On April 9, 1997, the Office of Management and Budget (OMB) announced the
adoption of a new industry classification replacing the Standard Industrial Classifications (SIC), a system used to
classify most of the data we have about industries or kinds of business in our economy. The
updating of industry classifications is nothing new. Since its origination in the 1930s, the SIC system has
been revised or updated every 10 or 15 years to reflect new developments in the American
economy and to address problems identified by data users and statistical agencies.
The most recent change to the SIC system occurred in 1987. That revision identified a
number of new high tech industries, tripled the number of classifications within computer-related services, and gave us our first industry categories for computer and software stores,
video tape rental stores, and manufacturers of plastic bottles. Nonetheless, the 1987
revision left three quarters of all industries unchanged, and left the broad structure and
hierarchy intact, including such basic sector groupings as manufacturing, retail trade,
services, and construction.
The objectives for the 1997 revision were much broader. Not only was the system to
identify new industries, but the process also sought to reorganize the system according to a
more consistent economic principle--according to types of production activities performed--rather than the mixture of production-based and market-based categories in the SIC. That
reorganization would allow for the presentation of more detail for the rapidly expanding
service sector that accounts for most economic activity but only 40 percent of SIC
categories. Further, the system was redefined jointly with Canada and Mexico so that
comparable statistics could be obtained for the three NAFTA trading partners.
In 1992, OMB formed the U.S. Economic Classification Policy Committee (ECPC), chaired by
the Bureau of Economic Analysis (BEA), and staffed by BEA, the Bureau of Labor Statistics,
and the Census Bureau. The ECPC worked together with Statistics Canada and Mexico's
Instituto Nacional de Estadistica, Geografia e Informatica (INEGI). This international
dimension became so important that, despite national variations, all three
countries are referring to the new industry classification as the North American Industry Classification System (NAICS).
The NAICS proposal was fleshed out, industry by industry, in a series of
Federal Register
notices from 1994 through 1996. The entire proposed classification
system appears in the Federal Register notice of April 9, 1997, and
a final revision is expected in the fall of 1997. The definitive reference,
the North American Industry Classification System--United States, 1997, counterpart
to the 1987 SIC Manual, will be published in spring 1998.
Table 1 lists some of the new industries being separately recognized for the first time with NAICS. A few of them reflect "high tech" developments such as fiber optic cable manufacturing, satellite communications, and the reproduction of computer software. More of them recognize less technological changes in the way business is done: bed and breakfast inns, environmental consulting, warehouse clubs, pet supply stores, credit card issuing, diet and weight reduction centers. Taken together, these new industries provide an interesting review of the profound ways our economy has changed over recent years. |
Table 1: Selected new U.S. industries being identified in NAICS
|
New Sectors
NAICS groups the economy into 20 broad sectors, up from the 10 divisions of the SIC
system. Many of the new sectors reflect recognizable parts of SIC divisions, such as the
Utilities and Transportation sectors, broken out from the SIC division Transportation,
Communications, and Utilities. Similarly, the SIC division for Service Industries has been
subdivided to form several new sectors with longer names: Professional, Scientific and
Technical Services; Management, Support, Waste Management, and Remediation Services;
Education Services; Health and Social Assistance; Arts, Entertainment, and Recreation; and
Other Services except Public Administration.
Other sectors represent combinations of pieces from more than one SIC division. The new
Information sector includes major components from Transportation, Communications, and
Utilities (broadcasting and telecommunications), Manufacturing (publishing), and Services
Industries (software publishing, data processing, information services, motion picture and
sound recording). The Accommodation and Foodservices sector puts
together hotels and other lodging places from Service Industries and eating and drinking
places from Retail Trade.
Table 2. NAICS Sectors and their Corresponding SIC Divisions
Code | NAICS Sectors | SIC Divisions Making the Largest Contributions |
11 | Agriculture, Forestry, Fishing and Hunting | Agriculture, Forestry and Fishing Manufacturing |
21 | Mining | Mineral Industries
Manufacturing |
22 | Utilities | Transportation, Communication, and Utilities |
23 | Construction | Construction Industries
Finance, Insurance, and Real Estate |
31-33 | Manufacturing | Manufacturing
Retail Trade |
42 | Wholesale Trade | Wholesale Trade |
44-45 | Retail Trade | Retail Trade Wholesale Trade |
48-49 | Transportation and Warehousing | Transportation, Communication, and Utilities
Service Industries |
51 | Information | Transportation, Communication, and Utilities Manufacturing Service Industries |
52 | Finance and Insurance | Finance, Insurance, and Real Estate
Retail Trade |
53 | Real Estate and Rental and Leasing | Finance, Insurance, and Real Estate Service Industries Transportation, Communication, and Utilities |
54 | Professional, Scientific, and Technical Services | Service Industries
Finance, Insurance, and Real Estate Agriculture, Forestry, and Fishing |
55 | Management of Companies and Enterprises | Financial, Insurance, and Real Estate auxiliary establishments in all industries |
56 | Administrative and Support, Waste Management and Remediation Services | Service Industries Transportation, Communication, and Utilities Manufacturing Construction Industries |
61 | Education Services | Service Industries |
62 | Health Care and Social Assistance | Service Industries
Transportation, Communication, and Utilities |
71 | Arts, Entertainment, and Recreation | Service Industries Retail Trade Finance, Insurance, and Real Estate |
72 | Accommodation and Foodservices | Retail Trade Service Industries |
81 | Other Services (except Public Administration) | Service Industries Finance, Insurance, and Real Estate Manufacturing Agriculture, Forestry, and Fishing |
92 | Public Administration | Public Administration Service Industries |
New Numbering System
NAICS industries are identified by a 6-digit code, in contrast to the 4-digit SIC code. The
longer code accommodates the larger number of sectors and allows more flexibility in
designating subsectors. It also provides for additional detail not necessarily appropriate for
all three NAICS countries. The international NAICS agreement fixes only the first five digits
of the code. The sixth digit, where used, identifies subdivisions of NAICS industries that
accommodate user needs in individual countries. Thus, 6-digit U.S. codes may differ from
counterparts in Canada or Mexico, but at the 5-digit level they are standardized.
Table 3 illustrates the hierarchic structure of the NAICS codes.
Table 3. Examples of NAICS Hierarchy
NAICS level | Example #1 | Example #2 | ||
NAICS code | Description | NAICS code | Description | |
Sector | 31-33 | Manufacturing | 51 | Information |
Subsector | 334 | Computer and electronic product manufacturing | 513 | Broadcasting and telecommunications |
Industry group | 3346 | Manufacturing and reproduction of magnetic and optical media | 5133 | Telecommunications |
Industry | 33461 | Manufacturing and reproduction of magnetic and optical media | 51332 | Wireless telecommunications carriers, except satellite |
U.S. Industry | 334611 | Reproduction of software | 513321 | Paging |
Dealing with Breaks in Time Series
The creation of new sectors and the expansion of the industry code from four to six digits
certainly suggest that this revision of the industry classification system is more profound than
earlier SIC revisions. Yet neither of these changes necessarily affects the ability to link old
data on an SIC basis to new data on a NAICS basis.
Data for more than two-thirds of all 4-digit SICs will be derivable from the NAICS system, either because the industry is not being changed (other than in code), or because new industries are being defined as subdivisions of old ones. For example--
On the other hand, many other industries are being changed more profoundly, leading to breaks in the availability of time series data.
The total number of industry classifications is increasing with NAICS (1170 industries
applicable in the U.S., up from 1004 SICs), as illustrated by the subdivision of SICs 4953
and 7389 cited above. Of the 1170 industries, 358 are new--not previously recognized
separately under SIC, 422 are substantially unchanged while 388 represent
revisions to the scope of existing industries. (Two are not applicable in the U.S.)
Correspondence Tables
Users may now examine the detail of the NAICS system using extensive tables recently
published in Federal Register notices. The first table (illustrated in table 4, below) shows the
full NAICS hierarchy, and the 1987 SICs or parts of SICs that comprise each
NAICS industry. A status code of E identifies existing industries
without significant change, R for revised
industries, and N for new industries. An asterisk (*) preceding an SIC code indicates that
only part of the SIC is contributing to the NAICS category on that line, a part defined in
parentheses in the 1987 SIC description.
Table 4. 1997 NAICS Matched to 1987 SIC (excerpts from
table 1 in Federal Register)
NAICS code | NAICS description | Status code | SIC code | 1987 SIC description |
5111 | Newspaper, periodical, book and database publishing | |||
51111 | Newspaper publishing | E | 2711 | Newspapers: publishing or publishing and printing |
51112 | Periodical publishing | E | 2721 | Periodicals: publishing or publishing and printing |
51113 | Book publishiing | E | 2731 | Books: publishing or publishingan printing |
51114 | Database publishing | N | *2741 | Miscellaneous publishing (database publishers) |
511191 | Greeting card publishing | E | *2771 | Greeting cards |
511199 | All other publishing | R | *2741 | Miscellaneous publishing (exc. database publishing) |
5112 | Software publishing | |||
51121 | Software publishing | R | *7372 | Prepackaged software (software publishing) |
5613 | Employment services | |||
56131 | Employment placement agencies | R | *7361
*7819
*7922 |
Employment agencies (except executive placing)
Services allied to motion picture production (casting bureaus) Theatrical producers and miscellaneous theatrical services (casting agencies) |
56132 | Temporary help services | N | *7363 | Help supply services (except employee leasing service) |
56133 | Employee leasing services | N | *7363 | Help supply services (except temporary help supply service) |
The second table (illustrated in table 5, below) lists all 4-digit SICs along with their
counterpart NAICS categories. The abbreviation "pt." means "part of", and indicates that this
component will be mixed with pieces from other SICs to form the NAICS industry.
A @ symbol in the SIC code column indicates that this category cannot reasonably be
approximated from NAICS-based data, i.e., there will be a break in the time series that is
estimated to be
greater than 3 percent of the 1992 value of output (e.g., sales, shipments) for the 1987
industry.
Table 5. 1987 SIC Compared to 1997 NAICS (excerpt from
table 2 in Federal Register)
SIC code | 1987 SIC description | NAICS code | 1997 U.S. description |
7361@ | Employment agencies Executive placing services Except executive placing services |
561612 56131 |
Human resource consulting (pt) Employment placement agencies (pt) |
7363 | Help supply services Temporary help supply Employee leasing services |
56133 56132 |
Temporary help services Employee leasing services |
The full detail of both tables is accessible
here in several formats, including a format easily navigated on the internet,
database files in ASCII or dBase formats, portable document format (PDF),
and a word processor format.
Other contents of Federal Register notices, issue papers, and implementation
schedules may also be accessed at the NAICS web site:
http://www.census.gov/epcd/www/naics.html.
Generating Comparable Data
Economic data for only one point in time have limited value. Most of the numbers we use
take on greater meaning when compared to data for other time periods, allowing inference of
development or change. The implementation of NAICS will cause major disruptions in the
availability of such time series information, not only for individual industries that are
redefined but also for the broad sectors, like manufacturing and retailing, that we use to
describe our economy in everyday conversation.
The U.S. Economic Classification Policy Committee recognized the problems with breaking
time series, but nonetheless chose to start with a "fresh slate." The Committee concluded a
July 1993 issues paper by stating--
Reclassifying existing data
Ideally, given the need to familiarize data users with the new classification system, and given
the need to start building the new generation of time series, it would have been desirable to apply
the new classification criteria to existing data, such as the results of the 1992 Economic
Census. This is possible where SICs have direct counterparts, or where census planners
identified kind-of-business categories in 1992 that anticipated the new classifications, such
as warehouse clubs, convenience stores, and refreshment places.
Unfortunately, most of the new classifications require information that we simply don't have
from 1992 census questionnaires: Was an office supply store set up more like a retail store
or a wholesale business? Did a retail bakery offer table service? Was a bank location
involved in commercial banking or credit card issuing?
Role of the 1997 Economic Census
The Economic Census has a unique role in supporting a change in the industry classification
system. Only the census gathers the comprehensive detail that allows for conclusive
determination of industry classification for all establishments. The timing for the entire
NAICS development process was largely predicated on the need to make decisions in time
for implementation in the 1997 Economic Census.
The 475 variants of the
1997 Economic Census questionnaire have been designed to
classify each establishment according to both SIC and NAICS. In some cases that means
more questions or categories to identify the type of store or types of services or commodities
produced or sold. Designing these questions is critical, more important to the
operational definition of these industries than any external documentation.
In 1987, collecting data in this way allowed the publication of basic summaries of 1987 data
according to both the old, 1972-based SICs and the new, 1987-based SICs in the earliest
final publications from each census, even though the rest of the 1987 census data were
classified exclusively according to the new system. Data were presented in two ways:
Comparative statistics tables
"Comparative statistics" tables present new data according to the old system, along
with data from one or more earlier censuses based on the same system for
comparison. In 1987, comparative statistics tables were shown for old, 1972-based
SICs at the state level for retail trade, wholesale trade, and services;
and at the national level for manufacturing and mining.
Table 6 illustrates a 1987 comparative statistics table, where each row is defined by
the smallest kind-of-business unit identifiable from the previous census. Since they are
constrained by the information available from the previous census,
comparative statistics tables ignore any new detail introduced in the more
recent revision.
Table 6. Comparative Statistics Based on SIC
The Committee recognizes . . . that it is unproductive to collect and maintain time series data
that have questionable value. Thus, it may be preferable to accept a one-time break in
historical continuity if the benefits of conversion to a new classification structure are apparent
and accepted by users.
1972 SIC Code |
1987 SIC Code |
Kind of business or operation |
Establishments | Receipts | |||
1987 | 1982 | 1987 ($1,000) | 1982 ($1,000) | % chg 82-87 | |||
7331 | 7331 | Direct mail advertising services | 3512 | 2145 | 4097739 | 1742417 | 135.2 |
7332 | 7334 pt. | Blueprinting and photocopying services | 3877 | 3233 | 1785951 | 856651 | 108.5 |
7333 | 7335,6 | Commercial photography, art, and graphics | 13826 | 10833 | 5243998 | 3095499 | 69.4 |
7339 | 7334 pt, 7338 |
Stenographic services; and reproduction services, n.e.c. | 6017 | 4920 | 1279613 | 795943 | 60.8 |
Bridge tables
"Bridge" tables take the interrelationships of the old and new classifications one step farther.
They present new data cross tabulated by both old and new classification systems
at the same time, identifying the lowest common denominators between the two
systems. In 1987, bridge tables were produced at the national level only for
manufacturing, mining, and construction.
Tables 7 and 8 illustrate 1987 bridge tables, one from the perspective of the new
classification, one from the perspective of the old. Each row is defined by the
lowest common denominator between the old and new systems.
Table 7. Bridge Table: (Distribution of 1987 SIC-Based Industries Among 1972 SIC-Based Industries)
Industry |
Estab- lish- ments |
Employees (1000) |
Payroll (million$) |
Value of shipment (million$) |
New 3571, Electronic computers | 974 | 151.9 | 4953.0 | 33626.5 |
Old 3573, Electronic Computing Equipment (pt.) | 969 | 151.5 | 4945.2 | 33591.9 |
Old 3662, Radio and Television Commun. Eqp. (pt.) | 4 | (D) | (D) | (D) |
All other | 1 | (D) | (D) | (D) |
New 3572, Computer storage devices | 106 | 43.3 | 1442.6 | 6394.8 |
Old 3573, Electronic Computing Equipment (pt.) | 106 | 43.3 | 1442.6 | 6394.8 |
New 3575, Computer terminals | 121 | 15.0 | 441.7 | 1799.0 |
Old 3573, Electronic Computing Equipment (pt.) | 115 | (D) | (D) | (D) |
Old 3661, Telephone and Telegraph Apparatus (pt.) | 6 | (D) | (D) | (D) |
New 3577, Computer peripheral equipment, n.e.c. | 549 | 76.2 | 2625.4 | 13965.5 |
Old 3573, Electronic Computing Equipment (pt.) | 549 | 76.2 | 2625.4 | 13965.5 |
Table 8. Bridge Table: (Distribution of 1972 SIC-Based Industries Among 1987 SIC-Based Industries)
Industry |
Estab- lish- ments |
Employees (1000) |
Payroll (million$) |
Value of shipment (million$) |
Old 3573, Electronic Computing Equipment | 1852 | 310.7 | 10119.4 | 59195.4 |
New 3571, Electronic computers (pt.) | 969 | 151.5 | 4945.2 | 33591.9 |
New 3572, Computer storage devices | 106 | 43.3 | 1442.6 | 6394.8 |
New 3575, Computer terminals (pt.) | 115 | (D) | (D) | (D) |
New 3577, Computer peripheral equipment, n.e.c. | 549 | 76.2 | 2625.4 | 13965.5 |
New 3661, Telephone and telegraph apparatus (pt.) | 39 | (D) | (D) | (D) |
New 3695, Magnetic & optical recording media (pt.) | 67 | 15.5 | 401.4 | 1755.4 |
Nonmanufacturing | 7 | 2.1 | 61.0 | 240.2 |
The fact that the comparative and bridge tables came out in the earliest final publications
was very helpful to users in both getting the trend data (based on the old system) early, and
in introducing the new classifications. Nonetheless, when further review resulted in
reclassification for some establishments, the bridge tables already published for
manufacturing and mining were no longer definitive.
Plans for 1997
Here are highlights of Census Bureau plans for publishing 1997 Economic Census data:
1. In early 1999, in the first "advance" report covering the entire economy, we will
publish 1997 national data on a NAICS basis (3-digit subsector detail), and
"comparative statistics" for 1997 and 1992 on a 1987 SIC basis (2-digit detail).
At the state level, advance data will be limited to totals for the 18 NAICS sectors
and comparative statistics for the 8 SIC divisions covered in the 1997
Economic Census (excluding agriculture and public administration, largely
covered in the independent censuses of agriculture and governments).
2. Roughly a year later, in early 2000, we will publish detailed bridge tables covering
all industries at the national level, and comparative statistics for all industries at
both state and national levels.
3. All other data will be published only on a new, NAICS basis. Thus, the first
industry-specific numbers, to be published in 1999, will not support comparisons to
earlier years for any industries that changed in definition. Outside of the limited
comparative statistics tables, historical data will appear only for those industries
that remained unmodified or can be converted cleanly into NAICS.
As in previous censuses following changes to the SIC, no comparability tables will be
available for metropolitan areas, counties or cities. All substate data from the 1997 census
will be published on a NAICS basis only.
The absence of 1997 data published on an SIC basis below the state level, coupled with the
magnitude of the changes in sector definitions, means that the popular
Economic Census Profiles can no longer include
historical data at the sector level, illustrated below,
that have been available on CD-ROMs from the last two censuses.
Table 9. Historical Data Displayed by Profile Software (not possible for 1997)
|
Limitations: Disclosure-Avoidance and Resources
The reader may well question why the plans to publish historically comparable data on an
SIC basis are so modest. After all, the 1997 census is being designed to allow the
classification of data according to both old and new system, and we know that users will
want as much historically comparable data as possible.
The Census Bureau protects the confidentiality of the information businesses provide by
avoiding the publication of any data that could be associated with any particular
establishment or company. While nearly all industries have been defined in such a way that
there are enough companies present to support tabulation at the national level, many of the
SIC-by-NAICS combinations may be so small that they are dominated by individual
companies. That will preclude the publication of some parts of the national bridge tables, as
illustrated by the (D)s in Table 7 above, as well as comparative statistics for some industries--or even sectors--for some states.
Disclosure analysis in any one Economic Census table is generally a straightforward process.
Disclosure analysis across multiple, non-nested classification systems can become
exceedingly complex, and the Census Bureau has not been able to reduce this complex analysis entirely to
computerized algorithms. It is the large amount of analyst time required to undertake
disclosure analysis, more than anything else, that limits the Bureau's ability to publish the 1997
Economic Census according to both the NAICS and SIC systems. Tight budgets in recent
years have not allowed the agency to fully maintain, much less build up, the kind of staff necessary to
take on this additional work, and tight budgets in future years could reduce further our ability
to complete this extremely important work in a timely manner.
Another impact of budgetary limitations is that the coverage of the 1997 Economic Census
will not be complete for several of the new NAICS sectors. That is because several activities
in agriculture services, forestry, and fisheries--which have not been included in previous
economic censuses--are moving to other sectors. For example, veterinary services and
landscape design are moving to the Professional, Scientific and Technical Services sector.
Summaries for that sector in the Economic Census will be incomplete until funding is
available to cover those firms not previously included.
Given that the Census Bureau will not be able to fully publish 1997 Economic Census data according to both
systems, some users might have preferred that SIC rather than NAICS be given the priority for
publication. Doing so would have given users the full comparable detail for now, provide for
enough NAICS data to get us used to the new concepts, and allow a gradual transition
toward more complete implementation of NAICS in 2002. That alternative was rejected
because the Bureau generally recognizes that NAICS is a better system than SIC, it includes many
new industry categories that many users are eager to have, and both the Bureau and
data users are going to have to
make the transition some time. In essence, let's get on with it.
User Reaction
All in all, the initial public reaction to the introduction of NAICS has been positive. Business
leaders and trade associations recognize that the new industry classification system will be
better than the SIC and more relevant to today's economy.
Those involved in studying the international economy are excited to gain new tools for
examining production and sales across Canada, Mexico and the U.S. The statistics may not
make audible the "giant sucking sound" Ross Perot predicted, but analysts will now be able
to make better inferences about the long-term effects of NAFTA.
A number of trade associations and companies associated with new industries are especially
eager to see statistics on the NAICS basis. The American Society of Interior Design is an
example of a group not particularly concerned with historical comparability in 1997. Interior
designers now have their own industry in NAICS, whereas under the SIC they were a tiny part of the large
category "business services not elsewhere classified."
Perhaps the group least well served in the changeover to NAICS are users of substate data. The
1997 Economic Census will publish some historically comparable data at the state level, but
people interested in smaller geographic areas will be left without answers to rather basic
questions like--
Alternatives for local data users
Data users will have a few alternatives.
1997 County Business Patterns (CBP). The Census Bureau creates CBP data largely by taking the results of the
5-year census and updating them on an establishment-by-establishment basis with
a survey of large multiunit companies coupled with employment and payroll
information for all companies from the quarterly payroll tax records reported to the
Social Security Administration. CBP does not pick up any new classification
information from the census until the nominal year after the census year, so the
1997 CBP will still be reported by SIC, and will not convert to NAICS until data year
1998 (to be published in 2000). CBP reports the number of establishments,
employment and payroll, but includes no measures of sales or output. And industry
classification in CBP that is based on updates to old information cannot be
as accurate as industry classification resulting from direct observation in an
economic census.
Synthetic estimates. Synthetic
estimation is a fancy term for what many users do rather routinely: apply
proportions or trends derived from one data set to a related data set or subclass.
Users could apply state trends published by SIC-based industry group to the local
level, or could prorate NAICS-based data for their locality into SICs using national
bridge tables. Those users may or may not be aware of how unreliable their results
could be.
Special tabulations. A proposal has been made that the
Census Bureau publish 1997 totals by SIC division, e.g., manufacturing and
retailing, for metro
areas, counties and places. Doing so would allow the publication of the economic
profiles illustrated in table 9, even though some of the results would have
to be suppressed to avoid disclosing information about particular companies.
Unfortunately, these tabulations will require more resources than the
economic census program expects to have in 1999 and 2000, and it may be
that local area information by SIC can be produced only on an ad hoc basis with
users paying the cost of the special tabulations.
Effects on Other Statistical Programs
The introduction of NAICS will, ultimately, have a far-reaching impact on the entire U.S.
statistical system. In particular, statistical agencies face many challenges in the conversion
to NAICS for the Census Economic Current Surveys and for national accounts statistics
produced by the Bureau of Economic Analysis, the Federal Reserve Board, and the Bureau
of Labor Statistics.
Consistent with past practices, the introduction of this new classification system scheme
begins with the quinquennial sEconomic Census. Other programs like the Annual Survey of
Manufactures, County Business Patterns, the Economic Current Surveys, and national
accounts statistics will convert to the new scheme in the several years that follow. Unlike
past SIC revisions, though, the change from an SIC to a NAICS basis is so large that it
requires that several agencies consider new approaches for converting their programs to
NAICS.
Examples of new approaches being considered by various agencies include--
Conclusions
In the long run, the conversion of economic statistics from the SIC system to NAICS will be
seen as an important step toward providing a strong foundation for statistical information in
coming decades. Nonetheless, the data user community faces a major
upheaval--losing some of their ability to look at consistent statistics
over time. Hopefully, statistical agencies and others
will undertake analytical and
educational activities to help us all through this transition.
There will doubtless be many expressions of dismay among users as they become aware of
the breaks in time series associated with the conversion to NAICS. The essential question
now is how users can be helped to prepare to deal with the situation,
given that there is still some time before the first NAICS-based data appear
in early 1999.
A NAICS Implementation Schedule lists
target publication dates for the first NAICS-based data from each of over 30
different surveys and other statistical products from Census, BEA and BLS.
The schedule, as well as a wide variety of tables, documents, and papers, are
accessible from the NAICS web site:
http://www.census.gov/epcd/www/naics.html.
Revised: 2/23/98