage to the ocean economy in the U.S. gulf region in 2005,” finding that gulf coast counties and
parishes affected by Hurricane Katrina—and, close to a month later, Hurricane Rita— sustained the largest insured
dollar losses in a year from catastrophes of that nature. As a measure of the momentousness of the loss, the
authors noted that the affected counties and parishes constituted 80 percent of employment in Louisiana, 33
percent in Texas, 14 percent in Alabama, and lesser, though by no means negligible, percentages in Florida and
Mississippi.
The Great Recession, rivaling 9/11 as the most significant U.S. event of the 21st century to date, received an
abundance of coverage that stretched over several years of issues of the Review. It all started with two March
2008 articles that, although published after the recession began, were actually harbingers of it, because, at the
time, the recession had not yet been officially declared. First, reporting the CPS annual story in an article titled
“Household survey indicators weaken in 2007,” James Marschall Borbely informed readers that “unemployment
rose, employment growth slowed, and the labor force participation rate and employment–population ratio trended
down.” Then, in “Payroll employment in 2007: job growth slows,” Robyn J. Richards seconded Borbely’s findings
as she reported the CES annual story. Richards found that “employment grew by just 0.8 percent in 2007, the
lowest rate in 4 years,” with the 1-month diffusion index for total private employment falling below 50 for the first
time since 2003—meaning that more industries lost than added jobs over the year. After that, the recession picked
up full steam, officially beginning in December 2007 and ending in June 2009,4 and the BLS annual stories
recorded its progress during that time every spring from 2009 through 2011. Once again, the titles of the articles
are by themselves descriptive: “U.S. labor market in 2008: economy in recession” (Borbely, March 2009);
“Substantial job losses in 2008: weakness broadens and deepens across industries” (Laura A. Kelter, March
2009); “Job openings and hires decline in 2008” (Katherine Klemmer, May 2009); “Payroll employment in 2009: job
losses continue” (Megan M. Barker and Adam A. Hadi, March 2010); “Job openings, hires, and separations fall
during the recession” (Mark deWolf and Klemmer, May 2010); “ Unemployment remains high in 2010” (Eleni
Theodossiou and Steven F. Hipple, March 2011); and “Payroll employment turns the corner in 2010” (John P.
Eddlemon, March 2011). Interspersed among the various annual stories were two broader, but still Great
Recession–related articles. In the first, in April 2009, Harley J. Frazis and Randy E. Ilg, analyzing “Trends in labor
force flows during recent recessions,” found that the Great Recession (still in progress at the time of publication)
was characterized by a decrease in flows into employment from March 2007 to December 2008 and a decrease in
flows out of unemployment in mid-2007, consistent with “a prolonged slowdown in job creation occurring alongside
10

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

an increase in job destruction.” Thus, maintained the authors, the Great Recession “differs from…most earlier
recessions, which were marked more by increasing flows out of employment.” In the second broad article, “The
nation’s underemployed in the ‘Great Recession’ of 2007–09,” Andrew Sum and Ishwar Khatiwada used data from
the CPS to show that “the less educated, those in low-skilled occupations, and those in low-paying occupations
had a higher incidence of underemployment during the 2007–09 recession.” The authors also found that
underemployment was concentrated among workers from lower income households.
As a measure of the importance the Review attached to the Great Recession, the April 2011 issue was devoted
entirely to that subject. A piece titled “Employment loss and the 2007–09 recession: an overview,” by Christopher J.
Goodman and Steven M. Mance, set the stage for nine articles that examined the effect of the recession on
employment in various industries. Observing that “the downturn in employment accompanying the 2007–09
recession was notable for its prolonged length, for affecting an especially wide range of industries, and for being
deeper than any other downturn since World War II,” the overview pointed out that, as of December 2010, a year
and a half after the trough of the recession, employment remained 7.7 million jobs below the prerecession peak.
Next, surely by design, the article “Employment in health care: a crutch for the ailing economy during the 2007–09
recession,” by Catherine A. Wood, set a positive tone, letting the reader know at the outset that there was at least
one bright spot in the economy: through all the job losses, the healthcare industry grew by 428,000 jobs during the
recession and continued to grow steadily thereafter. Wood concluded her exposition with a noteworthy
observation: “that [the healthcare industry] boosted employment at all during such a severe and prolonged
economic downturn is remarkable.” Following Wood’s article came Brian Davidson’s piece “Mining employment
trends of 2007–09: a question of prices,” in which the author noted that the mining industry managed to sustain job
growth through the first 10 months of the recession, buoyed by markedly higher oil prices; after that, when oil
prices fell substantially, employment fell as well, reaching a trough 4 months after the recession had ended. In
sum, the period of falling employment in mining was 6 months shorter than the duration of the recession.
Then came the parade of unpalatable news, leading off with the worst case of all: the construction industry. The
title of Adam Hadi’s article succinctly told it all: “Construction employment peaks before the recession and falls
sharply throughout it.” Job losses amounted to 1.5 million, a 20-percent decline in employment in the industry, the
largest percent decline of all industries during the Great Recession. Not far behind, however, was manufacturing:
fleshing out her article, “Manufacturing employment hard hit during the 2007–09 recession,” Megan M. Barker cited
the loss of 2 million employees, or 15 percent of the industry’s workforce, as indicative of the ongoing dire straits of
manufacturing, an industry on a downward trend since 1979, with job losses accelerating during every recession
thereafter. Then, in “Professional and business services: employment trends in the 2007–09 recession,” Frank
Conlon related that industry’s woes during the recession, including the loss of 1.6 million jobs, or nearly 9 percent
of the industry’s workforce. The loss was the largest ever recorded in the industry, in both percentage and absolute
number. Most striking were the losses in the administrative and waste services component of the industry, a
component that accounted for 3 of every 4 jobs lost in professional and business services. Next in the parade of
industries pummeled by the recession was the finance industry,5 whose troubles were aptly described in the title of
George Prassas’ article: “Employment in financial activities: double billed by housing and financial crises.” Indeed,
the industry had received a twofold hit: first, preceding the recession by more than a year and a half, the housing
market bubble burst after a peak in employment in the real estate credit industry and among mortgage and
nonmortgage brokers and real estate brokers in April 2006; then, in October 2008, after the start of the recession,
the financial markets began to experience large losses that lasted until almost a year after the National Bureau of

11

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Economic Research declared the recession over. All told, employment in the finance industry fell by 473,000, or
5.8 percent, during the recession.
Retail trade was next in the discussion of industries hit by the recession. Michael D. McCall described the “Deep
drop in retail trade employment during the 2007–09 recession,” a slide amounting to a loss of a little more than a
million jobs, or 6.7 percent of the industry’s employment. Motor vehicle and parts dealers lost the most jobs in the
retail trade industry, 271,000, followed by clothing and clothing accessories stores, which shed 161,000 jobs.
Then, in “Employment in leisure and hospitality departs from historical trends during 2007–09 recession,” Eliot
Davila told how the leisure and hospitality industry suffered the most severe and prolonged downturn in
employment in a long history in which the industry had experienced almost uninterrupted job growth. Finally,
Steven Kroll wound up the discussion with an article on “The decline in work hours during the 2007–09 recession,”
revealing the not-unexpected finding that average weekly hours for workers in private industry decreased across
all industries during the recession. Hours were pulled down further, said Kroll, “as a result of heavy job losses in
industries with above-average workweeks.” Summing up the effect of the Great Recession and the economic
situation as of December 2010, 4 months before publication of the series of articles in the Review, Goodman and
Mance said, in their overview, “The U.S. economy is recovering from one of the longest and deepest recessions
since the end of World War II…[in which] virtually no area of the economy remained unscathed,…particularly the
labor market.”
After the April 2011 issue, the Review continued to report on employment in a now recovering (though modestly)
economy. Following a gloomy August 2011 assessment titled “Job openings and hires show little postrecession
improvement,” by Katherine Bauer Klemmer and Robert Lazaneo, the titles of the publication’s “annual stories”
reveal the slow upward progression of employment from March 2012 to the present: “Payroll employment growth
accelerates in 2011” (Parth A. Tikiwala and Frank Conlon, March 2012—the CES annual story); “U.S. labor market
shows gradual improvement in 2011” (Eleni Theodossiou, March 2012—the CPS annual story); “Job openings and
hires continue to show modest changes in 2011” (Guy L. Podgornik, September 2012—the Job Openings and
Labor Turnover Survey, or JOLTS, annual story); “Slow and steady: payroll employment grew moderately in
2012” (Sutton E. Puglia and Parth A. Tikiwala, March 2013); “U.S. labor market continued to improve in 2012” (Lisa
Williamson, March 2013); “Job openings continue to grow in 2012, hires and separations less so” (Kendra C.
Hathaway, May 2013); “Nonfarm employment continued its road to recovery in 2013” (Kara Sullivan, March 2014);
“Unemployment continued its downward trend in 2013” (Catherine A. Wood, April 2014); “Continued improvement
in U.S. labor market in 2014” (Eleni Theodossiou Sherman and Janie-Lynn Kang, April 2015); “CES employment
recovers in 2014” (John P. Mullins and Brittney E. Forbes, April 2015); and “Job openings reach a new high, hires
and quits also increase” (Kevin S. Dubina, June 2015). The Review will continue to report regularly on employment
—the most basic aspect of the economy—for the foreseeable future.

Labor force
The labor force has been a key area of investigation for the Review since its inception. A sampling of articles from
the 1980s, the 1990s, and the period from 2000 to 2015 reveals a variety of labor force–related topics examined
from decade to decade: the demographic mix of the labor force, the labor force participation rate, the youth labor
force, the employment-to-population ratio, the women’s labor force, working mothers, the older labor force, baby
boomers, the racial and ethnic composition of the labor force, and veterans in the labor force, to mention just a
few. Articles from the 1980s include “The employment–population ratio: its value in labor force analysis” (February
12

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

1981), by Carol Boyd Leon, who touted the consistent cyclical properties of that statistic and the accuracy of its
seasonal adjustment in measuring the ability of the economy to provide jobs; “The labor market problems of older
workers” (May 1983), in which Philip L. Rones observed that “older workers do not have especially high
unemployment rates, but when they become unemployed, they are less likely to find a job, and more likely to leave
the labor force in discouragement”; Ellen Sehgal’s “Foreign born in the U.S. labor market: the results of a special
survey” (July 1985), which found that, despite initial hardship, recent entrants to the U.S. labor force saw their
employment and earnings approach those of native-born workers with the passage of time; “An international
comparison of labor force participation, 1977–84” (May 1986), Patrick J. McMahon’s comprehensive look at labor
force participation in six countries; and “Labor force status of Vietnam-era veterans” (February 1987), by Sharon R.
Cohany, who presented the results of a special survey which found that men who served in Southeast Asia—
especially those with service-connected disabilities—encountered labor market difficulties substantially greater
than those of nonveterans of the same era.

Among labor force articles published in the 1990s were Paul O. Flaim’s “Population changes, the baby boom, and
the unemployment rate” (August 1990), an article showing that, because of their large numbers, baby boomers
(those born between 1946 and 1964) had a substantial effect on the unemployment rate. In the late 1960s and
throughout most of the 1970s, they put upward pressure on the rate, with many of them having entered the job
market as teenagers and twenty-somethings looking for employment. In the 1980s, they exerted downward
pressure on the unemployment rate, when, in their thirties and forties, they found employment, and remained
employed, more easily. Flaim projected that the baby boomers would continue to put downward pressure on the
unemployment rate in the 1990s. Other articles of the same decade were “Working and poor in 1990” (December
1992), by Jennifer M. Gardner and Diane E. Herz, an article unequivocally linking poverty in working families
chiefly to their low wages—an issue receiving much attention today—but also to a small number of workers in the

13

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

family and to families maintained by women; Herz’s solo article examining “Work after early retirement: an
increasing trend among men” (April 1995), in which the author found that early pensioners were returning to work
at a faster pace than in the previous decade, a transition likely attributable to a number of factors, including
changes in the types and provisions of pensions, increases in healthcare costs, and longer life expectancies spent
more healthily; Alexander Kronemer’s intriguing “Inventing a working class in Saudi Arabia” (May 1997), an
international report on that country’s unusual labor force, in which 9 of every 10 private sector workers were
foreigners, and on the Kingdom’s effort to replace as many of them as possible with qualified Saudis in the face of
a culture that long has had a “strong distaste for the kinds of work found in most manufacturing and office-support
jobs”; and “The long-term consequences of nontraditional employment” (May 1998), by Marianne A. Ferber and
Jane Waldfogel, a followup to the Review’s October 1996 special issue on contingent and alternative work
arrangements. Ferber and Waldfogel found strong evidence indicating that a history of part-time work is associated
with lower pay for both men and women, except for men who were self-employed and women who were working
part time voluntarily.
The years from 2000 to 2015 saw the Review continue its examination of the many aspects of the labor force. A
sample of articles published during those years brings out the flavor of the issues discussed. Ten years after
Flaim’s piece on the initial negative and then positive unemployment effects of the baby boomers transitioning into
older age groups, Arlene Dohm picked up the story in her article “Gauging the labor force effects of retiring babyboomers” (July 2000). She began by observing that the boomers were approaching retirement age and their effect
on both the overall economy and certain occupations and industries, as in all the previous decades, would be
substantial. Then, she contemplated the prospect that many younger workers might not have the relatively high
level of skills required to fill the jobs vacated by retiring baby boomers. Dohm went on to say that the industries
most affected—those with the greatest percentage of workers 45 years and older who were likely to retire or
otherwise permanently leave the occupation—would be educational services, public administration, transportation,
and health services. She was encouraged, however—and in this, she was right, as was borne out in later Review
articles—by indications that older workers were delaying their departure from the labor force or returning to it after
retirement.
In another study of the labor force, Abraham Mosisa examined the “Labor force characteristics of secondgeneration Americans” (September 2006), focusing on the children of foreign-born workers. Comparing secondgeneration Americans—those with one or both parents foreign born—with their counterparts in third and higher
generations, he found that “the second generation is more racially diverse than the third generation; and secondgeneration individuals tend to have higher levels of education than their third-generation counterparts.” Thus, wrote
Mosisa, “it appears that members of the second generation of American workers have achieved [labor market]
parity with their third-generation counterparts; indeed, in some respects, they may have become more successful.”
He went on to point out that a key factor in the second generation’s success in the workforce is its level of
educational attainment: “The second generation has taken advantage of access to education,” and “38.0 percent
of those aged 25 to 54 years have at least a bachelor’s degree, compared with 29.7 percent of the third
generation.” In this regard, continued Mosisa, “second-generation workers are somewhat more likely than thirdgeneration workers to be employed in professional and related occupations, and in management, business, and
financial operations. The median annual earnings of second-generation workers are somewhat higher than those
of their third-generation counterparts.”

14

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

By 2010, Asians in the nation numbered 11.2 million and accounted for 4.7 percent of the U.S. civilian
noninstitutional population ages 16 and older. There were 7.2 million Asians in the labor force, and their
participation rate was almost 66 percent. In the November 2011 issue of the Review, Mary Dorinda Allard captured
the group’s diversity—and similarities—in her article “Asians in the U.S. labor force: profile of a diverse population.”
Examining the labor force statistics of the six major Asian groups populating the nation, Allard arrived at the
following results distinguishing the groups from one another: Chinese (22 percent of all Asians) workers were
overrepresented in food preparation and serving-related occupations and in computer and mathematical
occupations; Indians (18 percent) were most likely to be foreign born, be married, and have a bachelor’s degree or
higher; among Filipinos (17 percent), women outnumbered men and had a higher labor force participation rate
than women in any other Asian group and about a third of workers were in the health care and social assistance
industry; Vietnamese (11 percent) were the least likely of the groups to have a bachelor’s degree or higher, and
about one-fifth of Vietnamese workers were in personal care and service occupations; about one-quarter of
Korean (10 percent) workers were self-employed, a much higher percentage than that for any other Asian group;
and, unlike the other Asian groups, the Japanese (6 percent) were mostly native born, and they were more likely to
be ages 55 and older. (Other Asians, including Thais, Pakistanis, Cambodians, Hmong, and Laotians, together
made up the remaining 16 percent.) With regard to similarities, probably the most important economic one that
Allard found was that all the Asian groups suffered increased unemployment rates during the Great Recession—
though less so than non-Asians.
Rounding out the sample of Review articles on the labor force published in the period from 2000 to 2015 is an
October 2012 entry by Alix Gould-Werth and H. Luke Shaefer examining “Unemployment Insurance participation
by education and by race and ethnicity.” In it, the authors analyzed results from the 2005 Unemployment Insurance
(UI) Non-Filers Supplement to the CPS in order to learn “whether application for and receipt of benefits among
applicants varies systematically with two key demographic characteristics” of the labor force: “educational
attainment, and race and ethnicity.” Regarding education, the highly educated were more likely than the less
educated to apply for UI benefits and to receive benefits if they applied. If they perceived themselves as ineligible
for benefits (and therefore didn’t apply to the UI program), the highly educated were again more likely than the less
educated to attribute their perceived ineligibility to voluntarily quitting their job. By contrast, the reason given by
most of the less educated unemployed workers for perceiving themselves as ineligible for UI benefits (and hence
not applying for them) was that they did not work enough or earn enough to qualify. The latter finding suggests that
less educated workers may lack a sufficient understanding of UI eligibility criteria and is therefore an impetus for
further research.
With regard to race and ethnicity, Hispanic respondents to the survey were far less likely than White non-Hispanic
or African American respondents to apply for UI benefits and, when they did apply, to receive benefits. The authors
ruled out noncitizenship as a reason for both disparities, because both persisted when the Hispanic sample was
restricted to citizens. The authors found statistically significant differences between Hispanic respondents, on the
one hand, and White non-Hispanic respondents and African American respondents, on the other, in a number of
areas: (1) a greater proportion of Hispanic respondents than both White non-Hispanic respondents and African
American respondents indicated that they did not know where or how to apply as a reason for failing to file; (2) a
greater proportion of Hispanic respondents than both White non-Hispanic respondents and African American
respondents also indicated that they did not know that benefits existed; and (3) more than 5 percent of Hispanic
respondents (but no Hispanic citizens) listed inability to speak English as a reason for not filing, compared with

15

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

one-quarter of 1 percent of White non-Hispanic respondents and no African American respondents. All of these
findings suggest that future studies of application for, and receipt of, UI benefits should examine Hispanic workers
separately from African American workers and, indeed, from other minority workers in order to understand what
factors are driving the differences between them.

Conclusion
Since 1980, the Review’s coverage of inflation, employment, and the labor force has been extensive. Although
these topics were featured consistently in earlier periods, their treatment in the journal’s more recent issues has
added more to our understanding of, among other things, the causes and effects of inflation, the employment
impacts of recessions and other economically disruptive events, and the imprint of demographic change on the
labor force. The final installment in the series will focus on three additional topics discussed extensively in the
Review since 1980: employee benefits, industries and occupations, and worker safety and health.
Note: Monthly Labor Review articles published since 1980 are available online at https://www.bls.gov/opub/mlr/
2016/home.htm.
SUGGESTED CITATION

Brian I. Baker, "The Monthly Labor Review at 100—part III: inflation, employment, and the labor force since 1980,"
Monthly Labor Review, U.S. Bureau of Labor Statistics, June 2016, https://doi.org/10.21916/mlr.2016.27.
NOTES
1 See “Table of historical inflation rates by month and year,” Historical inflation rates: 1914–2016 (San Antonio: Coinnews Media
Group, January 20, 2016), http://www.usinflationcalculator.com/inflation/historical-inflation-rates.
2 Norwood referred to the PCE deflator as a product of the Commerce Department, Triplett as a product of the BEA. In actuality, it is
both: the BEA is an agency of the Commerce Department.
3 “Table of historical inflation rates by month and year.”
4 According to the National Bureau of Economic Research, the official arbiter of recession beginning and ending dates.
5 North American Industry Classification System sectors 52 (finance and insurance) and 53 (real estate and rental and leasing).

RELATED CONTENT

Related Articles
The Monthly Labor Review at 100—part II: the “middle years,” 1930–80, Monthly Labor Review, May 2016.
The Monthly Labor Review at 100—part I: the early years, 1915–30, Monthly Labor Review, May 2016.
Celebrating 100 years of the Monthly Labor Review, Monthly Labor Review, July 2015.
The Monthly Labor Review turns 100, Monthly Labor Review, July 2015.

Related Subjects
Bureau of labor statistics

History

Labor and economic history

16

June 2016

Measuring quarterly labor productivity by industry
Timely statistics on output, employment, and productivity
are essential to understanding the performance of the U.S.
economy. This study examines newly available quarterly
GDP-by-industry statistics to determine whether they can
be used to produce reasonable quarterly labor productivity
measures at the industry level. The results show that the
quarterly labor-productivity data at the industry level can
provide insights into which industries are driving current
aggregate economic performance. However, the quarterly
industry data are highly volatile and are most useful when
evaluated in conjunction with long-run trends in order to
Lucy P. Eldridge
eldridge.lucy@bls.gov

more precisely assess the business cycle dynamics.
Timely statistics on output, employment, and productivity
are essential to understanding the performance of the U.S.
economy. Labor productivity indicates how effectively labor

Lucy P. Eldridge is the Associate Commissioner
of the Office of Productivity and Technology, U.S.
Bureau of Labor Statistics.

inputs are converted into output and provides information
needed to assess changes in technology, labor share, living
standards, and competitiveness. The U.S. Bureau of Labor
Statistics (BLS) produces both quarterly labor productivity
measures for broad sectors of the U.S. economy and
annual labor productivity measures for industries.1
Quarterly labor productivity data are analyzed as indicators

Jennifer Price
price.jennifer@bls.gov
Jennifer Price is an economist in the Office of
Productivity and Technology, U.S. Bureau of
Labor Statistics.

of cyclical changes in the economy and are closely watched
by the financial community, nonfinancial businesses,
government policymakers, and researchers. Industry-level productivity statistics provide a means for comparing
trends in efficiency and in technological improvements across industries, and indicate which industries are
contributing to growth in the overall economy. Although annual industry productivity data can be used to analyze
past industry performance and long-term trends, they are not frequent enough to provide indicators of current
industry performance or identify which industries are driving current aggregate economic performance. Industrylevel labor input data are available on a quarterly basis, but corresponding quarterly industry-level output data for
nonmanufacturing industries—data that are necessary for constructing labor productivity measures—have not
been available until recently.

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

In April 2014, the U.S. Bureau of Economic Analysis (BEA) began releasing quarterly gross domestic product
(GDP)-by-industry measures.2 These new output measures were developed to be consistent with the annual
industry accounts, and they appear to provide the data needed to construct more timely labor productivity
measures. However, because complete output data are not yet available for all industries on a quarterly basis,
these higher frequency data rely on assumptions about the relationships among industry inputs, outputs, and value
added from the annual and benchmark statistics. This study examines the new quarterly GDP-by-industry statistics
to determine whether they can be used to produce reasonable quarterly labor productivity measures at the industry
level. This study develops quarterly labor hours and labor productivity measures for the 20 private industry groups
for which BEA is releasing GDP-by-industry data.3 In addition, the study evaluates the volatility in the quarterly
productivity measures to determine the value of these industry data for better understanding the sources of
economic growth—in order to provide recommendations.

BLS labor productivity measures
The preliminary and revised quarterly press release—“Productivity and Costs”—includes measures of labor
productivity for six major U.S. sectors: business, nonfarm business, manufacturing, durable and nondurable goods
manufacturing, and nonfinancial corporations.4 Labor productivity measures are calculated as growth in real output
relative to growth in hours worked. BLS calculates quarterly labor productivity for the business and nonfarm
business sectors by combining real output from the National Income and Product Accounts (NIPA), produced by
the BEA, with measures of hours worked, prepared by the BLS Productivity Program. Output for the business
sector is estimated as GDP less the output of general government, nonprofit institutions, and the household sector
(including owner-occupied housing). Because input-cost measures are used to measure the output for general
government services, the household sector, and nonprofit institutions, the trends in these output measures will, by
definition, move with measures of input data and will tend to imply little or no labor productivity growth. Therefore,
the business sector is the most aggregate sector for which reliable measures of productivity can be produced.5
Nonfarm business sector output further excludes the output of the farm sector, while the nonfinancial corporate
sector even further excludes the output of unincorporated businesses and those corporations classified as offices
of bank holding companies, offices of other holding companies, or offices in the finance and insurance sector.6
For the U.S. manufacturing sector, as well as for individual manufacturing industries, output is estimated by
aggregating U.S. Census Bureau industry shipments data to obtain gross output and then removing transactions
that occur within the sector or industry (intrasector or intra-industry transfers). This approach creates a measure of
sectoral output that excludes those outputs produced and consumed within the sector or industry. To derive
quarterly estimates from the annual manufacturing indexes, BLS adjusts the annual totals with the use of a
quarterly reference series and a quadratic minimization formula.7 The quarterly reference series is constructed
from the Federal Reserve monthly indexes of Industrial Production.8
Studies of output per hour in individual industries have been produced by BLS since the late 1800s. The BLS
industry productivity program has evolved from producing industry-specific studies to the regular publication of
annual measures of labor productivity for 199 unique 3- and 4-digit NAICS industries.9 BLS researchers construct
industry output measures by using data primarily from the economic censuses and annual surveys of the U.S.
Census Bureau, together with information on price changes primarily from BLS. Real output is most often derived

2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

by deflating nominal sales or values of production with the use of BLS price indexes and removing intra-industry
transactions; however, for a few industries, output is measured by physical quantities of output.10

Quarterly output by industry
BEA GDP-by-industry data are available from 2005 to the present, with data for the most current quarter released
120 days after the end of the reference quarter. BEA began working on the prototype for quarterly GDP-by-industry
data in 2007, and the measures have evolved over the past 8 years to reflect improved techniques.11 The quarterly
data were developed to be consistent with the methodology used to construct time series estimates of the annual
industry accounts, which are an extension of the annual input–output (I–O) accounts. The I–O accounts consist of
two basic national accounting tables: a make table and a use table. The make table shows the production of goods
and services by industry; the sum of the entries across all industries is the total output of commodity throughout
the domestic economy. The use table shows the consumption of goods and services by each domestic industry
and by final users. The use table also shows the compensation of employees; taxes on production and imports,
less subsidies; and gross operating surplus. Together, these three components compose total value added. The
make and use tables are constructed from various data sources and are balanced to align the estimates of
industry inputs, outputs, and value added across the economy.12
GDP by industry is a key component of the annual industry accounts, measuring each domestic industry’s
contribution to GDP.13 BEA uses the annual I–O table and annual GDP-by-industry measures as the starting point
for creating quarterly GDP-by-industry estimates. BEA describes five steps taken to estimate quarterly GDP by
industry: develop domestic supply by commodity, construct value added by industry, prepare initial use tables,
balance use tables, and estimate price and quantity indexes for GDP by industry.14 The five steps are described as
follows:
1. Measures of domestic supply by commodity—representing the value of goods and services produced by
domestic firms, plus imports and government sales, less exports and changes in inventory—are developed
from various monthly and quarterly surveys, and tested and adjusted for seasonality.
2. Data on value added by industry—representing the costs incurred and the incomes earned in production—
are estimated with the use of compensation of employees by industry; taxes on production and imports, less
subsidies; and gross operating surplus.
3. An initial use table—showing the consumption of intermediate inputs and final uses—is constructed for each
quarter with the use of the available annual use table for the year and is revised during annual revisions.
4. A balancing procedure is applied to ensure that each industry’s output equals its intermediate inputs plus its
value-added components and that the sum of intermediate and final uses for each commodity is equal to the
industry's gross output.
5. Finally, the initial nominal industry and commodity gross output, intermediate inputs, and value-added
results, and the corresponding quantity and price indexes are then interpolated (i.e., benchmarked) with
respect to the most recently published annual data in accordance with the Denton proportional firstdifference method. A double-deflation method is used to allow relative prices to affect output and
intermediate uses differently. Real value added is computed as the difference between real output and real
intermediate inputs.15

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

The new GDP-by-industry data provide more timely information on accelerations and decelerations in economic
growth at the industry level. These data are a useful addition to the annual industry accounts that BEA publishes.
This study makes use of the BEA 2005–14 quarterly output data published on January 22, 2015.

Output concepts
Labor productivity can be computed by using two different representations of output: sectoral or value added.16
Sectoral output is a broader measure of output that removes intermediate inputs produced within an industry or
sector from gross output—the total value of goods and services produced by an industry or sector. As previously
mentioned, GDP is a value-added measure of output and is equal to gross output less all purchased intermediate
inputs.
BLS prefers to use the sectoral output concept when measuring economic growth. This approach acknowledges
that changes in the price, quality, and availability of intermediate inputs will influence a firm’s decision concerning
its use of capital and labor.17 As such, any changes in labor productivity may be due to technological progress,
economies of scale, improved management techniques, and increased skills of the labor force, as well as changes
to nonlabor inputs produced outside the industry or sector (i.e., capital services, energy, purchased intermediate
materials, and purchased services).18 Labor productivity based on a sectoral output concept will therefore increase
with outsourcing and with improvements in the quality of purchased intermediate inputs. If these purchased
intermediate inputs are excluded from the value of output, they can no longer be a source of productivity growth.19
Therefore, BLS labor productivity measures for the manufacturing sector, individual manufacturing industries, and
NIPA-level nonmanufacturing industries are calculated under a sectoral output approach.
However, there may be circumstances when a value-added output approach to measuring labor productivity,
relating output solely to the primary inputs in production, is beneficial.20 For example, to study the relationship
between growth in wages and labor productivity, a preferred approach may be one that removes outsourcing and
the quality of intermediate inputs from the model.21 Unlike sectoral output measures, value-added output
measures will decline with labor as a result of outsourcing; thus labor productivity will be less affected.22 BLS
measures for business, nonfarm business, and nonfinancial corporate sector labor productivity are constructed
under a value-added approach. Because there few intermediate inputs are purchased from outside these
aggregate sectors, labor productivity measures based on value-added output and those based on sectoral output
will be similar—the largest difference is due to purchased imported materials.23
Data users may need different output concepts for measuring labor productivity, depending upon which questions
they are interested in answering. Value-added productivity measures more closely reflect an industry’s ability to
translate technical change into final income, while sectoral productivity measures more closely reflect the technical
efficiency with which industries transform inputs into output. Because the choice of sectoral output or value-added
output will result in different accelerations and decelerations in measured labor productivity, it is important to be
aware of which method is used when interpreting productivity data. For this study, labor productivity measures are
presented under both the sectoral and the value-added output approach.
To construct sectoral output measures, intra-industry transactions were removed from the BEA quarterly real gross
output-by-industry measures. These intermediate inputs were removed so that output is not overstated relative to
the labor hours used to produce that output. Their removal was accomplished by estimating ratios of sectoral
output to gross output with the use of industry current-dollar data from the BEA annual I–O use tables before

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

redefinition. Intra-industry transactions were calculated as the sum of all outputs that are produced and used within
the same industry group. These transactions were subtracted from gross output, and then a sectoral-output-togross-output ratio was constructed. The annual adjustment ratios for each industry group were converted into a
quarterly series by using a moving-average procedure to smooth the data. Estimates of real sectoral output by
industry were calculated by multiplying the sectoral adjustment ratios by the BEA quarterly real gross output-byindustry data.24

Independence of output and hours data
Because complete data for constructing quarterly I–O tables are not available, BEA relies on assumptions about
the relationships among industry inputs, outputs, and value added from the annual and benchmark I-O statistics to
estimate quarterly output data. Input measures, such as wages from the BLS Quarterly Census of Employment
and Wages (QCEW) or employment from the BLS Current Employment Statistics (CES) program, are available
more frequently than measures of output. Thus, it is important to determine the extent to which BEA uses these
input measures to supplement output data. Although such techniques are suitable for output measurement, they
can be troubling for productivity measurement if input and output measures are not sufficiently independent. If
similar source data are used in measuring inputs and outputs, then, by definition, labor productivity will be biased
toward zero.
Most BEA output measures are constructed from U.S. Census Bureau data; value-of-shipments data are used for
mining and manufacturing, revenues for utilities, sales for wholesale and retail trade, and commissions for
commodity brokerage. BEA makes strong use of the Census Bureau’s Quarterly Services Survey (QSS) and
Service Annual Survey (SAS).25 Industry coverage within the QSS and SAS has been significantly expanded over
the past 10 years, resulting in decreased dependence on input-based data for BEA output measures. Since its
initial publication of quarterly revenue and expenses for selected information industries in the fourth quarter of
2003, the QSS has added data for selected detailed industries within the following industries: health services
(fourth quarter of 2004, first quarter of 2009), professional and business services (third quarter of 2006),
administrative services (third quarter of 2006), transportation (first quarter of 2009, first quarter of 2010), leisure
(first quarter of 2009), other services (first quarter of 2009), finance (third quarter of 2009), utilities (first quarter of
2010), real estate (first quarter of 2010), educational services (first quarter of 2010), and accommodations (third
quarter of 2012).26 The SAS underwent a similar expansion to annual statistics.27 Many of these data have
become available only since 2009.
The direct and indirect use of input-based output data is found, to some extent, in 11 service-providing industries.
Direct use occurs within portions of seven industries, where input data are either used to estimate the initial annual
series or used as an extrapolator to construct the quarterly series. The primary source of input-based output data
for estimation of quarterly current-dollar statistics is the BLS QCEW. The information, real estate, management
services, administrative services, and other services industries all incorporate QCEW data into quarterly output
estimates.
It is difficult to quantify the impact of input-based data that are used indirectly, because such use often represents
only a small portion of the industry measure. Input-based data are used to estimate some price indexes in both the
professional and business services and the educational services industries. Indirect use of input-based data is also
present when estimates are based on NIPA Personal Consumption Expenditures that have been constructed from

5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

input-based data. Industries affected by the indirect use of input-based data include finance and insurance, real
estate, professional and business services, educational services, health services, leisure, and accommodations.28
Gross output for most service-sector industries is derived from QSS data. However, labor productivity measures
should be viewed with caution for those industries where input data are used to construct output measures.
BLS does not consider productivity for the total economy to be a reliable indicator, because of the correlation
between measuring output and measuring labor input for several segments of the economy, especially nonprofit
institutions serving households (NPISH). Because the output of NPISH cannot be measured independently of labor
inputs, productivity measures that include NPISH will have a downward bias.
Information on the presence of NPISH within each industry group is available, allowing data users to estimate
industry output and GDP share of nonprofits. Table 1 shows that NPISH are heavily concentrated in education (78
percent of educational services), health services (89 percent of hospitals and 63 percent of social assistance
programs), leisure services (91 percent of museums, historical sites, and similar institutions), and other services
(76 percent of religious, grantmaking, civic, professional, and similar organizations). From these data, it is
estimated that input-based methods are affecting approximately 5 percent of measured GDP.
Table 1. Impact of nonprofits on gross domestic product (GDP)
Industry group
Information
Professional and
business services
Educational services
Health services

Leisure

Accommodations
Other services

Nonprofit percentage of Industry share of Percent of GDP

Detailed industry affected

output

Broadcasting and telecommunications
Professional, scientific, and technical
services
Educational Services
Ambulatory health care services
Hospitals
Nursing and residential care facilities
Social assistance
Performing arts, spectator sports, and
related industries
Museums, historical sites, and similar
institutions
Amusement, gambling, and recreation
industries
Accommodations
Religious, grantmaking, civic,
professional, and similar organizations

GDP

affected

1.70

2.76

0.01

2.40
78.00
12.60
89.10
40.70
63.30

6.39
1.14
3.17
2.55
.76
.59

.12
.94
.35
2.17
.29
.38

25.10

.48

.04

91.10

.05

.04

22.50
1.90

.45
.72

.05
.00

76.10

.74

.67

Source: Authors’ calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

For the BLS quarterly business sector labor productivity measures, BEA provides aggregate business sector
output and BLS uses data from the Economic Census and BEA to remove labor hours for nonprofits. Because the
new, quarterly BEA industry output data do not exclude output for nonprofit institutions, this study adjusts both
output and hours measures to remove NPISH at the industry level. The data presented show industry measures,
less the nonprofit portion of the industry, corresponding to the private business sector portion of each industry.

6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Figure 1 presents the differences between GDP by industry measured as sectoral output and value-added output
for 2014. The dollar level of sectoral output will always be larger than the dollar level of value-added output. The
difference between the two series represents the value of the intermediate inputs that the industry is purchasing
from outside its borders. The largest differences are found in the manufacturing and real estate industries; the
smallest differences are found in education services, leisure services, and utilities.

Quarterly hours worked, by industry
BLS does not currently publish quarterly hours worked for all people by industry.29 Quarterly hours data have been
constructed for this research study and, unless otherwise noted, closely follow the methods used to calculate
quarterly estimates of hours worked that underlie the business sector productivity data.
The primary source of hours data is the average-weekly-hours-paid series for production workers in goodsproducing industries and for nonsupervisory workers in service-providing industries from the CES program.30 CES
program surveys approximately 146,000 establishments, collecting employment and hours-paid data. Seasonally
adjusted monthly data from the CES are used to construct quarterly averages of employment and quarterly
employment-weighted averages of average weekly hours.31 The CES average weekly hours for production and
nonsupervisory employees32 (AWHPCES ) are adjusted to an hours-worked basis by using an hours-worked-tohours-paid ratio (hwhpPNCS ) estimated from data provided by the National Compensation Survey (NCS).33 The
hours-worked adjustment controls for changes in vacation, holiday, and sick pay. Total hours worked by production
and nonsupervisory employees (HP) are calculated as
(1)

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

where NP is the CES employment of production and nonsupervisory employees.34
Average weekly hours for nonproduction and supervisory workers are estimated by applying a ratio adjustment
from the BLS Current Population Survey (CPS) to the hours data for production and non-supervisory employees.
The CPS ratio is equal to the average weekly hours worked by nonproduction and supervisory employees divided
by the average weekly hours worked by production and nonsupervisory employees.35 This ratio is seasonally
adjusted by means of an X-12-ARIMA program and is combined with the average-weekly-hours-worked series for
production and nonsupervisory employees, as well as CES employment data.36 Total hours worked by
nonproduction and supervisory employees (HNP) are estimated as

(2)
where AWHNPCPSand AWHPCPS represent CPS measures of average weekly hours for nonproduction and
supervisory employees and production and nonsupervisory employees, respectively. NNP is the CES employment
notation for nonproduction and supervisory employees.37
Total hours is the sum of all employee hours and the hours worked by self-employed and unpaid family workers.
Hours worked by self-employed and unpaid family workers are estimated by pooling 3 months of self-reported
weekly hours from the CPS for the 20 major industry groups that match the GDP-by-industry series.38 There are
too few observations from the CPS to construct data on self-employed and unpaid family workers for the
management-of-companies-and-enterprises industry group. Therefore, data on the self-employed and unpaid
family workers are created as the residual of all professional and business services, less professional and
technical services and administrative and waste management services. For the agricultural services industry
group, the category of hours worked on farms is constructed on the basis of CPS data.39 Although the use of
quarterly CPS data for the 20 industries of interest in this study is reasonable, further industry detail on a quarterly
basis may be beyond the limits of the available CPS data.

Quarterly labor productivity by industry
Quarter-to-quarter growth in labor productivity is calculated as quarter-to-quarter growth in output less quarter-toquarter growth in labor hours and is expressed as an annual rate to facilitate comparisons with annual growth
rates. Figures 2 and 3 use the sectoral output and value-added output approaches, respectively, to compare
annual average growth rates of labor productivity with corresponding quarter-to-quarter growth rates in the private
business sector.40

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Notice that, under both sectoral and value-added methodologies, the quarterly productivity growth rates provide
additional information that is not readily apparent from the annual labor productivity growth rates: the quarter-toquarter labor productivity growth rates show higher peaks and deeper troughs than the annual growth rates. (See
table 2). During the recessionary period, the fourth quarter of 2007 through the second quarter of 2009, average
annual growth was modest (1.2 percent under the sectoral output approach; 1.6 percent under the value-added

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

output approach), while quarterly data fluctuated considerably from period to period. (See tables 3 and 4.) Within
time period, quarterly labor productivity growth rates ranged from −2.4 to 10.1 percent under the sectoral output
approach and ranged from −2.8 to 7.3 percent under the value-added approach.
Table 2. Labor productivity growth for private business sector, annualized percent change from previous
quarter, 2005–14
Year and quarter

Value-added output

2005, Q2
2005, Q3
2005, Q4
2006, Q1
2006, Q2
2006, Q3
2006, Q4
2007, Q1
2007, Q2
2007, Q3
2007, Q4
2008, Q1
2008, Q2
2008, Q3
2008, Q4
2009, Q1
2009, Q2
2009, Q3
2009, Q4
2010, Q1
2010, Q2
2010, Q3
2010, Q4
2011, Q1
2011, Q2
2011, Q3
2011, Q4
2012, Q1
2012, Q2
2012, Q3
2012, Q4
2013, Q1
2013, Q2
2013, Q3
2013, Q4
2014, Q1
2014, Q2
2014, Q3
2014, Q4

Sectoral output
−0.01
2.48
−.09
2.90
.22
−2.21
1.14
.11
2.59
4.11
1.48
−2.79
3.80
.35
−1.56
4.06
7.27
6.79
3.78
.77
.68
2.20
1.11
−3.3
.47
−.68
2.54
−1.29
1.54
.84
−2.12
1.24
.59
3.05
2.11
−4.95
2.58
3.40
−2.91

Source: Authors' calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

10

0.00
1.98
3.15
−.73
−1.18
−2.76
.32
2.08
2.80
−.53
1.21
−2.35
−.45
.39
.40
.38
1.06
9.40
2.75
.88
−.67
.67
−2.28
−.77
−2.75
−2.86
1.44
3.03
.35
.81
.79
−.21
3.54
1.32
2.91
−2.33
2.92
2.40
−1.91

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 3. Labor productivity growth, sectoral output, annual average percent change, 2005−14
Industry
Private business
Agriculture services
Mining
Utilities
Construction
Manufacturing
Durable manufacturing
Nondurable manufacturing
Wholesale trade
Retail trade
Transportation
Information
Finance, insurance, and real estate
Finance
Real estate
Professional, management, and administrative services
Professional and business services
Management services
Administrative and waste management services
Education and health services
Education services
Health services
Leisure and hospitality
Leisure
Accommodations
Other services
Goods
Services

2005−14 2005, Q1− 2007, Q4 2007, Q4− 2009, Q2 2009, Q2− 2014, Q4
0.01
−.58
−.5
−1.46
−1.49
1.65
1.90
1.05
1.20
1.02
.43
4.80
1.65
1.18
1.55
.62
.15
1.31
1.33
1.25
1.49
1.20
.42
2.17
.01
.07
.70
.97

0.53
−1.75
−6.36
−.2
−5.27
2.02
2.33
.40
.94
1.15
2.03
4.91
2.20
3.05
1.51
−.25
−.99
−.92
1.66
.86
2.58
.59
1.02
5.19
.07
−.06
−.82
1.10

1.20
2.33
7.30
−8.79
2.60
−1.8
−5.87
.73
−9.68
−2.52
−3.36
2.69
.97
−.35
1.81
1.73
.47
−4.18
3.76
1.02
.92
1.00
−1.08
.90
−1.74
−2.2
2.24
−.01

0.88
−.86
.04
.36
−.82
2.62
4.56
1.47
5.25
2.11
.87
5.21
1.51
.61
1.41
.69
.66
4.38
.21
1.48
1.08
1.54
.58
.82
.57
.90
.96
1.18

Source: Authors' calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

Table 4. Labor productivity growth, value-added output, annual average percent change, 2005−14
Industry group
Private business
Agriculture services
Mining
Utilities
Construction
Manufacturing
Durable manufacturing
Nondurable manufacturing
Wholesale trade
Retail trade
Transportation
Information
Finance, insurance, and real estate

2005−14 2005, Q1− 2007, Q4 2007, Q4− 2009, Q2 2009, Q2− 2014, Q4
1.06
.44
.60
1.19
−1.58
2.26
3.27
1.08
.62
.55
.27
4.71
2.14

See footnotes at end of table.

11

1.06
−4.32
−4.95
1.60
−5.44
4.43
5.73
3.01
2.07
.22
2.93
7.06
1.80

1.59
11.41
16.75
−5.78
1.69
1.82
−.16
3.23
−5.22
−.06
−1.17
1.54
3.99

0.83
−.35
−1.75
3.21
−.57
1.12
3.01
−.74
1.85
.91
−.69
4.28
1.62

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 4. Labor productivity growth, value-added output, annual average percent change, 2005−14
Industry group
Finance
Real estate
Professional, management, and administrative services
Professional and business services
Management services
Administrative and waste management services
Education and health services
Education services
Health services
Leisure and hospitality
Leisure
Accommodations
Other services
Goods
Services

2005−14 2005, Q1− 2007, Q4 2007, Q4− 2009, Q2 2009, Q2− 2014, Q4
2.08
2.19
.62
.28
−.18
1.27
.50
−.3
.58
−.2
1.78
−.59
−.67
1.38
.94

−.89
3.58
−.34
−.55
−4.86
1.70
.18
.57
.27
−.13
2.84
−.6
−1.07
.92
1.09

7.94
3.20
2.20
1.25
−4.96
3.61
2.05
1.94
2.04
−2.59
.45
−3.35
−2.5
3.39
.93

1.77
1.00
.58
.39
4.14
.14
.17
−1.41
.26
.58
1.56
.39
.19
.85
.82

Source: Authors' calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

Industry labor productivity growth
The heterogeneity among individual industries is lost when the data are presented at the aggregate level, as
opposed to when data are shown in industry-level detail. Figure 4 presents the 2013–14 annual growth in labor
productivity across all industries, under both the sectoral output and value-added output approach. For some
industries, the story is the same for both output concepts, while in other industries the story is quite different. For
example, in the retail trade sector, there is negative labor productivity growth from the sectoral output approach,
but positive productivity growth with the value-added approach. Recall that the difference between sectoral and
value-added output is that sectoral output excludes only those materials purchased from within the retail trade
sector while value-added output excludes all intermediate purchases. Therefore, if sectoral labor productivity is
declining, and value-added labor productivity is growing in retail trade, then the materials that are purchased from
outside retail trade are declining.

12

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Under the sectoral output approach, labor productivity growth between 2013 and 2014 varied among individual
industries, ranging from a decline of 8.0 percent (agriculture, forestry, fishing and hunting) to an increase of 7.4
percent (management of companies and enterprises). Labor productivity growth of 1.1 percent reflects a growth of
1.1 percent for service-providing industries and a growth of 0.8 percent for goods-producing industries. Labor
productivity for service-providing industries was higher than that of goods-producing industries, a result of strong
growth in the management and information industries. Concurrently, the slower growth of goods-producing
industries was triggered by declines in agriculture. Under the value-added approach, labor productivity growth
between 2013 and 2014 also varied among individual industries, ranging from a decline of 13.0 percent
(agriculture, forestry, fishing, and hunting) to an increase of 8.5 percent (management of companies and
enterprises). Labor productivity growth of 0.3 percent for total private business reflects a growth of 0.5 percent for
service-providing industries and a decline of 0.5 percent for goods-producing industries.
The heterogeneity in labor productivity growth among industries is more pronounced in quarterly data than in
annual data. In any given quarter, labor productivity growth will vary dramatically across industries. Over the most
recent quarters, the smallest across-industry variation was 17.0 percent in the second quarter of 2013, with
accommodations and food services declining 4.0 percent and administrative services growing 13.2 percent. In
contrast, the largest variation across industries was 47.0 percent in the first quarter of 2013, with administrative
and waste management services declining 15.6 percent and agriculture growing 31.7 percent. We do not see a
similar range of variation in the annual data. Table 5 presents quarterly labor productivity growth rates for the last
eight quarters of the reference period under the sectoral output approach; value-added data are presented in table

13

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

6. It is clear that the quarterly data present a more dynamic picture than the annual data. In the nondurable
manufacturing industry (see figure 5), annual labor productivity grew 1.7 percent from 2013 to 2014, representing
five quarters of moderate growth (from 1.7 percent to 4.3 percent) interspersed with three quarters of decline (from
−0.1 percent to −1.7 percent). During this same period, the wholesale trade industry experienced six quarters of
increases with only two periods of decline, an observation that is not evident from the 3.0 percent annual growth
rate, visible in figure 6. Such volatility occurs within each of the 20 industries, but is not readily apparent in annual
data.
<id="columnhead8">
Table 5. Labor productivity growth, sectoral output, annualized percent change from previous period, first
quarter 2013 to fourth quarter 2014
2013

2014

2012– 2013– Minimum

Industry group
Q1
Agriculture, forestry,
fishing, and hunting
Mining
Utilities
Construction
Durable goods
Nondurable goods
Wholesale trade
Retail trade
Transportation and
warehousing
Information
Finance and
insurance
Real estate and
rental and leasing
Professional,
scientific, and
technical services
Management of
companies and
enterprises
Administrative and
waste management
services
Educational services
Health care and
social assistance
Arts, entertainment,
and recreation
Accommodation and
food services
Other services,
except government
Private Business
Sector

31.72
−3.6
−4.33
−4.48
2.44
3.39
3.91
1.04

Q2

Q3

Q4

Q1

Q2

Q3

Q4

11.32 −13.83 −4.65 −7.78 −1.66 −26.33 −1.96
3.84
8.02 −14.17 −2.87 26.07
1.10 −17.34
4.02 −3.55 22.86 −9.78 −8.15
2.57
6.87
3.87
6.66
2.67 −3.91
.42 −2.63 −3.08
5.56
1.90
2.08 2.90 4.68
3.62
.11
1.68 −1.67
−.12
2.28 4.05
4.33
−.94
.75
4.76
9.05 −5.02 5.57
7.30
1.37
.02
2.19
2.26 −8.35 7.04
.66 −7.81

−5.8
.66

4.35
1.80

.25
13.24

12.57

−.04

.41

−4.33

−.32

−1.25

−8.11

5.82

2.54

−1.40

−1.63

1.72

7.63

3.54

−15.63 13.20
3.09 3.56

−5.47
3.26

.08 −2.57
1.97 −1.16

13

14

Maximum

growth rate growth rate

9.07
4.09
−.16
−.48
3.08
1.05
1.34
3.24

−7.97
1.42
.47
.10
3.04
1.70
3.02
−.61

−26.33
−17.34
−9.78
−4.48
.11
−1.67
−5.02
−8.35

31.72
26.07
22.86
6.66
5.56
4.33
9.05
1.04

.59
.83
8.36 −2.56

2.93
5.92

2.26
.68

4.82
4.92

−1.66
4.59

1.75
4.19

−5.8
−2.56

4.82
13.24

8.83

1.83

1.92

−1.34

−1.21

3.67

2.23

−1.34

12.57

.33 −3.73

.55

2.89

−2.34

−1.25

−.71

−4.33

2.89

5.18

3.37

−.08

−2.7

−.20

−8.11

5.82

7.02 13.39

8.69

−2.41

.87

7.43

−1.63

13.39

1.20
2.30

−4.11
−.24

3.72
1.76

−1.46
2.52

−.61
1.22

−15.63
−1.16

13.20
3.56

−7.9

−1.01

4.95

3.55

4.48 −5.42

6.55

6.08

3.56

1.81

2.34

−5.42

6.55

−14.49

8.57

−.50

3.31 −4.47 −8.16

5.49

2.40

−1.20

−.78

−14.49

8.57

.53 −4.02

−2.52

−6.35

−.85

.58

−.21

3.54

1.32

3.88 −3.76

5.58

.64

.70

−.41

.35

−4.02

5.58

2.77

.42

2.48

7.66

−3.00

−.57

1.89

−6.35

7.66

2.91 −2.33

2.92

2.40

−1.91

1.23

1.05

−2.33

3.54

See footnotes at end of table.

14

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 5. Labor productivity growth, sectoral output, annualized percent change from previous period, first
quarter 2013 to fourth quarter 2014
2013

2014

2012– 2013– Minimum

Industry group
Private goodsproducing
industries
Private serviceproviding
industries

Q1

Q2

Q3

2.33

6.65

1.29

−1.14

2.46

1.28

Q4

Q1

13

14

growth rate growth rate

Q2

Q3

.15

1.68

1.69

−6.78

2.91

.80

−6.78

6.65

3.75 −3.10

3.05

2.37

−.01

.67

1.09

−3.10

3.75

.56

Q4

Maximum

Source: Authors' calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

Table 6. Labor productivity growth, value-added output approach, annualized percent change from
previous period, first quarter 2013 to fourth quarter 2014
2013

2014

Minimum

Industry group

2012−132013−14
Q1

Q2

Q3

Agriculture, forestry,
fishing, and hunting
82.35 15.87 −2.17
Mining
−9.54 2.43 −0.19
Utilities
5.68 −2.38
.44
Construction
−3.73 −3.77 4.41
Durable goods
−1.14 3.03 3.53
Nondurable goods
12.20
.11 5.66
Wholesale trade
1.16
3.28 2.10
Retail trade
12.76 −5.26 1.66
Transportation and
warehousing
1.54 −3.57 2.82
Information
17.45
2.29 12.51
Finance and
insurance
2.49
4.68 5.09
Real estate and
rental and leasing
.21 −3.1
1.26
Professional,
scientific, and
technical services
−14.17
2.06 7.17
Management of
companies and
enterprises
−20.49
9.60 12.13
Administrative and
waste management
services
−7.21 2.56 −0.76
Educational services
−4.42 −0.78 1.24
Health care and
social assistance
2.93
.54 1.71
Arts, entertainment,
and recreation
−14.02 11.93 −2.7
Accommodation and
food services
−1.13 −5.9 −1.26

Q4

−6.54
.38
5.00
−2.13
−1.19
9.56
1.06
3.82
3.66
12.60

Q1

Q3

−32.17 6.41 −19.17
−24.34 12.09
19.64
−22.87 16.49
16.67
−10.19 −5.01 −1.36
−3.42 2.49
3.91
3.36 5.41 −5.23
−6.48 3.77
7.82
1.65 3.72
.11
−3.4 −2.83
−9.73 7.10

2.82 −10.22
1.94

Q2

5.22

Q4

rate

−3.34
8.64
−7.88
−5.77
−3
3.46
1.23
−4.43

16.68 −12.99
3.43 −0.86
−0.39 −1.01
−1.69 −4.18
.61
.25
.85
3.78
.56
.73
2.69
1.42

−9.43
1.02

−2.33
4.82

19.98 −10.53

2.46
2.76

growth

Maximum
growth rate

−32.17
−24.34
−22.87
−10.19
−3.42
−5.23
−6.48
−5.26

82.35
19.64
16.67
4.41
3.91
12.20
7.82
12.76

−0.93
3.22

−9.43
−9.73

3.66
17.45

4.60

1.32

−10.53

19.98

−6.72 −0.57

3.51

−3.13

−1.8

−1.21

−6.72

3.51

−4.79

5.38

2.14

1.54

−2.35

.63

−14.17

7.17

8.04

5.28 14.47

9.46

−4.31

1.24

8.49

−20.49

14.47

.65
−0.15

−6.02 2.41
−2.63 −2.57

−1.73
3.10

−2.3
−1.36

−1.17
−1.27

−1.27
−0.78

−7.21
−4.42

2.56
3.10

−3.07

−0.61

3.14

3.74

−0.12

−0.19

1.16

1.60

−0.61

3.74

2.08

−7.47

1.04

8.94

.43

−0.47

.31

−14.02

11.93

2.91

−2.61

1.47

.74

−2.83

−1.56

−0.46

−5.9

2.91

See footnotes at end of table.

15

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 6. Labor productivity growth, value-added output approach, annualized percent change from
previous period, first quarter 2013 to fourth quarter 2014
2013

2014

Minimum

Industry group

2012−132013−14
Q1

Other services,
except government
Private Business
Sector
Private goodsproducing
industries
Private serviceproviding
industries

Q2

−0.93 −0.98

growth

Maximum
growth rate

Q3

Q4

Q1

Q2

Q3

Q4

3.10

4.97

−1.73

−2.5

10.10

−4.36

.33

1.35

−4.36

10.10

rate

1.24

.59

3.05

2.11

−4.95

2.58

3.40

−2.91

.73

.25

−4.95

3.40

5.34

1.49

2.58

1.93

−5.36

1.78

−0.87

−1.48

1.42

−0.54

−5.36

5.34

.01

.32

3.19

2.17

−4.82

2.82

4.69

−3.32

.52

.48

−4.82

4.69

Source: Authors' calculations based on data from U.S. Bureau of Labor Statistics and U.S. Bureau of Economic Analysis.

16

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Data users are often interested in short-term movements. However, such movements are subject to greater
volatility than longer term movements. Macroeconomic series, such as GDP, frequently fluctuate around a trend
that varies over time. Because of this tendency, the permanent trend should be separated from the transitory
component. Most frequently, volatility is estimated on the basis of the standard deviation of the growth rate.41
Referring back to tables 5 and 6, we see that quarterly growth rates are much more volatile than annual average
growth rates. From 2013 to 2014, growth in labor productivity in the retail trade industry declined an average of 0.6
percent across eight quarters under the sectoral output approach. The quarterly data reveal that growth ranged
from a minimum of −8.3 percent in the first quarter of 2014 to a maximum of 10.0 percent in the first quarter of
2013. The value-added approach yields similar results for this industry, with quarterly growth ranging from a
minimum of −5.3 percent in the second quarter of 2013 to a maximum of 12.8 percent in the first quarter of 2013.
By contrast, the average annual growth rate from 2013 to 2014 was a constant 1.4 percent.

Industry contributions to labor productivity growth
To examine how individual industries affect growth in the private business sector, industry contributions were
calculated as the individual industry’s growth in labor productivity weighted by its average share of output in the
two periods of interest.42 For ease of exposition, figures 7 and 8 show how the broad groups of private goodsproducing and service-providing industries contribute to the growth in aggregate labor productivity measures.
Notice that, in most quarters, service-providing industries are contributing to both the majority of gains and the
majority of declines in aggregate labor productivity.

17

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

A closer look at individual service-providing industries reveals that, under the sectoral output approach, the
finance, insurance, and real estate industry contributed to strong gains in the second quarter of 2009. (See figure
9.) In the fourth quarter of 2014, the losses in labor productivity are also primarily from this industry while the gains
can be attributed primarily to the information and utilities industries.

18

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Among the goods-producing industries, there were early gains in construction and nondurable goods in the second
quarter of 2009 followed by gains in all goods-producing industries in the third quarter of that year. (See figure 10.)
However, in the fourth quarter of 2014, productivity declines were due largely to negative productivity growth in
mining. These quarterly data on labor productivity by industry provide new insights into economic activity and
highlight the heterogeneity among industries, complementing the existing aggregate labor productivity measures.

19

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Under the value-added methodology, the finance, insurance, and real estate industry is the greatest contributor to
aggregate gains in service-providing industries coming out of the recessionary period. (See figure 11.) However, in
the fourth quarter of 2014 (the most recent quarter included in the figures), this industry contributed heavily to the
decline in aggregate productivity. Among goods-producing industries, durable goods manufacturing exhibited
strong growth, contributing significantly to aggregate labor productivity coming out of the recession. In this most
recent quarter, productivity growth in durable manufacturing and construction were offset by productivity declines
in mining and nondurable manufacturing, resulting in little growth in the goods-producing sector. (See figure 12.)

20

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Conclusion
Sustained growth in labor productivity enables an economy to produce additional goods and services without an
increase in labor resources, resulting in higher standards of living. Given the newly available quarterly GDP-byindustry data, this study showed that it is feasible to create reasonable quarterly hours-worked measures for 20
industry groups of interest; however, further industry detail on a quarterly basis may be beyond the limits of the
available CPS hours data. Although quarterly labor productivity data at the industry level offer users insights into
which industries are contributing the most to aggregate productivity growth, the high volatility in the data limit their
usefulness.
This quarterly labor productivity measures developed in this study at the industry level are presented as growth in
output relative to growth in hours worked. Labor productivity series were constructed out of both GDP-by-industry
data and a broader measure of sectoral output.
It is apparent from the data that the choice of output approach affects labor productivity growth rates as well as
individual industry contributions to aggregate economic growth. It is therefore important for data users to be aware
of whether intermediate purchases from outside the industry are included or excluded from measures of output. In
addition, adjustments were made to both output and labor data to eliminate known sources of productivity bias
resulting from the use of input data in the construction of measured output. That is, the output and hours worked
by NPISH were removed from the industry data. Although their removal improved the data, in some industries
input and output data remain correlated, and users should be cautious when interpreting these data, particularly in
the information, real estate, management services, administrative services, finance and insurance, professional
and business services, leisure, accommodations, and other services industries.

21

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Under both the sectoral and value-added methodologies, quarterly productivity growth rates provide additional
information that cannot be gleaned from existing aggregate quarterly or annual industry measures. Heterogeneity
among individual industries is lost when the data are presented at the aggregate level, and quarter-to-quarter labor
productivity growth rates show higher peaks and deeper troughs than the annual growth rates exhibit for specific
industries. In addition, the heterogeneity in labor productivity growth among industries is more pronounced in the
quarterly data. However, because quarterly labor productivity data at the industry level are highly volatile, data
users should use the information to supplement long-run analysis and should be cautious when drawing
conclusions about the state of the economy on the basis of a single quarterly data point.
ACKNOWLEDGMENTS: The authors thank Mark Dumas, John Glaser, Christopher Kask, Paul Kern, Sabrina
Pabilonia, Steven Rosenthal, John Ruser, Chris Sparks, Jay Stewart, Erich Strassner, Victor Torres, and David
Wasshausen for helpful comments. All views expressed in this article are those of the authors and do not
necessarily reflect the views or policies of the U.S. Bureau of Labor Statistics.
SUGGESTED CITATION

Lucy P. Eldridge and Jennifer Price, "Measuring quarterly labor productivity by industry," Monthly Labor Review,
U.S. Bureau of Labor Statistics, June 2016, https://doi.org/10.21916/mlr.2016.28.
NOTES
1 Major sectors include business, nonfarm business, manufacturing, and nonfinancial corporations. Annual industry measures are
calculated for two-, three-, four-, five-, and six-digit industries as defined by the North American Industry Classification System
(NAICS). For more information, see Labor Productivity and Costs, (U.S. Bureau of Labor Statistics), https://www.bls.gov/lpc/
home.htm.
2 “New quarterly statistics detail industries’ economic performance: statistics span first quarter of 2005 through fourth quarter of 2013
and Annual Results for 2013,” news release (U.S. Bureau of Economic Analysis, April 25, 2014), https://apps.bea.gov/newsreleases/
industry/gdpindustry/2014/gdpind413.htm.
3 The government sector is not included in the study.
4 The press release includes quarterly and annual indexes, and percentage changes, for output per hour and related measures, such
as unit labor costs, real and current dollar compensation per hour, and unit nonlabor payments. (See Labor Productivity and Costs.)
5 Although quarterly labor productivity measures are produced for the total economy, the methods for estimating output for some
components of the economy are problematic for productivity measurement. Thus, measures of productivity for the total economy are
considered less reliable than business sector measures and are not included in the press release.
6 Although the farm sector in the United States is small, it is highly volatile. For more information on BLS methods, see “Technical
information about the BLS major sector productivity and cost measures” (U.S. Bureau of Labor Statistics, March 11, 2008), https://
www.bls.gov/lpc/lpcmethods.pdf.
7 Frank T. Denton, “Adjustment of monthly or quarterly series to annual totals: an approach based on quadratic minimization,” Journal
of the American Statistical Association, vol. 66, no. 333 (March 1971), pp. 99–102, http://www.oecd.org/std/21779760.pdf. The Denton
proportional first-difference method preserves the pattern of growth in quarterly indicator series by minimizing the proportional periodto-period change while meeting the average annual level constraints.
8 Because of a lag in the availability of the annual benchmark data, more recent quarterly and annual manufacturing output measures
are also extrapolated on the basis of changes to the indexes of Industrial Production.

22

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

9 Industry measures produced include output per hour, output per employee, output, implicit price deflators, employment, hours of
employees, labor compensation, and unit labor costs. Separate news releases are issues for selected services, manufacturing, and
trade.
10 BLS Handbook of Methods (U.S. Bureau of Labor Statistics), chapter 11, p. 2, https://www.bls.gov/opub/hom/pdf/homch11.pdf.
11 For more information on the evolution and early phases of the development of quarterly GDP-by-industry statistics, see Carol A.
Robbins, Thomas F. Howells, and Wendy Li, "Experimental quarterly U.S. gross domestic product by industry statistics," Survey of
Current Business (U.S. Bureau of Economic Analysis, February 2010), pp. 24–31, https://www.bea.gov/scb/pdf/2010/02%20February/
0210_gdp_indy.pdf.
12 Annual I–O accounts are available for 1997–2012 and include data on 69 industries. Benchmark I–O accounts include more
detailed information for about 400 industries. Benchmark I–O accounts are prepared roughly every 5 years and are based on detailed
data from the Economic Census conducted by the Census Bureau. The 2007 benchmark was released in December 2014. For more
information, see Donald D. Kim, Erich H. Strassner, and David B. Wasshausen, “Industry economic accounts: results of the
comprehensive revision, revised statistics for 1997–2012,” Survey of Current Business (U.S. Bureau of Economic Analysis, February
2014), pp.1–18, https://www.bea.gov/scb/pdf/2014/02%20February/0214_industry%20economic%20accounts.pdf.
13 See “Measuring the nation’s economy: an industry perspective, a primer on BEA industry accounts” (U.S. Bureau of Economic
Analysis, May 2011), https://www.bea.gov/resources/methodologies/industry-primer.
14 For a complete description of methods and source data, see Erich H. Strassner and David B. Wasshausen “New quarterly gross
domestic product by industry statistics,” Survey of Current Business (U.S. Bureau of Economic Analysis, May 2014), https://
www.bea.gov/scb/pdf/2014/05%20May/0514_gdp-by-industry.pdf.
15 The domestic and foreign portions of intermediate inputs are deflated separately to account for commodities purchased as inputs
from both domestic and foreign sources.
16 Labor productivity is not measured under a gross output concept because, under that concept, intermediate inputs made within an
industry or sector would be double counted—counted by both the establishment producing the product and the establishment
consuming the product.
17 Edwin R. Dean, Michael J. Harper, and Mark S. Sherwood, “Productivity measurement with changing weight indices of outputs and
inputs,” Industry Productivity: International Comparison and Measurement Issues (Paris: Organisation for Economic Co-operation and
Development, Washington, DC, 1996), chap. 7, pp. 183–215, http://www.oecd.org/sti/ind/1825894.pdf.
18 Multifactor productivity (MFP) data give a more comprehensive picture of productivity change over time, and they provide a
decomposition of labor productivity change into sources of growth. However, because of the complexities associated with constructing
MFP measures, these data are not available on a quarterly basis. BLS publishes MFP on major sectors and selected detailed
industries on an annual basis. For more information, see https://www.bls.gov/mfp/.
19 See William Gullickson, “Measurement of productivity growth in U.S. manufacturing,” Monthly Labor Review, July 1995, pp.13–37,
https://www.bls.gov/mfp/mprgul95.pdf.
20 For a complete discussion of the advantages and disadvantages of the two output concepts, see “Measuring productivity:
measurement of aggregate and industry-level productivity growth,” OECD Manual (Paris: Organisation for Economic Co-operation
and Development, 2001), chapter 3, pp. 23–38 https://www.oecd.org/std/productivity-stats/2352458.pdf.
21 Ibid., p. 28. If technical change within an industry does not affect all factors of production but operates primarily on the primary
inputs, then the value-added approach is preferable.
22 Value-added labor productivity measures are generally less sensitive to outsourcing than are sectoral measures. But for multifactor
productivity, sectoral measures are less sensitive to outsourcing than value-added measures are.

23

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

23 For more information on how imports affect productivity measures, see Lucy P. Eldridge and Michael J. Harper, “Effects of
imported intermediate inputs on productivity,” Monthly Labor Review, June 2010, pp. 3–15, https://www.bls.gov/opub/mlr/2010/06/
art1full.pdf.
24 Sectoral output measures for manufacturing industries are derived from BLS quarterly labor productivity data. Measures for
nonmanufacturing industries were calculated for this study and may differ from annual BLS multifactor productivity because of data
vintages.
25 See “NIPA Handbook: Concepts and methods of the U.S. national income and product accounts” (Bureau of Economic Analysis,
February 2014), chapter 3, https://apps.bea.gov/national/pdf/NIPAhandbookch1-4.pdf.
26 “Annual benchmark report for services: fourth quarter 2003 to fourth quarter 2013” (U.S. Census Bureau, June 2014), https://
www2.census.gov/services/qss/2013/benchmark_text-2013.pdf.
27 Annual SAS reports are available at the Census Bureau's "Annual and quarterly services," https://www.census.gov/services/sas/
historic_data.html.
28 For more information on BEA data sources, see Strassner and Wasshausen, “New quarterly gross domestic product by industry
statistics,” (BEA Briefing, May 2014), pp. 10–11, https://www.bea.gov/scb/pdf/2014/05%20May/0514_gdp-by-industry.pdf.
29 BLS does produce quarterly hours for wage and salary workers on nonfarm payrolls for 14 major industry groups, available in
tables at https://www.bls.gov/lpc/special_requests/tableb10.txt.
30 For more information on the CES, see Current Employment Statistics–CES (National)" (U.S. Bureau of Labor Statistics), https://
www.bls.gov/ces/.
31 Seasonally adjusted three-digit CES data are used for nonmanufacturing industries, and two-digit data are used for durable and
nondurable manufacturing. National Compensation Survey data are used at a slightly higher level of aggregation for
nonmanufacturing. Employee data are then aggregated for the 20 industries of interest.
32 In goods-producing industries, workers are divided into production and nonproduction workers. Nonproduction workers include
professional, specialty, and technical workers; executive, administrative, and managerial workers; sales workers; and administrative
support workers, including clerical workers. In service-providing industries, workers are divided into supervisory and nonsupervisory
workers. Supervisory workers include all executives and administrative and managerial workers. The CES program began collecting
data on earnings and hours for all employees in September 2005. The BLS Productivity Program is currently evaluating whether to
start using this new series.
33 Estimates of quarterly hours-worked-to-hours-paid ratios are derived from annual data at the three-digit industry level via a
smoothing procedure. The BLS major sector productivity program makes use of ratios at a more aggregate level. For more
information on the NCS, see "National Compensation Survey" (U.S. Bureau of Labor Statistics), https://www.bls.gov/ncs/.
34 To facilitate comparisons across various periods, quarterly estimates are expressed as annualized levels and quarterly growth
rates are expressed as annual growth rates via the following formula:

.

35 See Lucy P. Eldridge, Marilyn E. Manser, and Phyllis Flohr Otto, “Alternative measures of supervisory employee hours and
productivity growth,” Monthly Labor Review, vol. 127, no. 4, April 2004, pp. 9–28, https://www.bls.gov/opub/mlr/2004/04/art2full.pdf.
36 The X12-ARIMA program was developed by the U.S. Census Bureau. It is the same adjustment technique that CES employs to
adjust employment and average weekly hours, and the same program used by Census and BEA to adjust output. Indirect seasonal
adjustment (seasonally adjusting the components of the hours calculation rather than the final value) is preferred when component
series are suspected of having distinct seasonal patterns. (See "X-13-ARIMA-SEATS Seasonal Adjustment Program" (U.S. Census
Bureau), https://www.census.gov/srd/www/x13as/.) Given the limited observations for some industry groups, the CPS data are
seasonally adjusted quarterly rather than monthly.

24

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

37 For this study, CPS ratios were constructed for 20 selected industry groups; BLS quarterly major-sector productivity measures use
14 industry categories, while annual labor productivity measures use ratios constructed at the three- and four-digit industry level. For
more information on the CPS, see "Labor force statistics from the Current Population Survey" (U.S. Bureau of Labor Statistics),
https://www.bls.gov/cps/.
38 The published quarterly labor productivity statistics aggregates employee hours for 14 major industry groups, then adds an
aggregate value of hours worked for the self-employed and unpaid family workers. This aggregation is done because the major sector
is the only series of interest.
39 Because the CES collects employment and average weekly hours only for the logging industry, nonfarm agricultural services
employment data from the QCEW are combined with CES logging data to construct nonfarm employee hours. The data for
constructing these estimates were released by the CES program in February 2015.
40 All series presented in this article were constructed by the authors. Differences between these data and the published BLS
productivity statistics are a result of the difference in data vintage, as well as the adjustments made at different levels of industry
aggregation.
41 See Joël Cariolle and Michaël Goujon, “Measuring macroeconomic instability: a critical survey illustrated with exports series,”
Journal of Economic Surveys, vol. 29, no. 1, February 2015, pp. 1–26.
42 The authors explored different approaches for estimating the contributions to labor productivity growth, including those discussed
in Evsey D. Domar, “On the measurement of technological change,” Economic Journal, December 1961, pp. 709–729, http://
www.jstor.org/stable/2228246?seq=1#page_scan_tab_contents; and Marshal Reinsdorf and Robert Yuskavage, “Exact industry
contributions to labor productivity change,” in Price and productivity measurement, vol. 6, chapter 5, (2010), http://
www.indexmeasures.com/Vol6_10,09,26.pdf. Both approaches produce similar results. The authors used the Domar approach but
scaled the contributions to the aggregate level in order to capture interactions between industries.

RELATED CONTENT

Related Articles
New measure of labor productivity for private community hospitals: 1993–2012, Monthly Labor Review, October 2015.
Effects of imported intermediate inputs on productivity, Monthly Labor Review, June 2010.
Nonmanufacturing industry contributions to multifactor productivity, 1987–2006, Monthly Labor Review, June 2010.
Productivity trends in business cycles: a visual essay, Monthly Labor Review, June 2009.

Related Subjects
Productivity
studies

Statistical programs and methods

Occupations

25

Industry and Occupational studies

Industry

June 2016

The effect of the Affordable Care Act’s Medicaid
expansion on health insurance coverage in 2014
Richard Works
The major components of the Affordable Care Act (ACA) went into effect in 2014 with Medicaid expansion being
optional. How was health insurance coverage affected in states that expanded versus those that did not?
According to “Impacts of the Affordable Care Act on Health Insurance Coverage in Medicaid Expansion and NonExpansion States” (National Bureau of Economic Research working paper no. 22182, April 2016) by Charles
Courtemanche, James Marton, Benjamin Ukert, Aaron Yelowitz, Daniela Zapata, gains in insurance coverage were
largest for nonwhites, young adults, unmarried individuals, and those with incomes below the threshold for
Medicaid eligibility.
Data were collected through the American Community Survey administered by the Census Bureau. This survey
samples approximately 1 percent of the U.S. population: over 3-million people per year. The sample for this study
consisted of 18–64 year olds from calendar years 2011 to 2014. The researchers incorporated multiple control
variables for demographic characteristics, family structure, and economic/labor force characteristics. Researchers
also controlled for the seasonally adjusted monthly state unemployment rate published by the Bureau of Labor
Statistics. These controls addressed concerns that Medicaid expansion may be correlated with other factors.
The preferred specification for the regression analysis was a difference-in-difference-in-differences design that
disentangles year-to-year changes from causal effects of non-Medicaid portions of the ACA. The full-sample
regressions estimate the ACA, including the Medicaid expansion, increased insurance coverage by 5.9 percentage
points at the sample mean pretreatment uninsured rate; this compares with 3.0 percentage points without the
expansion. The effect reached as high as 15.4 percentage points (compared with 7.8 without the expansion) in the
area with the largest uninsured rate. Results passed falsification tests and remained similar across checks for
robustness.
Results indicate the ACA increased private insurance coverage by 2.4 percent. The authors indicate this is due to
the increased take-up rates for employer-sponsored plans that resulted from the individual mandate. No
consequential difference was found between coverage gains among low-income and middle-income earners in
non-Medicaid-expansion states. Private insurance was the source of coverage gains in non-Medicaid-expansion
states. Gains in Medicaid-expansion states were exclusively attributable to increased Medicaid coverage.
However, some evidence suggest that reduced private coverage crowded out a portion of gains in Medicaid
coverage from the expansion.
This study extends literature through methodological approaches. The authors state, “Our identification strategy for
the non-Medicaid expansion portion of the ACA can potentially be used in future research to identify the impacts of

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

the ACA on other outcomes such as health care utilization, health, and personal finances.” However, data
availability is indicated to be a limitation. Additional research is recommended as more data become available.

2

June 2016

Long may they live: cities and neighborhoods over
the centuries
Edith S. Baker
Regions come, and regions go. Cities arise, flourish for a time, and then fall into decline. Neighborhoods change.
But not always. In some—even many—cases, cities and towns persist over centuries and even millennia.
Neighborhoods get new residents, but the character of the neighborhood remains. Why is all this so? Why is it that
some places last for a long time and others don’t? In “The puzzling persistence of place” (Federal Reserve Bank of
Philadelphia Business Review, Second Quarter 2015), Jeffrey Lin asks three questions about neighborhoods,
cities, and regions that have endured despite changes in their circumstances that would have caused other
neighborhoods, cities, and regions to fall by the wayside in one way or another: “Why haven’t these urban patterns
changed over decades, centuries, or even millennia? Is such persistence desirable? And what does persistence
imply about the prospects for ‘place-making’ policies aimed at generating development or attracting [development]
to particular locations?” The rest of his article seeks to answer these questions and to shed light not only on the
reasons that certain places remain practically perpetually livable, but also on what we can do to encourage policies
which feed into the natural attributes that make them so livable.
Lin cites three factors that economists have identified to “account for the remarkable long-run persistence of
place”: natural geographic advantages; human geography, or agglomeration economies; and the human
geography of the past, or sunk factors.
Natural geographic advantages are features of the territory, such as natural harbors, navigable rivers, and
defensible hills, that attract households and businesses to the area. Their value may change over time—for
example, natural harbors once were a magnet for trade and development, but now are more likely to attract
residents and tourists because of the beauty of the landscape—but they still retain value.
Human geography is represented by those features of the environment which are valuable because they offer
proximity to households and businesses whereby people can work, shop, be entertained, and give and receive all
sorts of services. In other words, people benefit from the very fact that they agglomerate and form a thriving
economic unit. Places that have done so tend to persist over time, despite both internal and external changes that
might otherwise be disruptive.
Finally, the human geography of the past gives locations “durable capital left over from decades or centuries ago.”
These sunk factors are a firewall against degradation and decay, providing a legacy of infrastructure and
institutions that keeps a city or neighborhood vibrant while other areas fall into decline. Often, they serve purposes
different from those they originally served—as, for example, a dilapidated bridge that once was a busy
thoroughfare connecting heavy traffic between two “Rust Belt” towns. The bridge was never torn down and

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

replaced because it would have been too costly to do so. Now it is of use only to local residents and businesses—
a purpose different from the original one, but a purpose nonetheless.
As evidence of the role of natural geographic advantages in persistence, Lin points to the strong correlation
between the distribution of population among Japanese cities today and the population distribution of those same
areas 8,000 years ago, as discovered by archaeologists. Moreover, despite heavy, random bombing of those cities
during World War II, population returned to its prewar distribution shortly after the war. Regarding the role of
human geography in persistence, Lin recalls the many U.S. cities that owe their establishment and early existence
to some natural geographic advantage they had over other cities—for example, being near a waterfall that
provided waterpower or being a port city when river traffic was the only way commerce could be conducted with
other parts of the nation. Later, when that advantage was lost because of technological advancements
(electrification and new land transportation technology, respectively), those cities still thrived because of the strong
agglomeration economies that remained.
In a different, but related, vein, some among the first Europeans to America settled in certain New York City
marshlands whose poor drainage resulted in flooding and disease. These areas became low-income areas of
cities, with poor amenities and meager public services. Later, despite improvements in drainage and sewerage
which made that natural disadvantage disappear, those areas remained impoverished, likely because of their bynow long established human geography—and they remain so today. Thus, there can be “bad” agglomeration
economies that lead to the persistence of cities with undesirable characteristics.
Finally, confirming the role of sunk factors (the human geography of the past) in persistence, Lin calls the reader’s
attention to Sweden, where towns that grew up along that country’s budding railroad system grew faster than other
towns and remain larger today; to sub-Saharan Africa, where cities and agricultural development continue to follow
long-abandoned rail lines; and to Los Angeles, California, where “population density today is strongly correlated
with the location[s] of streetcar stops in the 1910s, and this correlation has been increasing over time.” All of these
situations provide evidence that historical investments in infrastructure (i.e., sunk factors) may keep a location
viable beyond what would otherwise be expected.
So, what can we learn from the factors underlying persistence about policies aimed at creating or attracting
economic activities to particular locations? What kinds of policies can we adopt that offer a reasonable probability
of success? One thing we can learn, says Lin, is that “policies that work against these three factors are unlikely to
succeed.” For example, consider the prospect of building a new airline hub from scratch. Given that existing hubs
are characterized by large sunk costs and economies of scale, building a new hub would require overcoming those
advantages. Similarly, policies that seek to improve the condition of Los Angeles neighborhoods with poor natural
amenities would be almost doomed to failure from the start because it would be difficult to overcome that large
“sunk” disadvantage. To bring the point home via a concrete example, it would take “an implausibly large
investment…to improve South Los Angeles to the level of Beverly Hills.” But policies that recognize the factors
underlying persistence and that take full advantage of them “may be most effective in creating long-lasting change
in neighborhoods and cities.” For example, if it is known that certain neighborhoods have businesses that engage
in economic activities which strongly benefit other businesses and households, then policies that encourage those
activities might have permanent beneficial effects.

2

June 2016

Labor productivity growth in elementary and
secondary school services: 1989–2012
New measures of labor productivity in elementary and
secondary schools show that labor productivity declined
from the mid-1990s through the first decade after 2000.
However, it increased from 2010 forward because of labor
input declines combined with modestly increasing
enrollments and test scores.
Education is a primary driver of economic growth and
stability for nations and for individuals.1 Investments in
education affect the ability of a country to compete in
international markets and to ensure increases in living
standards for its residents. On an individual level, returns to
investments in education are well documented. For
example, high school dropouts earn less over their lifetimes
than high school graduates. In addition, the penalty
associated with lower educational attainment has become
more pronounced over time: after the earnings are adjusted

Susan G. Powers
powers.susan@bls.gov

expected lifetime earnings of high school dropouts in

Susan G. Powers is a research economist in the
Division of Productivity Research and Program
Development, Office of Productivy and
Technology, U.S. Bureau of Labor Statistics.

1970.2

Steven Flint

for inflation, economists project that expected lifetime
earnings of high school dropouts today are less than the

In 2014, nearly 9 percent of all U.S. workers, almost 13
million individuals, were employed in the educational
services sector.3 In fact, the education sector now employs
more workers than the entire manufacturing sector. Within

Steven Flint was formerly an economist in the
Division of Industry Productivity Studies, Office of
Productivity and Technology, U.S. Bureau of
Labor Statistics.

the educational services sector, the elementary and
secondary schools industry employed just over 8 million
individuals, or 65 percent of employees in the broader
education sector.4 For the 2013–14 school year, expenditures for educational services in the United States are
estimated at $1,194 billion, or 7.1 percent of the United States Gross Domestic Product (GDP). Of this total, $682
billion, or 4.1 percent of GDP, were expenditures for educational services at public and private elementary and

1

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

secondary schools and $512 billion, or 3.1 percent of GDP, were expenditures for education at postsecondary
degree-granting institutions.5
Labor productivity is a statistical measure that relates an industry’s output of services to the quantity of labor
required to produce those services. Productivity data are essential for one to understand the education industry
because they provide information on the efficiency of the production of services.
The U.S. Bureau of Labor Statistics (BLS) has developed measures of labor productivity and related series,
including unit labor costs, for North American Industry Classification System (NAICS) 6111, elementary and
secondary schools, the second largest component industry in the education sector by employment and receipts,
after NAICS 6113, colleges, universities, and professional schools.6 This is the first detailed industry in the
education sector for which BLS has developed labor productivity series. The new measures reflect BLS
commitment to expand its coverage of service industries, including those industries for which developing reliable
series presents a significant challenge. This article introduces the new measures and examines productivity trends
in this industry from 1989 to 2012.
The elementary and secondary schools industry includes both public and private schools providing educational
and related services that constitute a basic preparatory education, typically from kindergarten through the 12th
grade.7 BLS measures for this industry include data for all schools, as well as separate data on public and private
schools. During the 2012–13 school year, 98,454 public schools and 30,651 private elementary and secondary
schools, serving over 55 million students, were operating in the United States.8 Over 90 percent of these students
attended public schools, while less than 10 percent attended private schools.9 Although public school students
overwhelmingly attended traditional public schools, the number of public school students attending charter schools
increased from 0.7 percent in the 1999–2000 school year to 4.6 percent in 2012–13.10 Charter schools are
independent public schools operating under a contract or “charter,” with a state agency or local school board. In
exchange for greater flexibility and independence, the charter school agrees to be accountable for goals outlined in
the charter, such as improving student performance. Failure to achieve these goals may cause the school’s charter
to be revoked.11 Finally, 3.4 percent of school-age children were homeschooled during the 2012–13 school year.12

U.S. education system
Educational services in the United States have their roots in early colonial days. The first law establishing universal
public education was the Massachusetts General School Law of 1647, which ordered, “. . . that where any town
shall increase to the number of one hundred families or householders, they shall set up a grammar school . . . .”13
From these rudimentary beginnings, the elementary and secondary education system in the United States has
evolved and striven, guided by landmark legislation such as the 1965 Elementary and Secondary Education Act
and the 1975 Education for All Handicapped Children Act, to educate children with varied economic circumstances,
language skills, and mental and physical capabilities appropriately.14
The U.S. public school system historically has been the responsibility of local school districts and states. State
education departments set general education requirements, whereas local school boards manage school district
level decisions.15 To receive federal funds for education, states must comply with various guidelines and standards
that the federal government has established, through legislation.16 Public school systems, which are required to
provide free educational services to all, provide the majority of educational services to elementary and secondary
school-age individuals with disabilities. In addition, public school systems serve the majority of economically
2

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

disadvantaged students.17 States generally assign responsibility for providing adult education services, such as
General Educational Development (GED) programs, literacy programs, and English as a Second Language
(ESOL) programs, to local public school systems.
Educational standards for grades K–12 in public schools are established at the state level and are overseen by the
individual state’s education department. These standards typically include requirements for proficiency in basic
subjects such as reading, writing, mathematics, and science; minimal achievement gaps by race, gender, ethnicity,
and socioeconomic status; college readiness preparation, including availability and participation in college credit
programs such as Advanced Placement (AP) and International Baccalaureate (IB) programs; and minimum
academic credits for graduation. Requirements for a high school diploma vary by state and include achieving a
minimum number of credits in English, mathematics, science, social studies, arts or vocational education, physical
education and, in some states and school districts, a foreign language.18
Private schools must observe federal, state, and local laws, such as maintenance of state-required attendance,
curriculum, and safety records and reports; compliance with local building, fire, and sanitation codes; and annual
reports to the Internal Revenue Service.19 Private schools may receive services under the federal Elementary and
Secondary Education Act (ESEA), including participating in programs such as the National School Lunch Program,
receiving funds from Individuals with Disabilities Education Act grants, and receiving funds from the Title I,
Improving the Academic Achievement of the Disadvantaged, program for supplemental education.20 Note,
however, that private schools serve a small percentage of individuals with disabilities. Often, regional or national
accrediting agencies, such as the Association of Independent Schools of New England, or the National Christian
School Association, accredit private schools.
Although private schools are generally required to provide a curriculum of study meeting the state minimum
requirements, they often offer broader curriculums, lower pupil–teacher ratios, additional extracurricular activities,
more qualified teachers, and even longer school years.21 Graduation requirements of private high schools typically
are more demanding than the requirements of public high schools.22 From 1999 to 2000, private high schools
required more coursework in social studies, mathematics, science, foreign language, and computer science than
did public high schools.23 For example, private schools required, on average, 3.1 years of mathematics and 1.5
years of foreign language, whereas public schools required 2.7 years and only 0.5 years, respectively.24 Most
private elementary and secondary schools are nonprofit; a very few are for profit, particularly schools run by
Education Management Organizations, private organizations that manage charter schools.25
Today, public and private schools in the United States provide students with not only instructional services but also
supplementary student support services, such as guidance counseling, healthcare services (including school
nurses; school psychologists; vision, dental, audiology and speech screenings; and speech therapy services), food
services, and transportation services (primarily offered by public schools). Additional support services for learning,
emotionally, or physically disabled students are also provided, when appropriate.
Major challenges that the elementary and secondary schools industry has experienced include large increases in
the number of non-English-speaking students, an increase in the number of students from impoverished
circumstances, the integration of mentally and physically challenged students into general classrooms, and a
cultural shift toward more working mothers, particularly mothers with preschool children. Demand for early

3

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

education programs has increased markedly because a growing number of women continue to work while raising
young children.26 Other additional challenges include the following:
·

English language learners increased in all but 11 states and grew from 5.0 percent of students in 1993

to 9.2 percent in 2012.27
·

Students living in poverty increased from 17 percent in 1990 to 21 percent in 2012.28

·

Students participating in the free or reduced price lunch program increased from 32 percent in 1990 to

51 percent in 2012.29
·

Students receiving special education services under the Individuals with Disabilities Education Act

increased from 11 percent in 1990 to 13 percent in 2012.30 The percentage of disabled students spending
80 percent or more of their time in a regular classroom environment increased from 33 percent in 1990 to
61 percent in 2012.31
·

Enrollment of 3- and 4-year-olds in academic preschool programs increased from 32 percent in 1975 to

44 percent in 1990 and 55 percent in 2013.32
·

Women in the labor force with children under age 6 increased from 39 percent in 1975 to 58 percent in

1990 and 65 percent in 2013.33

Education output
Constructing a labor productivity measure for elementary and secondary schools (NAICS 6111) first requires
developing industry output and labor input measures. The output produced by establishments in NAICS 6111
includes educational and related services that develop the knowledge, skills, and competencies of students,
culminating in completion of a basic preparatory education. However, educational services are difficult to measure
directly.34 In the United States, educational services are provided by a mixture of public and private organizations
that include both nonprofit and for-profit entities. Measuring education output involves all the difficulties associated
with measuring service outputs, further complicated by production in a nonmarket setting, such as production of
educational services by public schools.35 Services produced in a nonmarket setting are sometimes measured in
terms of employee hours; that is, the service is defined as the employees’ time. However, this definition of output is
not appropriate for use in measuring productivity. If output growth is based on the related change in labor, then
measured labor productivity is by definition constant and no information regarding industry efficiency can be
determined.
Various countries use different methods of measuring education output. These output measures range from the
very simple (such as a count of students enrolled) to more complex measures, which include a quality adjustment
reflecting some aspect of educational outcome.36 For primary and secondary education, several countries,
including Australia, Finland, Germany, Greece, the Netherlands, and New Zealand, use a volume measure such as
the number of pupils or number of teaching hours with no further adjustment.37 Other countries, such as Hungary,
Italy, Poland, and Spain, use the number of pupils as a volume measure and adjust for differences in class size.38
The United Kingdom constructs an education output measure that uses the number of students as a volume

4

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

measure and the average point score per student in the 11th-year General Certificate of Secondary Education test
as a quality adjuster.39
Although the number of students enrolled is a useful starting point for measuring elementary and secondary
education output, it does not reflect changes in the educational attainment level of students. Educational
attainment has been shown to vary over time, depending on teacher quality, class size, curriculum quality, and
other factors. However, the extent to which the various characteristics of the educational environment and activities
influence education output is not always clear. Fraumeni, Reinsdorf, Robinson, and Williams evaluated various
combinations of characteristics, such as improved student–teacher ratios, changes in teaching staff composition,
and high school dropout rates, and found that, although the direction in which a particular characteristic of the
educational environment affects education output may be known, quantifying the impact on education output is
difficult.40
A wealth of information is available on school performance. The federal government, through Title I of the federal
ESEA and reauthorized by the federal Every Student Succeeds Act of 2015, requires states to create annual
assessments of schools and school districts.41 In addition, the Every Student Succeeds Act requires states to hold
schools, districts, and states to yearly standards of achievement of students on standardized tests in reading and
mathematics.42 These standards are used to determine if schools, districts, and states are making adequate yearly
progress (AYP) as a whole and for specific subgroups of students (including racial and ethnic groups, special
education students, and English language learners). Schools, districts, and states failing to meet the AYP levels of
achievement for 2 consecutive years in the same subject are considered to need improvement and must take
specific steps to improve performance of their students.43
States also generally maintain testing programs and meet federal requirements for testing. Individual states
perform testing in public and charter schools using standardized tests as required by their state department of
education.44 The charter agreement typically requires charter schools to participate in state and national testing
programs.45 Private schools may or may not be required by a state to participate in state-level academic testing.46

BLS education output measure
The BLS output measure for elementary and secondary schools introduced in this article adopts an approach that
relies on student performance on standardized tests for capturing the effects of quality change. Separate
attendance-adjusted series on numbers of students in public and private schools are used as proxies for the
volume of output. To account for the effects of quality change, BLS analysts then applied adjustments based on
national mathematics and reading test score data. Finally, BLS staff aggregated the public and private school
quality-adjusted output measures using expenditure share weights to obtain a measure of overall output for the
elementary and secondary schools industry.
BLS obtained public school enrollment data for students in prekindergarten to grade 12 (pre-K–12) from the
National Center for Education Statistics (NCES), State Nonfiscal Survey of Public Elementary/Secondary
Education.47 Private school enrollment data for students in pre-K–12 are obtained from the NCES Private School
Universe Survey.48 Public and private school enrollments in kindergarten through 12th grade are adjusted for
variations in daily attendance with the use of NCES data on the average daily attendance of public school
students.49

5

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Separate public and private school quality-adjustment series are developed on the basis of students’ mathematics
and reading test scores from the National Assessment of Educational Progress (NAEP) long-term trend (LTT)
testing program for public and private schools.50 These LTT test scores are available (in certain years only) in
various subjects for students ages 9, 13, and 17. BLS interpolates the test score data between testing years in
order to estimate test scores for nontesting years. As shown in this article, for each of the three categories of
schools (all schools, public schools only, and private schools only), a ratio of the reading or mathematics test score
to the perfect score was computed for each year and the mathematics and reading test scores were averaged
together. When the attendance-adjusted student enrollment series for public and private schools were multiplied by
the appropriate test score ratio series, quality-adjusted student output was obtained.
Although quality-adjusted output measures similar to the measure introduced in this article are used in other
countries, such as the United Kingdom, research efforts to account for other dimensions of quality are ongoing.
Research continues toward establishing empirical relationships between other educational characteristics of U.S.
schools and the resulting education outputs. Industry providers themselves in all three types of schools (traditional
public, charter, and private) track numerous metrics, in addition to the number of students enrolled, to measure
their own output. These metrics include
·

performance measures such as student–teacher ratios;

·

parental involvement proxies;

·

high school course difficulty rankings;

·

lesson quality rankings;

·

teacher experience and qualification;

·

student composition;

·

number of AP, IB, or dual credit courses completed;

·

standardized testing of student achievement in selected subjects;

·

percentage of pupils moving up each year;

·

average daily attendance;

·

high school dropout rates;

·

graduation rates;

·

percentage of graduating students enrolling in college; and

·

percentage of transfer requests out of a specific school.

These are only some of the numerous measures often cited in assessing educational programs, and in the future,
they may be used to develop more sophisticated methods of adjusting for quality change in the provision of
educational services. Advances in the economics of education are discussed further in the Appendix.

Labor input and expenditures
6

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

BLS traditionally measures the labor input component of productivity as the total number of hours worked in an
industry. However, not many data of these types are available for primary and secondary education. Instead, BLS
measures labor inputs for NAICS 6111, elementary and Secondary Schools, using data on the number of full-time
equivalent (FTE) employees in detailed employment categories.51 Public school labor input is measured with the
use of NCES data on the number of FTE employees in 16 detailed employment categories.52 These categories
include not only teachers but also other school employees, such as librarians, guidance counselors,
administrators, and student support staff. Private school labor input is drawn from the NCES Schools and Staffing
Survey, Private School Questionnaire.53 BLS combines counts of FTE employees in the detailed employment
categories into broad employment categories. Then to develop labor input measures for public and private
schools, BLS further combines the counts using expenditure share weights.54 BLS constructs these weights for
public school labor inputs using NCES National Public Education Financial Survey data on salaries and benefits.55
Total labor input for the industry is measured as an aggregate of the public and private school labor inputs,
aggregated with the use of the public and private school expenditure share weights.

Trends in elementary and secondary schools
Output. Output in all elementary and secondary schools increased at a 1.2-percent average annual rate from 1989
to 2012, with the greatest rate of growth occurring in the 1990s, as shown in figure 1. Output grew 1.8 percent per
year between 1990 and 2000, slowed to 0.5 percent per year between 2000 and 2007, and then slowed even
further to a 0.1-percent average annual rate from 2007 to 2012. Because public schools constituted just over 90
percent of all elementary and secondary schools, output trends in public schools were similar to those for all
elementary and secondary schools. For private elementary and secondary schools, output increased at only a 0.5percent average annual rate from 1989 to 2012, with the highest period of growth from 1990 to 2000.

7

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

One factor that strongly affects the output measure is school enrollment, which changed along with the population
of school-age children.56 Public school enrollment grew 1.2 percent per year from 1990 to 2007 but only increased
0.2 percent per year from 2007 through 2012.57 Private school enrollment increased steadily from 1990 to 2000 at
an average annual rate of 1.8 percent, then declined 0.5 percent per year from 2000 to 2007, and fell 2.8 percent
per year from 2007 to 2012.58 Increased tuition and fees, a decline in the economic well-being of families, and
increased competition from public schools, particularly charter schools, have been cited as possible explanations
for the decline in private school enrollment.59 Public charter schools have served an increasing number of
students, with enrollment rising from 0.3 million students in 1999 to 2.3 million students in 2012.60
The output measures also reflect variation in the NAEP LTT mathematics and reading student test scores of public
and private schools, which were used as output quality adjusters. Private school student test scores ranged from
8.6 to 22.0 points higher than public school student test scores during the 1989–2012 period (on a 500-point scale)
and were, on average, 14.89 points higher than public school student scores, with the largest differences found in
reading scores.61 Both public and private school student test scores have increased gradually since the long-term
trend assessments of private schools began in 1978.62 For public schools, incorporating the quality adjustment
increased output growth by 0.25 percentage points from 1989 to 2012, while private school output increased by
0.44 percentage points because of the test-score-based quality adjustment.

Labor input. Labor input grew 1.3 percent annually from 1989 to 2012 in elementary and secondary schools, as
shown in figure 2. Average annual growth rates for labor input dropped in each successive subperiod examined, a
pattern also seen for output. Labor input grew 2.1 percent per year from 1990 to 2000, increased more slowly at
1.5 percent per year in 2000–07, and then declined at a 0.5-percent rate from 2007 to 2012. Labor input at private
schools increased 2.3 percent per year from 2000 to 2007, much faster than that for public schools. From 2007
through 2012, labor input declined more rapidly in private schools (1.7 percent per year) than in public schools (0.4
percent per year).

8

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Labor productivity. Labor productivity in elementary and secondary schools, calculated as output per FTE of
labor input, rose from 1990 to 1995 and then declined steadily until 2009. Although productivity rose from 2010
forward, it did not regain the level that it had reached in 1989. The average change in labor productivity for all
elementary and secondary schools over the period, as a whole, was a slight decline of 0.2 percent per year. Figure
3 displays labor productivity trends for the industry and for public and private schools, from 1989 to 2012. Because
public schools educate over 90 percent of the students, the industry and the public school trends are nearly
identical. Public school labor productivity peaked in 1992 and 1995 and then declined steadily until 2009, reflecting
declining growth in output and a somewhat slower decline in labor input growth. The decline in the public school
output growth rate primarily reflects the slowing growth rate of the school-age population. From 1990 to 2012, the
school-age population grew at an average annual rate of 0.79 percent, with a growth rate of 2.02 percent from
1990 to 2000 and 0.26 percent from 2000 to 2012.63 As shown in table 1, beginning in 2009, public school output
growth continued to slow while labor input fell off significantly, with a –0.91-percent growth rate of labor input for
2009 to 2012, resulting in an upturn in public school labor productivity. Labor productivity in private schools
declined sharply after 1993 and even more severely after 2001 than in public schools. This decline appears to be
the result of a more cautious reduction in labor inputs by private schools than by public schools. In addition, private
schools did not experience the same uptick in productivity after 2010 as their public counterparts. Private school
output growth remained positive from 1993 to 2001, while labor input growth increased markedly. From 2001 to
2012, private school output growth retreated to negative values and while labor input growth fell dramatically, labor
productivity growth for private schools remained negative. Figure 4 displays output, labor input, and labor
productivity annual growth rates for the industry overall, public schools only, and private schools only, from 1990 to
2012.

9

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Table 1. Average annual percentage-growth rates of labor productivity, output, and labor input for all
schools, public schools, and private schools, 1989–2012
All schools
Time span

Labor
productivity

1989–2012
1989–93
1993–2001

Output

Public schools
Labor

Labor

input

productivity

Output

Private schools
Labor

Labor

input

productivity

–0.17
.99
–1.08

1.20
2.06
1.66

1.37
1.06
2.77

–0.11
.80
–1.05

1.26
1.98
1.63

1.37
1.17
2.71

–0.91
3.25
–1.40

.08

.56

.47

.25

.73

.47

–2.02

1989–95
1995–2009

.58
–.84

2.07
.98

1.48
1.83

.49
–.73

1.97
1.07

1.48
1.82

1.63
–2.13

2009–12

1.49

.51

–.97

1.62

.70

–.91

–.20

2001–12

Output
0.49
3.09
2.07
–
1.58
3.20
–.14
–
1.90

Labor
input
1.41
–.15
3.53
.46
1.54
2.03
–1.71

Source: U.S. Bureau of Labor Statistics.

Unit labor costs and compensation. The concept of “unit labor costs” compares labor compensation with output
and is a useful gauge of how much output is received over time relative to labor costs, or the “cost
competitiveness” of labor input in the production of output. One may also calculate unit labor costs by dividing
hourly compensation (labor cost per hour) by labor productivity (output per hour). Therefore, an increase in labor
productivity growth offsets the growth of hourly compensation in calculating unit labor costs. A greater rate of labor
productivity growth relative to growth in hourly compensation will result in lower unit labor costs.

10

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

11

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

BLS calculated unit labor costs for public elementary and secondary schools using data on public school salaries
and benefits from the NPEFS. Unit labor costs increased at an average annual rate of 3.4 percent from 1989 to
2012.64 Public school unit labor costs varied over this period, increasing at a rate of 3.6 percent from 1990 to
2000, 4.3 percent from 2000 to 2007, and declining to 1.2 percent from 2007 to 2012. Regarding data from the
National Association of Independent Schools, private school unit labor costs increased at a rate of 5.4 percent for
the 1998–2012 period.65 From 1990 to 2000, private school unit labor costs rose at a rate of 4.2 percent,
increasing to 7.7 percent from 2000 to 2007 and falling to 5.0 percent for 2007–12. For the industry overall, unit
labor costs rose at an average annual rate of 3.6 percent from 1989 through 2012, with a variation of 3.7 percent
from 1990 to 2000, and at a rate of 4.6 percent from 2000 to 2007 and declined to 1.5 percent from 2007 to
2012.66
Labor compensation, the sum of salaries and benefits paid to employees, increased at a rate of 4.8 percent for the
elementary and secondary schools industry from 1989 to 2012. Labor compensation increased steadily in the
industry overall, growing at a rate of 5.5 percent from 1990 to 2000 and 5.6 percent from 2000 to 2007 before
declining to 1.7 percent for 2007–12. This 4.8-percent annual growth in labor compensation from 1989 to 2012 for
the elementary and secondary schools industry is similar to the 4.4-percent annual growth in labor compensation
found in the nonfarm business sector as a whole, one of the broadest aggregates for which productivity measures
are published. However, when combined with flat or declining labor productivity in the industry, this increase in
hourly compensation resulted in relatively rapid increases in unit labor costs.
Public school labor compensation increased at a rate of 4.7 percent during this period (1989–2012), whereas
private school labor compensation increased at a rate of 5.9 percent. Public school labor compensation grew at a
steady 5.4-percent growth rate from 1990 to 2007 before declining to 1.6 percent growth for 2007–12. Labor
compensation of private schools grew at a rate of 6.7 percent from 1990 to 2000 and 7.6 percent from 2000 to
2007 before declining to 2.3 percent from 2007 to 2012. Looking at the components of labor compensation, we find
that benefits grew at a faster rate than salaries in the elementary and secondary schools industry. The gap
between growth in benefits and growth in salaries was particularly wide between 2000 and 2007, with benefits
rising 4.1 percent faster than the rate salaries rose.

Conclusion
The elementary and secondary schools industry must be responsive to changes in the population requiring
educational services. Increased enrollments of students with specific needs, such as English-language learners,
disabled students, and impoverished students, challenge the industry. The industry is also subject to variation in
economic conditions, with public schools facing tighter budget constraints during periods of economic downturn
and private schools facing families with more limited budgets. Variations in factors influencing student educational
outcomes, such as teacher quality, student–teacher ratios, and curriculum quality, also play a role in determining
the output of educational services. In the future, we hope to provide additional information on changes in these
underlying factors and their quantitative impact on educational services. This new measure is a first step toward
understanding the relationship between production of educational services and the labor inputs used in this
production.

12

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

ACKNOWLEDGMENT: We gratefully acknowledge the contributions of Leo Sveikauskas, a research economist
with the Bureau of Labor Statistics, in the summary of recent advances in the economics of education.

Appendix: advances in education economics
Over the last few years, economists have progressed tremendously in understanding some of the central issues in
education. Many research studies have used large datasets to understand school performance and to unravel
connections between school performance and students’ economic success later in life. This appendix summarizes
much of the recent economic research and guides interested readers toward other research topics in which they
may be interested.
We begin with three studies conducted shortly after 2000 that influenced the economics of education considerably.
In the first study, Hanushek and Kimko come to two central conclusions.67 First, countries that have students who
score high on international tests in science and mathematics also have higher rates of economic growth. Second,
immigrants to the United States who come from countries with higher scores also earn more in the United States.
These results suggest that these countries produce high-quality human capital and are thus able to grow more
quickly. Such evidence is also consistent with an emphasis on science and technology education.
In a second study, Hanushek shows that, in most contexts, more resources devoted to education do not lead to
better results.68 A few exceptions to this general rule exist, mostly among young children and disadvantaged
groups. For example, Hanushek remarks that if disadvantaged students were fortunate enough to have strong
teachers, at the 85th percentile, for 5 consecutive years, such a boost in itself would be sufficient to eliminate the
entire gap between mainstream and disadvantaged students.69 What stands out most strongly from this study is
how additional resources generally do not lead to improved results. Findings such as these have led scholars to
conclude that, since added resources do not work, educators will have to fundamentally change the structure of
schools and their incentives to produce better outcomes.
In the final study, Rivkin, Hanushek, and Kain studied the value added of learning of students in Texas schools.70
“Value added” is a measure of a student’s learning in a given year, and it is measured by the increase from the
previous year’s test scores. The value-added measure reflects the “gain” in a student’s test scores compared with
previous years’ scores and controls for family, neighborhood, and school influences on a student. This value-added
approach makes adjusting for individual student differences in learning capability possible. Teacher evaluation by
year-to-year gains in student achievement then become a useful additional measure of teacher effectiveness.
Teacher effects are generally found to be consistent over time: Teachers with high value-added scores within a
given year tend to have similar scores in other years; teachers with low value-added scores tend to have similar
scores in other years. This result has been the basis for a renewed emphasis on measuring and rewarding good
teachers.
After the Rivkin et al. study and other similar work showed that teacher value added could be estimated, further
work analyzing education in terms of teacher value-added data then exploded. New teachers were found to have
below-average teacher scores in their earlier years, particularly in their first year.71 Many teachers with especially
low scores in their early years soon left the profession. Having shown that low income and minority students are
taught more frequently by beginner teachers and experience higher teacher turnover rates, Rivkin et al. and others
argued for implementing policy incentives such as higher pay to retain more experienced, qualified teachers for

13

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

disadvantaged students.72 Teacher scores were uncorrelated with many factors often used in teacher pay, such as
the presence of a graduate degree.73
A recent study has also illustrated the very strong returns associated with early childhood education. James
Heckman demonstrated that attempts to improve learning are much more effective in children’s early years, when
the brain is more malleable.74 Heckman explained many of the basic ideas in clear, nontechnical language.
Heckman’s work often distinguishes between cognitive (learning) skills and noncognitive skills, such as
perseverance and reliability, which have proven to influence children’s future economic success extensively.
In the last few years, economists have published several articles showing the important and enduring affect
teachers have over a student’s lifetime. For example, Chetty et al. showed that unusually effective kindergarten
teachers could create $8,500 to $10,700 greater lifetime earnings per student, in present value, or $170,000 to
$214,000 greater earnings for a class of 20 students.75 Interestingly, the effects of class quality on test scores
faded over time, but the effect on eventual adult income remained operating through noncognitive effects.
Similarly, Chetty, Friedman, and Rockoff found that, as long as they controlled for previous test scores, valueadded measures are an unbiased measure of teachers’ effect on student achievement.76 In further work, Chetty,
Friedman, and Rockoff showed that [measures of teacher value added] are not just measures of effectiveness in
teaching for the test but are useful predictors of future adult income.77 Finally, Chetty and Hendren showed that
students benefit from moving from poor to higher income areas as long as they move before age 13.78
The evidence just summarized illustrates the importance of teacher value-added effects. Nevertheless, valueadded methods have become a controversial topic, especially because, depending on the construction of the
particular measure and the accuracy of student testing, they may mischaracterize the contribution of individual
teachers. The American Statistical Association has cautioned about the use of value-added methods in evaluating
individual teachers.79 Rothstein tested three different value-added measures and found that the measures fail to
uphold some underlying assumptions, including that fifth-grade teacher assignments should not be correlated with
fourth-grade student gains.80 Rothstein finds that students who do exceptionally well in fourth grade trend
downward in gains in fifth grade as their achievements fall back toward the mean gain, and students who do poorly
in fourth grade trend upward in gains in fifth grade as they advance toward the mean. According to Rothstein, the
value-added measures he tested credited teachers for the students assigned, rather than accurately capturing the
value added by the individual teacher.81 Despite such concerns, measures of the value added by teachers are
important. The Gates Foundation (Measures of Effective Teaching [MET] Project) found that constructing a
composite measure of teacher effectiveness that combines information from test score growth, classroom
observations, and student-perception surveys of the classroom environment results in a fair and reliable
measure.82 In addition, using classroom observations of teacher performance and student-perception surveys to
evaluate teachers generates valuable measures in their own right.83 In a pioneering effort, the Gates Foundation is
underwriting a project that will videotape teachers with high value added, analyze the characteristics that high
value-added teachers have in common, and explore how the secrets of their effectiveness can be taught to other
teachers.84
SUGGESTED CITATION

14

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

Susan G. Powers and Steven Flint, "Labor productivity growth in elementary and secondary school services:
1989–2012," Monthly Labor Review, U.S. Bureau of Labor Statistics, June 2016, https://doi.org/10.21916/mlr.
2016.29.
NOTES
1

Michael Greenstone, Max Harris, Karen Li, Adam Looney, and Jeremy Patashnik, “A dozen economic facts about K–12 education,”
Policy Memo, The Hamilton Project (Washington, DC: The Brookings Institution, September 2012), p. 1.
2

Ibid. See figure 1, pp. 1–2.

3 According

to the U.S. Bureau of Labor Statistics, employment in all occupations in North American Industry Classification System
(NAICS) code 61, educational services, was 12,758,610 in May 2014.
4

Ibid. Elementary and secondary schools, NAICS code 61111, employed 8,308,980 individuals in all occupations in the industry as of
May 2014.
5

Digest of education statistics, table 106.10, “Expenditures of educational institutions related to the gross domestic product, by level
of institution: selected years, 1929–30 through 2013–14” (Washington, DC: U.S. Department of Education, Institute of Education
Sciences, National Center for Education Statistics, 2014), https://nces.ed.gov/programs/digest/d14/tables/dt14_106.10.asp.
6

See 2007 County Business Patterns and 2007 Economic Census. Also, see 2013 County Business Patterns, Statistics of U.S.
Businesses Main (U.S. Bureau of the Census), https://www.census.gov/econ/susb/.
7

Prekindergarten students attending academic prekindergarten programs affiliated with elementary and secondary schools in this
industry are typically included in data for this industry.
8

Digest of education statistics, table 203.10, “Enrollment in public elementary and secondary schools, by level and grade;” table
214.10, “Number of public school districts and public and private elementary and secondary schools”; and table 205.10, “Private
elementary and secondary school enrollment and private enrollment as a percentage of total enrollment in public and private schools,
by region and grade level (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for
Education Statistics, 2014), https://nces.ed.gov/programs/digest/d14/.
9

Ibid.

10

For more information, see Susan Aud, William Hussar, Frank Johnson, Grace Kena, Erin Roth, Eileen Manning, Xiaolei Wang, and
Jijun Zhang, The condition of education 2012, NCES 2012-045, appendix A, table A-4–2, “Number, percentage, and percentage
distribution of public charter schools and students, by region and state or jurisdiction” (Washington, DC: U.S. Department of
Education, Institute of Education Sciences, National Center for Education Statistics, May 2012), p. 134, http://nces.ed.gov/
pubs2012/2012045.pdf; and Digest of education statistics, table 216.30, “Number and percentage distribution of public elementary
and secondary students and schools, by traditional or charter school status and selected characteristics: selected years, 1999–2000
through 2012” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education
Statistics, 2014), https://nces.ed.gov/programs/digest/d14/tables/dt14_216.30.asp.
11

For additional information, see Center for Research on Education Outcomes, Stanford University, National Charter School Study,
2013; and The nation’s report card: national assessment of educational progress, “America’s charter schools: results from the NAEP
2003 pilot study,” NCES 2005-456 (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center
for Education Statistics, December 2004), http://nces.ed.gov/nationsreportcard/pdf/studies/2005456.pdf.
12

Digest of education statistics, table 206.10, “Number and percentage of homeschooled students ages 5 through 17 with a grade
equivalent of kindergarten through 12th grade, by selected child, parent, and household characteristics” (Washington, DC: U.S.
Department of Education, Institute of Education Sciences, National Center for Education Statistics, 2014), https://nces.ed.gov /
programs/digest/d14/tables/dt14_206.10.asp.

15

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

13

Frederick Huntley Magison and Thomas Tracy Bouvé, The statute law of municipal corporations in Massachusetts (Albany, NY:
Mathew Bender and Company, 1917), p. 208; and American history online, Facts on File, Inc., http://online.infobase.com/HRC/
Search/Details/201886?q=laws%20of%201642.
14

For a chronology of federal education legislation, see Digest of education statistics 2012, chapter 4, “Federal programs for
education and related activities,” NCES 2014-015 (Washington, DC: U.S. Department of Education, Institute of Education Sciences,
National Center for Education Statistics, December 2013), pp. 589–597, http://nces.ed.gov/pubs2014/2014015.pdf.
15

School boards develop policies and regulations to control the operation of the schools in a school district, including system
organization, school site location, school finance, equipment purchase, staffing, attendance, curriculum, extracurricular activities, and
other functions, and may also be authorized to levy taxes, initiate eminent domain proceedings, acquire land, and assume bonded
indebtedness. See Joseph Beckham and Barbara Klaymeier Wills, Education encyclopedia, “Duties, responsibilities, decision-making
and legal basis for local school board powers,” http://education.stateuniversity.com/pages/2391/School-Boards.html.
16

Digest of education statistics 2012, chapter 4, “Federal programs for education and related activities.”

17

These students often require additional support through the public school system, such as breakfast and lunch programs, safe
after-school programs, and tutoring, to compensate for their reduced access to in-home and in-community resources.
18

The National Council of State Supervisors for Languages (NCSSFL) provides documentation on foreign language high school
graduation requirements by state at http://ncssfl.org/.
19

State regulation of private schools (Washington, DC: U.S. Department of Education, Office of Innovation and Improvement, July
2009), https://www2.ed.gov/admins/comm/choice/regprivschl/regprivschl.pdf.
20

Council on American private education, “Private schools and the Every Student Succeeds Act” (Germantown, MD: Council for
American Private Education, January 2016), http://www.capenet.org/pdf/ESSACAPE.pdf.
21

State regulation of private schools, 2009.

22

Martha Naomi Alt and Katharin Peter, Findings from the condition of education 2002, “Private schools: a brief portrait,” NCES
2002–013 (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics,
2002), pp. 21–23, http://nces.ed.gov/pubs2002/2002013.pdf.
23

Ibid.

24

Ibid.

25

See William C. Symonds, “For-profit schools: They’re spreading fast. Can private companies do a better job of educating America’s
kids?” BusinessWeek, February 7, 2000. In 2000, the United States had about 200 for-profit elementary and secondary schools
serving roughly 100,000 of 53 million students in grades K–12. Although nominally for-profit organizations, few of these schools are in
fact profitable, http://www.bloomberg.com/news/articles/2000-02-06/for-profit-schools.
26

For more information on mothers who continue to work while raising their young children, see Robert Kaestner, Darren Lubotsky,
and Javaeria Qureshi, “Mother’s employment by child age and its implications for theory and policy,” paper submitted to National
Bureau of Economic Research, April 12, 2016, pp. 2–31; Digest of education statistics, table 202.20, “Percentage of 3-, 4-, and 5year-old children enrolled in preprimary programs, by level of program, attendance status, and selected child and family
characteristics: 2014” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education
Statistics, 2014), http://nces.ed.gov/programs/digest/d15/tables/dt15_202.20.asp; and Digest of education statistics, table 42,
“Number of preschool children under 6 years old, percentage in center-based programs, average hours in nonparental care, and
percentage in various types of primary care arrangements, by selected child and family characteristics: 2005” (Washington, DC: U.S.
Department of Education, Institute of Education Sciences, National Center for Education Statistics, 2006), http://nces.ed.gov/
programs/digest/d06/tables/dt06_042.asp.

16

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

27

Issue brief, “English language learner students in U.S. public schools: 1994 and 2000,” NCES 2004–035 (Washington, DC: U.S.
Department of Education, Institute of Education Sciences, National Center for Education Statistics, August 2004), p. 1, http://
nces.ed.gov/pubs2004/2004035.pdf; Digest of education statistics, table 204.20, “Number and percentage of public school students
participating in programs for English language learners, by state: selected years, 2002–03 through 2011–12” (Washington, DC: U.S.
Department of Education, Institute of Education Sciences, National Center for Education Statistics, 2014), https://nces.ed.gov /
programs/digest/d13/tables/dt13_204.20.asp; and Jennifer Robinson, Xiaolei Wang, Amy Rathbun, Jijun, Zhang, Sidney WilksonFlicker, Amy Barmer, and Erin Dunlop Velez, The condition of education 2015, NCES 2015-144 (Washington, DC: U.S. Department of
Education, Institute of Education Sciences, National Center for Education Statistics, May 2015), p. 85, http://nces.ed.gov/
pubs2015/2015144.pdf.
28

Ibid, p. 50.

29

Digest of education statistics, table 362, “Public school students receiving publicly funded free or reduced price lunch, by selected
school characteristics: school year 1990–91” (Washington, DC: U.S. Department of Education, Institute of Education Sciences,
National Center for Education Statistics, 1993), https://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=93292 ; and Digest of education
statistics, table 204.10, “Number and percentage of public school students eligible for free or reduced-price lunch, by state: selected
years, 2000–01 through 2011–12” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center
for Education Statistics, 2014), https://nces.ed.gov/programs/digest/d13/tables/dt13_204.10.asp.
30

Digest of education statistics, table 204.30, “Children 3 to 21 years old served under Individuals with Disabilities Education Act, Part
B, by type of disability: selected years, 1976–77 through 2012–13” (Washington, DC: U.S. Department of Education, Institute of
Education Sciences, National Center for Education Statistics, 2014), https://nces.ed.gov/programs/digest/d14/tables/dt14_204.30.asp.
31

Digest of education statistics, table 204.60, “Percentage distribution of students 6 to 21 years old served under Individuals with
Disabilities Education Act, Part B, by educational environment and type of disability: selected years, fall 1989 through fall
2012” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics,
2014), https://nces.ed.gov /programs/digest/d14/tables/dt14_204.60.asp.
32

Digest of education statistics, table 103.20, “Percentage of the population 3 to 34 years old enrolled in school, by age group:
selected years, 1940 through 2013” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center
for Education Statistics, 2014), https://nces.ed.gov /programs/digest/d14/tables/dt14_103.20.asp.
33

See table 7, “Employment status of women by presence and age of youngest child,” March 1975–March 2013, in BLS reports,
“Women in the labor force: a databook,” Report 1052 (U.S. Bureau of Labor Statistics, December 2014), p. 25, https://www.bls.gov/
opub/reports/womens-databook/archive/women-in-the-labor-force-a-databook-2014.pdf.
34

For further discussion, see Barbara M. Fraumeni, Marshall B. Reinsdorf, Brooks B. Robinson, and Matthew P. Williams, “Price and
real output measures for the education function of government: exploratory estimates for primary and secondary education,” NBER
Working Paper no. 14099 (Cambridge, MA: National Bureau of Economic Research, June 2008), http://www.nber.org/papers/w14099.
35

Mark K. Sherwood, “Difficulties in the measurement of service outputs,” Monthly Labor Review, March 1994, pp. 11–19, https://
www.bls.gov/mlr/1994/03/art2full.pdf.
36

Paul Schreyer, Towards measuring the volume output of education and health services: a handbook, Working Paper no. 31 (Paris,
France: Organisation for Economic Co-operation and Development, April 28, 2010), pp. 49–52, http://www.oecd-ilibrary.org/
economics/towards-measuring-the-volume-output-of-education-and-health-services_5kmd34g1zk9x-en.
37

Ibid.

38

Ibid.

39

Sources & methods, “Public service productivity estimates: education” (Newport, South Wales: Office for National Statistics, July
2012), p. 5.
40

Fraumeni et al., “Price and real output measures for the education function of government,” p. 20.

17

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

41

Every Student Succeeds Act, Pub. L. No. 114–95, December 10, 2015, https://www.congress.gov/114/plaws/publ95/
PLAW-114publ95.pdf.
42

Ibid.

43

Ibid.

44

Commonly used standardized tests include the Iowa Tests of Basic Skills for K–8 and the Iowa Tests of Educational Development,
used for grades 9–12, published by Riverside Publishing/Houghton Mifflin; the Stanford Achievement Test, Ninth Edition, published by
Harcourt Educational Measurement; and the Michigan Educational Assessment Program (MEAP), devised to assess educational
achievements of students in the Michigan school systems.
45

Lee Anderson, Nancy Adelman, Kyo Yamashiro, Mary Beth Donnelly, Kara Finnigan, Jose Blackorby, and Lynyonne Cotton,
“Evaluation of the public charter schools program: year one evaluation report” (Washington, DC: U.S. Department of Education, Office
of the Under Secretary, Planning and Evaluation Service, Elementary and Secondary Education Division, 2000), pp. 43–58, https://
www2.ed.gov/rschstat/eval/choice/pcsp-year1/year1report.pdf.
46

Josh Cunningham, Improving school choice in the states, “Accountability in private school choice programs” (Washington, DC:
National Conference of State Legislatures, December 2014), p. 4, http://www.ncsl.org/documents/educ/
AccountabilityInPrivateSchoolChoice.pdf.
47

The State Nonfiscal Survey of Public Elementary/Secondary Education is one of five annual surveys comprising the U.S.
Department of Education, National Center for Education Statistics Common Core of Data (CCD). The CCD is a national statistical
program that collects and compiles administrative data from state education agencies covering the universe of all public elementary
and secondary schools and school districts in the United States. Only prekindergarten students enrolled in a group or class that is part
of a public school program taught during the year or years preceding kindergarten, excluding Head Start students unless part of an
authorized public education program of a local education agency, are included in the survey. For more information, see http://
nces.ed.gov/ccd/stnfis.asp.
48

The National Center for Education Statistics (NCES) Private School Universe Survey of the U.S. Department of Education
produces data similar to that of the NCES Common Core of Data for the public schools. Private schools are included in the survey
when they teach at least one of grades 1–12, ungraded students between 5 and 18 years old, kindergarten (traditional year of school
primarily for 5-year-olds before first grade), transitional kindergarten (extra year of school for kindergarten-age children who are
judged not ready for kindergarten), or transitional first grade (extra year of school for children who have attended kindergarten but
have been judged not ready for first grade). Early childhood programs and daycare centers that teach kindergarten, transitional
kindergarten, or transitional first grade are also included in the survey. For more information, see http://nces.ed.gov/surveys/pss/
index.asp.
49 Average

daily attendance data are obtained from the U.S. Department of Education, National Center for Education Statistics,
National Public Education Financial Survey, http://nces.ed.gov/ccd/stfis.asp. Because average daily attendance data are not available
for private schools, the public school data on average daily attendance are used for adjusting both public and private school
enrollment for variations in daily attendance. The attendance of prekindergarten student enrollment is not adjusted because
attendance is not compulsory for this group of students.
50

The National Assessment of Educational Progress (NAEP) is the largest nationally representative and continuing assessment of
elementary and secondary school students in the United States. Assessments are conducted periodically in mathematics, reading,
science, writing, the arts, civics, economics, geography, and U.S. history. The NAEP (also known as “The Nation’s Report Card”; see
endnote 11) is a congressionally mandated assessment in various subject areas administered by the National Center for Education
Statistics, a branch of the U.S. Department of Education. Results are summarized only at the national, state, and trial urban district
(Trial Urban District Assessment [TUDA]) levels. The NAEP created the TUDA in 2002, beginning with six urban districts participating
in reading and writing assessments, to support the improvement of student achievement in the nation’s large urban districts. In 2009,
18 districts participated in mathematics, reading, and science. In 2011, 2013, and 2015, 21 districts participated. NAEP assessments
are administered uniformly with use of the same sets of test booklets across the nation and, as a result, serve as a common metric for

18

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

all states and selected urban districts. The assessment stays essentially the same from year to year, with only carefully documented
changes. This process permits NAEP to provide a clear picture of student academic progress over time. Long-term trend assessment
in mathematics and reading is conducted differently from the NAEP’s main assessments, and the two types of assessments are not
comparable. For additional information, see http://nces.ed.gov/nationsreportcard/about/.
51

These categories include elementary and secondary school teachers, instructional aides, guidance counselors, librarians and
library support staff, school administrators (principals, assistant principals, head masters, assistant heads), school administrative
support staff, psychologists, speech therapists, audiologists, school nurses, attendance officers, cafeteria staff, bus drivers, custodial
and building maintenance staff, and security staff. Public school systems also employ local education agency administrators such as
school district superintendents, assistant superintendents and deputy superintendents, administrative assistants, business managers,
administrative support staff, and instructional coordinators. Although private schools do not require local education agency staff, they
may employ business officers, admissions officers, development officers, and directors of studies.
52

These data are obtained from the National Center for Education Statistics, State Nonfiscal Survey of Public Elementary/Secondary
Education. The 16 public school detailed employment categories include prekindergarten teachers, kindergarten teachers, elementary
school teachers (excluding prekindergarten and kindergarten teachers), secondary school teachers, ungraded teachers, instructional
aides, instructional coordinators, guidance counselors, librarians, library support staff, local education agency administrators, local
education agency administrative support staff, school administrators, school administrative support staff, student support services
staff, and all other support staff.
53

Private school detailed employment categories include principals, assistant principals, other managers, instruction coordinators,
teachers (grades K–12), teacher aides, other aides, guidance counselors, librarians and media specialists, librarians and media
center aides, nurses, student support staff (includes student support services professional staff, such as school psychologists, social
workers, and speech therapists), secretaries and clerical staff, food service personnel, custodial and maintenance staff, and other
employees (includes health and other noninstructional aides and other employees not identified by function). These data are available
every 4 to 6 years. Data points between available years are interpolated with use of average annual growth rates. Because the
Schools and Staffing Survey, Private School Questionnaire, gathers data on number of teachers for grades K–12 only, we estimate
the number of private school prekindergarten teachers for each year by dividing annual private school prekindergarten student
enrollment by the private school student–teacher ratio for that year.
54

We defined the six broad employment categories to closely match the six public school expenditure categories available in the
National Center for Education Statistics (NCES), National Public Education Financial Survey. Expenditure share weights for each of
the six broad employee categories are calculated with the use of NCES National Public Education Financial Survey data on salaries
and benefits ( http://nces.ed.gov/ccd/stfis.asp). The use of expenditure share weights in constructing nonmarket output and input
index measures is discussed by W. Erwin Diewert in “The measurement of nonmarket outputs and inputs using cost weights,”
Discussion Paper no. 08-03 (Vancouver, BC, Canada: University of British Columbia, April 2008); and in chapter 16, System of
National Accounts 1993, Eurostat, IMF, OECD, UN, and the World Bank (New York: United Nations, 1993).
55

See Stephen Q. Cornman, “Documentation for the NCES Common Core of Data, National Public Education Financial Survey
(NPEFS), school year 2010–11 (fiscal year 2011),” NCES 2014-343 (Washington, DC: U.S. Department of Education, Institute of
Education Sciences, National Center for Education Statistics, December 2013), http://nces.ed.gov/ccd/pdf/stfis111agen.pdf.
56

Digest of education statistics, 2012, chapter 1, “All levels of education” (Washington, DC: U.S. Department of Education, Institute of
Education Sciences, National Center for Education Statistics, December 2012), http://nces.ed.gov/programs/digest/d12/ch_1.asp.
57

See the State Nonfiscal Survey of Public Elementary/Secondary Education, 1985–86 through 2011–12, U.S. Department of
Education, National Center for Education Statistics Common Core of Data, https://nces.ed.gov/ccd/stnfis.asp.
58

For more information, see U.S. Department of Education, National Center for Education Statistics, Private School Universe Survey,
https://nces.ed.gov/surveys/pss/.
59

Stephanie Ewert, “The decline in private school enrollment,” Working Paper no. FY12-117 (U.S. Census Bureau, Social, Economic,
and Housing Statistics Division, January 2013), https://www.census.gov/hhes/school/files/ewert_private_school_enrollment.pdf.

19

U.S. BUREAU OF LABOR STATISTICS

60

MONTHLY LABOR REVIEW

Digest of education statistics, table 216.30.

61

Because private school participation rates in the NAEP long-term testing program fell overall below the required standard for
reporting results in 2012, private school test scores for 2012 are estimated with the use of the 2008 values.
62

Differences between long-term trend reading and mathematics test scores administered by Catholic schools and public schools are
presented in The nation’s report card: trends in academic progress, 2012, NCES 2013-456 (Washington, DC: U.S. Department of
Education, Institute of Education Sciences, National Center for Education Statistics, 2012), pp. 42–43, http://nces.ed.gov/
nationsreportcard/subject/publications/main2012/pdf/2013456.pdf.
63

Digest of education statistics, table 101.40, “Estimated total and school-age resident populations, by state: selected years, 1970
through 2013” (Washington, DC: U.S. Department of Education, Institute of Education Sciences, National Center for Education
Statistics, 2014), https://nces.ed.gov/programs/digest/d14/tables/dt14_101.40.asp.
64

For additional information, see U.S. Department of Education, National Center for Education Statistics, Common Core of Data
National Public Education Financial Survey, https://nces.ed.gov/ccd/stfis.asp.
65

The National Association of Independent Schools (NAIS) generously provided mean and median salary data by employee category
for private schools. Total annual private school salary expenditures were estimated with use of data on the number of private school
employees by employee category from the U.S. Department of Education, National Center for Education Statistics, Schools and
Staffing Survey, Private School Questionnaire, and NAIS salary data by type of employee. Because private school benefits data were
unavailable, private school benefits were estimated with the use of private school salary data and the ratio of public school benefits to
salaries. This estimation approach constrains estimated private school benefits to the public school benefits growth rate, and as a
result, the private school benefit data are not published separately. The U.S. Bureau of Labor Statistics, Occupational Employment
Statistics program (OES), began publishing wage data by occupation for private schools in this industry in 2009. The OES data may
eventually be used to replace the NAIS data, which are not from a representative sample of private schools.
66

Labor compensation is defined to include salaries and benefits. Labor compensation for public schools is measured with use of
data on salary and benefit expenditures from the National Center for Education Statistics (NCES), National Public Education Financial
Survey. While extensive and detailed expenditure data are available for public schools, little expenditure data are available for private
schools. Using NCES data on number of employees and historical salary data from the National Association of Independent Schools
(NAIS), we construct estimates of salary expenditures for each of 10 detailed private school employee categories. Benefits
expenditures for private schools are not available from the NAIS. To estimate private school benefits, we computed the ratio of public
school benefits expenditures to public school salaries expenditures and applied this ratio to total salaries expenditures obtained from
the NAIS. We obtained aggregate elementary and secondary industry compensation by summing total public and private labor
compensation.
67

Eric A. Hanushek and Dennis D. Kimko, “Schooling, labor-force quality, and the growth of nations,” American Economic Review,
vol. 90, no. 5, December 2000, pp. 1,184–1,208, https://www.aeaweb.org/articles.php?doi=10.1257/aer.90.5.1184.
68

Eric A. Hanushek, “The failure of input-based schooling policies,” The Economic Journal, vol. 113, no. 485, February 2003, pp.
F64–F98, http://onlinelibrary.wiley.com/doi/10.1111/1468-0297.00099/abstract.
69

Ibid, pp. F31–F32.

70

Steven G. Rivkin, Eric A. Hanushek, and John F. Kain, “Teachers, schools, and academic achievement,” Econometrica, vol. 73, no.
2, March 2005, pp. 417–458, http://onlinelibrary.wiley.com/doi/10.1111/j.1468-0262.2005.00584.x/abstract.
71

Ibid, pp. 447–448.

72

Ibid, p. 450.

73

Ibid, p. 419.

74

James J. Heckman, “Schools, skills, and synapses,” Economic Inquiry, vol. 46, no. 3, July 2008, pp. 289–324, http://
onlinelibrary.wiley.com/doi/10.1111/j.1465-7295.2008.00163.x/abstract.

20

U.S. BUREAU OF LABOR STATISTICS

MONTHLY LABOR REVIEW

75

Raj Chetty, John N. Friedman, Nathaniel Hilger, Emmanuel Saez, Diane Whitmore Schanzenbach, and Danny Yagan, “How does
your kindergarten classroom affect your earnings: evidence from Project STAR,” Quarterly Journal of Economics, vol. 126, no. 4,
November 2011, pp. 49, 1,593–1,660, http://qje.oxfordjournals.org/content/126/4.toc.
76

Raj Chetty, John N. Friedman, and Jonah E. Rockoff, “Measuring the impacts of teachers I: evaluating bias in teacher value-added
estimates,” American Economic Review, vol. 104, no. 9, September 2014, pp. 2,593–2,632, https://www.aeaweb.org/articles.php?
doi=10.1257/aer.104.9.2593.
77

Chetty et al., “Measuring the impacts of teachers II: teacher value-added and student outcomes in adulthood,” pp. 2633–2679,
https://www.aeaweb.org/articles.php?f=s&doi=10.1257/aer.104.9.2633.
78

Raj Chetty and Nathaniel Hendren, “The impacts of neighborhoods on intergenerational mobility: childhood exposure effects and
county-level estimates,” unpublished paper, May 2015, http://scholar.harvard.edu/hendren/publications/impacts-NeighborhoodsIntergenerational-Mobility-Childhood-Exposure-Effects-And.
79

“ASA statement on using value-added models for educational assessment” (American Statistical Association, April 8, 2014), http://
www.amstat.org/policy/pdfs/asa_vam_statement.pdf.
80

Jesse Rothstein, “Teacher quality in educational production: tracking, decay, and student achievement,” The Quarterly Journal of
Economics, vol. 125, no. 1, February 2010, pp. 175–214, http://qje.oxfordjournals.org/content/125/1/175.abstract.
81

Ibid.

82

“Ensuring fair and reliable measures of effective teaching: culminating findings from the MET project’s three-year study” (MET
Project: Policy and Practitioner Brief, January 2013), p. 5, http://www.edweek.org/media/17teach-met1.pdf.
83

Ibid, p. 20.

84

Valerie Strauss, “Bill Gates’s $5 billion plan to videotape America’s teachers,” The Washington Post, May 10, 2013, https://
www.washingtonpost.com/news/answer-sheet/wp/2013/05/10/bill-gatess-5-billion-plan-to-videotape-americas-teachers/.

RELATED CONTENT

Related Articles
Do performance-based incentives work for teachers? Monthly Labor Review, May 2014.
Are student test scores an accurate measure of high school teachers’ effectiveness? Monthly Labor Review, August 2013.
Measuring productivity in service industries, Monthly Labor Review, June 1982.

Related Subjects
Income
methods

Education and training

Labor dynamics

Survey methods

Employment

21

Productivity

Statistical programs and