The location is the production location where, at a minimum, employment data are available. The establishment is the most homogeneous unit of production which maintains accounting records from which data elements can be drawn from the operating structure of the business. An establishment can consist of several locations (a dry cleaner with several common sale counters but with only one accounting), or there could be several establishments in a single location (a hotel with a bar or a restaurant within a gas station, for example). The company is the entity for which income, expenditure and balance sheet accounts are maintained, where operating profits and the rate of return on capital may be calculated. A company can count several establishments. The enterprise is the structure which oversees everything, being the unit which directs and controls the allocation of resources in all its components, and for which consolidated financial and balance sheet statements are held, making it possible to evaluate the entity's general financial situation. It is also the institutional unit for the System of National Accounts.

Evidently, this division is useful to distinguish the manufacturing unit in a complex organisation and these four dividing levels are identical in a small business.

Without considering legal or financial aspects, the NAICS will classify establishments whether they are related to small or very large businesses. It will also classify two establishments with similar production processes together.

WHAT IS THE REASON FOR THIS TRANSITION FROM THE SIC TO THE NAICS?

Data resulting from classifications allow for a historical description of a phenomenon which can be counted or measured. These data are organised in time series with continuity as its main virtue by preserving the existing relationship between the observations. In addition to the change in the classification system, this continuity concept also reacts to significant structural changes in the economy, in the data collecting method of surveys or even in questionnaires.

Minimal changes in the classification methods threaten continuity of the time series. This type of change generates enormous costs for statistical agencies who, on one hand, must produce these classifications and manage change, and for establishments, on the other hand, which must adapt to new survey procedures. Users must also adapt to the new concepts and change the parameters of their models which were established according to the previous classification.

When questioning the relevance of the change towards the NAICS, it must be remembered that the SIC was abandoned for two major reasons instead of being revised for a fourth time. First of all, regarding the actualisation of the classification structure, the three SIC revisions added new sectors without basically changing the classification structure which would now be over 50 years old. The economic activity changed as well as the structure of the SIC making it impossible to be displayed adequately.

Secondly, the SIC industry grouping method by establishments was ambiguous. Indeed, the SIC gathered establishments by using two different criteria. Users were unaware of which criteria was used for their grouping: Did a common product link these establishments? Was this grouping by establishment responsible for the majority of the production of this certain good or service?

The several changes brought on by the NAICS and the concepts which followed its development proved beneficial for the following reasons:

• Similarity of the production function, which is a unique criteria, prevailed within the NAICS for the grouping of industrial class establishments, conforming with the objective aimed during the production of industrial statistics.
• Particular consideration is given to service related industries, new industries or emerging industries, in order to increase the similarity between the classification structure and the economic activity.
• The continuity of the time series is an objective to be reached following the priority to untie itself from a classification which produces low analytical relevance series. Basket item categories, such as Other Services to Businesses Not Elsewhere Classified (category 7799 of the 1980 SIC), provide an example of significant weakness, because it is impossible to base an analysis on an industry named in a category where the grouping does not have any economic direction.
• So, on one hand, the NAICS supports international comparisons between NAFTA countries in response to the inherent need to follow and compare National Industries and, on the other hand, supports other countries while adapting to the new standard of international comparison, the International Standard Industrial Classification of the United Nations (ISIC revision 3).

THE CODING STRUCTURE OF THE NAICS

The hierarchical structure of the NAICS, or tree structure, is represented in the following table. The lowest level, entitled National Industries, carries a six-digit code. By dissecting this code, it is possible to find, for each level of the hierarchy, to which group belongs the National Industry.

In order to compare, the three countries using the NAICS worked out a common portion. To satisfy their different needs, the countries accepted to establish a more detailed version, the National Edition, which was obtained by adding elements to the common structure. For the examples below, codes up to their fifth level, known as Industry Codes, belong to the common-core structure, whereas codes with six levels, known as National Industries, belong to the Canadian National Edition, without being in the American Edition.

The main point is for the sixth digit to allow each country to develop its own National Industry, therefore recognising the need for each one to have distinct tools to display different economies, and this, for geographical as well as cultural or economic considerations.

VARIOUS NAICS

Starting from the subsector level, the NAICS Canada has a more developed tree structure than the common-core, entitled NAICS, when referring to the table below relating to the structure comparisons of the various NAICS. It is therefore said that the point stated above may contain some exceptions.

These exceptions relate to the sectors of construction, wholesale trade, retail trade, and public administration for which the common-core ends as of the second level. For other groupings, it is on the third or fourth level that the common-core ends. This situation sets the comparability limit between the three countries at the end of the common-core and not at the fifth level of the NAICS code.

During the next NAICS revision in 2002, the sectors of construction and wholesale trade will be the development subject all the way up to the Industry level.

When a group is present in the NAICS, the hierarchy level to which it belongs corresponds to the industrial orientation of the three countries. Forestry, for example, which ranked at the top of the Canadian Classification hierarchy (1980 SIC), ranked at the subsector level in the NAICS, since there is very little importance given to this activity in Mexico.

At the bottom of the hierarchy, if one of the three countries should break a formed group, then the three countries will adopt a minimal classification necessary for this country, no matter the relevance to the other countries. Therefore, in order to satisfy the needs for the Mexican NAICS, the Canadian version has an industry and a National Industry on "Tortilla Manufacturing" (311830). Similar compromises have been made for the United States making the Canadian Edition of the NAICS contain National Industries related to "Orange Groves" (111310) or "Cotton Farming" (111920), although these two products are by no means cultivated in Canada.

It is understood that in the development of various NAICS Canada groupings, the comparability requirements are more important than each country's faithful representation of economic reality. However, when a group belongs only to the Canadian Edition, it must then satisfy certain criteria regarding its importance in the Canadian economy.

Moreover, in order to increase the comparability of their statistical production, Canada and the United States developed their respective National Edition by supporting resemblance, either by granting the same definitions to the same codes or by avoiding¹ that the same National Industry Code be used in two different ways. Therefore, by reading the NAICS Canada handbook, it is possible to distinguish the groups which form the International Edition of the NAICS, the groups which are developed along with the United States and, finally, the groups which belong to Canada only.

THE DIFFERENCES BETWEEN THE NAICS AND THE SIC (1980)

The major difference lies in industry formation, based on the similarity of the production process. As opposed to the SIC, the demand aspects do not constitute a grouping formation pole. The SIC includes, in the wholesale trade sector for example, establishments dedicated only to Sold Equipment Maintenance Services. In the NAICS, what is not from the production process leading to wholesale trade is unfortunately dissociated from the industry and rather tossed in subsector 811, entitled Repair and Maintenance.

Usually, SIC industries can be found several times in the NAICS, but there are also new industries which display revisions of the classification system or its organisation change. Even though an industry can be found in both classification systems, its grouping level could differ. Therefore, accounting offices and research laboratories, both users of an input named Skilled Labour, are now grouped together.

Rearranging the border between the retail trade and wholesale trade is another major difference because different groups are formed. Whereas, in the SIC, these activities belonged to either one of them and was determined by the type of customers served, the NAICS grouping was done according to the production mode which is defined here by marketing elements.

The following table makes it possible to notice that the NAICS proposes a structure oriented towards new and emergent sectors and displays a non comprehensive list of the changes which occurred in various sectors.

• In addition to the presentation of a new aggregation of its activities, the sector of Agriculture, Forestry, Fishing and Hunting stands out by isolating, in this production sector, agricultural support activities. Therefore, veterinary laboratories, whose production mode differs from that of breeders are no longer confused in agricultural production.

• Similarly, in the Utilities sector, electric power distributors and electric power generators are now in distinct industries.

• The former division entitled Manufacturing Industries is now entitled Manufacturing, therefore illustrating that it is now production-oriented. The Publishing Industries are now excluded from Manufacturing to be included in sector 51, entitled Information and Cultural Industries. There is a new group for data-processing and electronic product manufacturing (it should be noted that the United States NAICS classification has much more information on this industry).

• The new boundary between the two types of trade, the Retail Trade and Wholesale Trade, based on the similarity of the means used to sell and not on the supposed customers, makes it possible to adapt classification to current reality. Therefore, an industry will belong to the retail trade rather than to the wholesale trade if they are single sales rather than sold in great quantities, if advertisement is directed towards widely distributed media rather than towards specialised works, if goods are distributed in stores rather than in catalogues, and if the operation site is localised in order to obtain passing customers or to help truck traffic. With the former classification, establishments whose activities consisted of building material or computer sales were associated to wholesale trade. In the NAICS, the retail trade takes part in these types of trades.

• A new group, entitled Information and Cultural Industries, allows the NAICS to group establishments which have production, transformation or distribution of informational and mass cultural products in common.

• Professional, Scientific and Technical Services group industries whose production processes strongly use aptitude and knowledge, thus forming 32 new industries.

• Activities related to daily operation support of other organisations are joined together in the Administrative and Support Services. It is about a new activity group which can be associated, among others, to registered office activities.

• Changes brought in the Educational Service Sectors shows contemporary economic structure adaptation, by adding an industry relating to non-institutional training and by associating places whose vocation is connected to learning and teaching, such as museums and libraries.

• Accommodation and Food Services are adapted to modern times with Resorts, Bed and Breakfast, Limited-Service Eating Places and Food Service Contractors.

SIMILARITIES BETWEEN BOTH CLASSIFICATION SYSTEMS

Classification changes cause interruptions in the time series. However, trend establishment, a typical activity of economic analysts, requires time series with a certain background. It is therefore necessary to reconcile the data resulting from former classifications with new data. This section aims at displaying data reconciliation modes, bonding approaches between past and present time series.

The first approach can only be practised with micro data held by the statistical agency. Statistics from new groups must then be compiled once more, starting by recording responses from all respondents for every period. Easier alternatives were developed.

It is therefore possible to proceed to interpolation recording according to the method adopted by Statistics Canada for the Labour Force Survey. Then proceeding by double coding the years for the beginning and the end of an existing period, in view of 1987 and 1997, and charging the respondents for 1988 to 1996 similar values to double coded years. It is then possible to produce new statistics from new groups. This method requires a certain amount of information on the respondents in order to be able to double code the limiting years of the sample, but also to conclude for the years inside the perimeters.

When their is less information available, it is possible to double code only one period, often the last year, and to make up a coefficient matrix of conversion in order to establish an empirical form of concordance between two classifications. This coefficient matrix, calculated over one year, is then applied to the previous periods. It is therefore an easier method because it is sensitive to the choice of the base year. The obtained results should be used carefully. This method was used for the 1990-1996 background data of the annual manufacturers survey. The publication entitled Labour Force Update, vol. 3, no. 2, 1999 offers such a matrix for employment in Canada.

However, after the first publication of the results according to a classification, the statistics are blocked from the respondent by the legal confidentiality agreement. Users without access to confidential data of the survey must resort to the theoretical concordance tables. These tables are proposed by Statistics Canada and act as a plotter, making it possible to follow activities between groups of two classifications. The use of these tables display a significant challenge: being able to compose with complex relations between two classifications, complex because they only deal with part of a category, and not the whole. The table below displays the types of concordance relations which exist.

Statistics Canada discourages the use of these tables for time series connection because this tool alone is unable to compose with complex concordance relations.

THE TIMING OF THE NAICS

The arrival of the NAICS represents a significant change which allowed data users to work from a classification system that is more coherent and more representative of the economic activity's actual state. It should also be noted that, by coincidence, certain improvements in no way related to the NAICS were brought to the survey, giving even more amplitude to this change period. With this said, the passage from the SIC to the NAICS is going well and will reach its peak in 2003, according to the Statistics Canada calendar which also disseminates other information on this subject such as the classification and its concordance tables, the establishment program by survey, as well as various comparisons with the former classification.

Data drawn from annual sectional surveys, manufacturing sector surveys for example, are very advanced in their passage to the NAICS. In fact, by the end of 2000, all annual sector-based surveys, besides some service surveys, will be classified according to the NAICS. The passage is done regarding monthly sector-based surveys, except for wholesale and retail trade surveys which will become part of the NAICS in 2003.

Concerning general surveys on all sectors, such as the Labour Force Survey and those on Transportation and Business Finance, the passage is finished, except for the Employment, Earnings and Working Hours Survey, whose passage is planned for the beginning of 2001, and for Provincial Gross Domestic Product by Industry, whose passage is planned for the end of 2001.

CONCLUSION

The NAICS was created to provide common definition keys of economic activity structures for the three countries who are members of NAFTA. In order for each country to satisfy the need for a comprehensive classification and to have a common application, it was decided to work out a common-core from which each country would add new elements in order to form its National Edition.

The Canadian National Edition of the NAICS will gradually replace the previous classification, the Standard Industrial Classification (SIC), revised in 1980. From now until 2003, the NAICS Canada will be the only establishment classification standard for surveys on production sectors. The end of the SIC marked a turning point since, after having experienced some revisions, existed for over 50 years.

The NAICS is different from the SIC according to the following observations. Initially, the NAICS used only one method of industry grouping. Then, wholesale and retail trade were divided in an adequate way. Finally, it should be noted that the NAICS offered a higher comparability than the SIC. Indeed, in addition to the comparability with NAFTA countries, it is possible to establish some form of correspondence with the United Nations Classification (ISIC revision 3).

The breaking of continuity inherent to the change of classification system is a major cost for this operation. Existing surveys often propose background material, gathered according to methods adapted to the availability of financial and informational resources. Some of these background materials must therefore be carefully consulted. Users also have balance tables between the NAICS and the 1980 SIC, but the use of these tables are challenging.

Beyond all these developments, a revision is planned every five years guaranteeing greater adaptation capacity. Already expected for the next revision in 2002, the International Edition of the NAICS is already expecting an upgrade of several levels of details in the construction and wholesale trade sectors presently lying at the minimal level of comparability: the sectors.

__________________
1. There is however, three exceptional cases (323119, 332999, 334512) for no codes were available in the National Industry of the American Edition.