The Statistics Corner: The NAICS Is Coming.
Will We Be Ready?

By Maurine A. Haver

The year 2000 will offer more challenges to the business economist than the simple conversion of software to deal with the turn of the century and a year divisible by four that is not a leap year. It also marks the start of the conversion of our monthly economic indicators to the North American Industrial Classification System (NAICS).

Many macro forecasters have dismissed this change based on a belief that it will only impact those involved in industry analysis. They are wrong. If current plans are implemented, by 2003 consistent data on aggregate measures such as nonfarm payroll employment in manufacturing will be reported retroactive to 1995 at the earliest. Familiar categories, such as construction and manufacturing, will still be found on statistical releases, but the establishments they represent will have changed.

ADVANTAGES OF NAICS

We might ask if the benefits of NAICS are worth the costs of adjusting to new definitions and truncated and/or inconsistent time series. It seems our answer has been "Yes." The decision to move to NAICS has been made over a number of years with requests for public comment each step along the way. And business economists have supported the program in theory each step of the way. Adoption of NAICS has several advantages:

1. It is based on a single conceptual framework- producing units that use similar production processes are grouped together. The Standard Industrial Classification (SIC) had both production-oriented and demand-based categories.

2. Provision is made for new and emerging industries. It not only recognizes many new industries that were not recognized explicitly in the old SIC, but it also provides a framework in which new industries can be more easily introduced in the future.

3. The system was developed jointly with Canada and Mexico, so there will be comparability among the data of the three countries up to the five-digit level. NAICS is also compatible with the two-digit level of the International Standard Industrial Classification of All Economic Activities of the United Nations (ISIC, rev. 3).

NAICS DEVELOPMENT

The development of NAICS began in Williamsburg, Virginia, in 1991 at a conference on the international classification of economic activities. Many NABE members participated in this meeting. Several of them criticized the SIC system's haphazard structure and limited coverage of services and high-tech industries. At this meeting, NABE past president Joe Duncan called for "a clean slate" in the classification of industries. Shortly thereafter, the Office of Management and Budget (OMB) appointed an Economic Classification Policy Committee (ECPC) to work with Canada and Mexico in the development of a system for classifying establishments by type of economic activity. This committee was chaired by Jack Triplett, Bureau of Economic Analysis, with members from the Census Bureau, the Bureau of Labor Statistics and the Department of Commerce.

ECPC working papers and additional information on NAICS can be found at www.census.gov/epcd/www/naics.html. A document with the full NAICS specification as well as a cross-reference table between the SIC and NAICS are also available.

SIC/NAICS COMPARISON

The SIC system we use today was first developed in the 1930s when the United States was primarily a manufacturing economy. It has been revised periodically to incorporate services and new industries, but the basic structure has remained unchanged. The most recent revision was in 1987. It has ten divisions with 1004 four-digit industries.

NAICS is organized in a hierarchical structure like the SIC. In coding NAICS, the number of divisions was expanded to twenty and divisions were renamed sectors. Industries are represented by five- or six-digit codes. (Five digits represent a NAICS industry. A sixth digit is added to designate U.S. detailed national industries.)

NAICS currently recognizes 1174 industries as shown below.

The manufacturing sector was reorganized to improve international comparability and to recognize new high-tech industries. One new subsector in manufacturing will be Computer and Electronic Product Manufacturing. Computers, communications equipment, audio and video equipment and semiconductor manufacturing fall into this subsector along with the reproduction of packaged software, which will move from services into manufacturing. Software development will remain in services.

Publishing except for printing will move out of manufacturing into a new sector called Information. This new sector groups together industries that primarily create and disseminate products with intellectual property content. Of the thirty-four industries in the Information sector, twenty are new.

The 1987 SIC Services division has been divided into eight new NAICS sectors. At Williamsburg, Courtney Slater commented that "there seems to be no consistent definition or classification concept underlying decisions on what to include in services or on how to arrange the categories" in the SIC. She went on to conclude that "the resulting conglomerate is too diverse to be analytically useful." The ECPC has listened. We will soon have a well-organized framework for services industries. But will our analysis not now be plagued by lack of consistent history?

Finally, it is important to note that NAICS classifies auxiliary establishments- units that exist to perform support services not intended for sale outside the enterprise- according to their primary activity rather than according to the classification of the establishment they serve, as was done with the SIC. For example, the accounting unit of an automotive manufacturing establishment will be classified under accounting in NAICS, not under automotive manufacturing, as was done previously. This may lead to a further decline in manufacturing employment with a commensurate gain in services employment in industries such as accounting and data processing.

NAICS TRANSITION PLAN

The federal statistical agencies are committed to providing some history and bridge tables to help data users move to the new classification scheme. Data from the 1997 economic censuses will be reported in 1999 on both an SIC and NAICS basis. Census will code auxiliary establishments by primary activity and by the industry the establishments serve to help provide a link between 1992 and 1997 data.

When the Economic Census data are ready for use in benchmarking, the transition to the NAICS for monthly indicators at Census will begin. The 1998 Annual Survey of Manufacturers will be undertaken only on a NAICS basis, and there are no plans at present to report monthly data on both an SIC and NAICS basis. BLS will begin a three-year implementation plan in 2000 with the first payroll employment data reported on a NAICS basis with the benchmark revision of 2002 data in 2003. Changes to BLS price programs are expected to begin in 2004. As a consequence, it is likely that there will be a period of time during which Census reports industry data according to NAICS, while BLS continues to report employment and prices according to SICs.

WHY THE DELAY?

The burden placed on the statistical agencies is enormous. Each establishment in their files must be reclassified, and auxiliary units must be separated and reclassified. Unlike Statistics Canada, where this will be done once, in the United States each agency will replicate the work of the others. (At least we hope that the assignments of establishments to NAICS codes will be consistent from agency to agency.) Even if data sharing legislation were passed today, it is unlikely that the workload could be reduced materially if the existing timetable remains unchanged. BLS faces the task of working with the states to get state systems converted as well.

In addition to coding establishments, historical data need to be constructed. Current budget plans provide for construction of history sufficient to calculate seasonal factors. Over one-third of all SICs will be split into two or more NAICS codes. Of the so-called "directs" (those that map directly from an SIC into a NAICS), many actually must be combined with parts of other SICs to form the NAICS code. For example, even with something as simple as corn (which is called a direct), SIC code 0115 (corn) and part of SIC 0119 (cash grains, nec) form the new NAICS code 11115 called corn farming. Therefore, to calculate history for NAICS code 11115, it is necessary to allocate a portion of SIC 0119 and add it to corn. Likewise, to construct SIC figures from data collected under NAICS, one needs to allocate corn to two separate SICs.

WHAT DOES THIS MEAN FOR BUSINESS ECONOMISTS?

At a minimum, we have lots of uncertainty ahead concerning the comparability of our data both over time and across data groups. If schedules are maintained, it appears that Census data will be reported on a NAICS basis well ahead of BLS. Economists at BEA who use data from both agencies may find themselves in the same predicament as most users _ construction, shipments, inventories and orders data will be on the NAICS basis while employment and prices will continue to be reported according to the old SIC structure. (What is this going to do to productivity!)

And what about history? Statistical agency representatives fear that they may not get funding for the development of any historical data. Will we work with a number of different versions of our history, each built independently by different consulting firms? If it is too costly for government with the access they have to micro data, what will be the cost to the private sector? How reliable will these private estimates be?

Business economists began this process with complaints about the limitations of the SIC. Unfortunately, most economists probably did not anticipate that our history would be decimated. One of the principles adopted by the ECPC was that "time series continuity be maintained to the extent possible." This principle sounds good but can it really mean only three to five years of history?

The time for NABE members to address these issues has come. The changeover has begun with the 1997 economic censuses. There is no turning back. A major change in the underpinings of our statistical system is beginning even at a time when statistical agency budgets have been barely adequate to maintain comprehensive, high-quality national statistics.

The NABE Statistics Committee has moved NAICS to the top of our list of concerns, and we need your support. That means enlisting your company's management in a campaign to protect the data used by your company. It means letters to the agencies, to OMB, to Congress and to anyone involved in economic statistics and statistical agency budgets. The agencies who produce "estimates" are best suited to the task of rebuilding our history, and they need the funds to do it. Building a new system for the twenty-first century is important, but it cannot be done at the cost of losing the basic building blocks of our work.

Back to the Statistics Page