ments. A company operating establishments in more than one county is generally required to submit a report for each location; also, companies engaged in different lines of activity at the same general location are required to submit separate reports if their operations are large and distinct in character. Thus, pulp mills were separated from paper mills; and coke ovens and blast furnaces from steel mills. (See "General Explanations, " 1947 Census of Manufactures, Volumes I, II, or III, Section 5.)

b. Definition of Industries:

An industry consists of a group of establishments primarily engaged in the same line or similar lines of economic activity. In the manufacturing field, the line of activity is generally defined in terms of the products made, although in some cases the determining factor is the process of manufacture used. (See "General Explanations, 1947 Census of Manufactures, Volume I, Section 10, or Volumes II or III, Section 9.)

c. Scope of the Surveys:

The annual surveys, as did the 1947 Census, include establishments primarily engaged in manufacturing as defined in the Standard Industrial 'Classification Manual, Volume I, Manufacturing Industries, dated November 1945. The Standard Industrial Classification Manual describes manufacturing establishments as those engaged in the mechanical or chemical transformation of substances into new products. These activities are usually carried on in plants, factories or mills, which in the United States characteristically use power-driven machines and materials-handling equipment.

Manufacturing production is usually carried on for the wholesale market, for inter-plant transfer, or to the order of industrial users, rather than for direct sale to the household consumer. Some manufacturers in a few industries sell chiefly at retail to domestic consumers through the mail, through house-to-house routes, or through salesmen. Some activities of a service nature (china decorating and engraving, etc.) when performed primarily for the trade are included in the manufacturing division; when performed primarily to the order of the household consumer, they are considered nonmanufacturing.

The scope of the annual surveys coincided with the Standard Industrial Classification System, with the following minor exceptions:

(1) Manufacturing establishments with no employees were omitted from the surveys (except for certain jobbers as described in (2) below).

no effect on the statistics except those for number of establishments. (See "General Explanations, 1947 Census of Manufactures, Volume I, Section 9, or Volumes II or III, Section 8.)

(2) Textile, apparel, and leather jobbers who assign materials to contract factories for fabrication were covered in the annual survey although they are not classified as manufacturers by the Standard Industrial Classification. These establishments were included mainly to obtain more complete data on cost of materials consumed, value of products shipped, inventories, and value added by manufacture for the industries to which they are assigned. Their inclusion has virtually no effect, however, on the figures for employment, earnings, and man-hours.

The annual surveys covered the following types of establishments which were omitted from the 1947 Census: (1) Logging camps and logging contractors and (2) Sawmills that produce less than 200,000 board feet of lumber a year.

3. Design and Selection of the Annual Survey Sample

Nearly 45,000 manufacturing establishments were selected for the 1949 and 1950 Annual Survey samples. They were drawn from the 1947 Census of Manufactures directory, supplemented by (1) a 1947 Bureau of Old Age and Survivors Insurance list of companies classified in manufacturing by that agency but which could not be matched readily to the Census list, and (2) a BOASI list of manufacturers that had begun operating after 1947. Small Eastern sawmills, which are not well covered by the available lists, were represented by an area sample. These several sources together included more than 300, 000 establishments.

The sample design emphasized industry group (3-digit) estimates for 1949, and important individual (4-digit) industries for 1950. All large companies (single-unit as well as multi-unit) were included in the sample for both years, however, regardless of industry.

All "independent" establishments (i.e., single-unit companies) with 250 or more employees, all multi-unit companies with any establishments of 250 or more employees, and all other multiunit companies with aggregate employment of 1, 000 or more in the 1947 Census were selected for the sample. Smaller companies were sampled at rates that decreased with decreasing size in terms of their 1947 employment, the specific pattern varying by industry group for 1949, and by individual industry for 1950. A similar pattern was used in sampling from the two supplementary

Companies rather than individual establishments were used as sampling units, and those selected for the sample were required to file separate reports for each of their establishments. This requirement avoided potential confusion as to which manufacturing establishments of a company should be reported; made it possible to detect reporting errors, where interplant transfers within a company were important; provided a convenient method for representing new units of existing companies in the sample; and greatly simplified the task of coordinating the Census reports with BOASI reports.

(For a detailed explanation of the sampling technique, as well as of the methods used in drawing the sample from the BOASI list and the sawmill area sample, see Appendix A.)

4. Canvassing Methods Used in the Surveys

The report forms for both the 1949 and 1950 Annual Survey were distributed and, for the most part, returned by mail. An important exception to this procedure was made in the case of the small, highly mobile Eastern sawmills which were canvassed almost entirely by field enumeration.

The mailing, collection, screening for completeness and consistency, and follow-up of reports for all independent and small multi-unit companies were handled entirely by the various Field Offices of the Census Bureau. For the large multi-unit companies, these operations were conducted directly by the Census Bureau in Washington. To facilitate screening of the reports, the field offices were provided with a series of editing checks patterned after those used by respondents. (See Appendix D, Form MA-100 Item 12.) Those manufacturers failing to send in reports and those whose reports were revealed by these checks to be incomplete or otherwise inadequate were reached by letter, telephone, or (in some few instances) by personal visit.

5. The Annual Survey Establishment Report Forms

One standard report form, MA-100, was sent to all establishments included in the annual surveys, except sawmills and planing mills (See Appendix D). The latter plants were canvassed on Form MA-110 which called for information on production and stocks of rough lumber, in addition to the items included on the standard form.

a. Comparison with the 1947 Census Forms:

The 1949 and 1950 report forms represented, for the most part, merely an abridgement of the standard schedule used in the 1947 Census of Manufactures. The basic statistical measures

pay rolls, value of shipments, cost of materials, etc. were defined in essentially the same way for

each of the 3 years. Consequently, historical series published in this report may be considered to be comparable, except as specifically noted in the tables. Three 1947 Census inquiries--analysis of pay roll ended nearest October 15, electric energy generated and sold, and highway-type motor vehicles in use--were dropped entirely, and one new inquiry--products bought and resold without further manufacture--was added to the annual survey forms. Other important changes made in the report form were: (a) the provision for reporting value of shipments in terms of approximately 1,000 product classes, instead of some 6,600 individual products, as was the case in the 1947 census; (b) the substitution of product class reference lists, on the basis of which respondents entered on the report forms (in addition to the value of shipments figures) descriptions of the product classes and their codes, in place of preprinted product descriptions and codes appearing on previous census forms; and (c) the addition of a check list of questions at the end of the form to enable respondents to examine their figures for completeness and accuracy.

b. "Self-Editing" Checks for Completness and Consistency

A check-list of five questions was added to the annual survey report forms, by means of which the respondent could edit his own figures for completeness and consistency. Although no comprehensive statistical study has as yet been made of the effectiveness of this technique, there is reason to believe that it has resulted in an improvement in the quality of reporting. At the same time, it has spared the Census Bureau the expense of numerous letters, telephone calls, and field visits to manufacturers involved in the editing and followup of defective reports.

c. Product Classes and Product Class Reference List.

The limitation of the Annual Survey budget (approximately one-twentieth as large as that for a complete census) required the substitution of a single basic product inquiry for the more than 200 different product forms, each containing a preprinted list of products, used in the 1947 Census. This drastic simplification of the product inquiry made it desirable to use a reference list, containing the principal classes of products made in each of the major segments of manufacturing activity, to assist respondents in reporting their shipments of products. The product class reference list also was used as a device to achieve self-coding" of product information by respondents, as they were asked to assign appropriate 5-digit product class codes shown on the reference list to their products, as well

The Annual Survey of Manufactures product classes are generally comparable to those published for the 1947 Census. In 1947, the approximately 6, 600 individual products for which statistics were collected were grouped into nearly 1, 000 product classes. Both the 1947 Census product descriptions and to a lesser extent, the annual survey product classes were reviewed and approved by a number of interagency committees set up within the Federal Government. In addition, trade association officials and company representatives from nearly all industries had actively participated in assuring that the commodities and classes as defined were realistic in terms of business practices and records.

d. Limitations in Reference List Approach

One problem involved in this approach was the failure to define each product class in terms of all of its constituent products. For some classes, this made it necessary for respondents to establish their own definitions in order to report a breakdown of their value of shipments. As a result, there were a large number of instances in which plants believed to be making the same products in 1949 and 1950 as in 1947 reported substantial shipments of a given product class in 1 year but no shipments of the same class in the other 2 years. For the product classes shown in this report, the majority of these differences were resolved by determining whether they were due either to reporting errors (and correcting the report if they were) or to actual shifts in production between 1947 on the one hand and 1949 and 1950 on the other.

Data have been withheld for any product classes for which (1) the standard error of the estimates exceeded 15 percent, (2) the estimates, regardless of sampling errors, were considered incorrect as indicated by a lack of consistency with related data, or (3) the estimates could not be appraised adequately and checked in the time alloted to such work.

It should also be noted that the statistics for about 70 of the product classes were drawn from the Bureau's Facts for Industry Series (which are based on 63 monthly, quarterly, and annual commodity surveys) or from commodity surveys of other Federal agencies in which the coverage was more complete than in the annual surveys. 6. Classification of Establishments into Industries a. Method of Industry Classification:

An establishment is assigned to, or classified in, an industry, generally on the basis of the principal products made. The products made

the industries to which they "belong, and the group of products accounting for the principal portion of the total value of shipments of the establishment determines its industry classification. This group of products is said to be the primary products of the establishments as well as of the industry which it defines; all other products made by establishments classified in the industry are referred to as secondary products. (For a description of each of the manufacturing industries see Appendix C, "Industry Descriptions," 1947 Census of Manufactures, Volume I, II, or III.)

While some establishments produce only the primary products of the industry in which they are classified, this is rarely true of all the establishments in an industry. The general statistics (employment, pay rolls, value added by manufacture, etc.) shown for an industry, therefore, reflect not only the primary activities of the establishments in that industry, but also their activities in the manufacture of secondary products and, in some instances, their auxiliary nonmanufacturing activities as well.

In some of the borderline areas between manufacturing and other major divisions of the classification system, such as retail trade, wholesale trade, or construction, products are made to a significant extent by establishments primarily engaged in nonmanufacturing activities. Among these products are: venetian blinds, awnings, millwork, dried fruit, ice, prepared feed, poultry dressing, house furnishings, canvas products, and dairy products. Since no reports were obtained from establishments primarily engaged in nonmanufacturing activities, national totals for these products as published in table 1, Chapter III, are understated to the extent of such nonmanufacturing production.

b. Historical Shifts in Classification:

Difficulties arise arise in the assignment of industry classification to establishments wherever their major activities change from one period to the next. For example, if in 1947, the value of an establishment's output consisted of 51 percent creamery butter and 49 percent natural cheese, it was classified in the Creamery butter industry. However, if these percentages are reversed in 1949 or 1950, application of the principal activity rule would require shifting of 100 percent of the statistical data for the establishment in the current year to the Natural cheese industry because of a slight change in its activity. If the establishment is large, the statistics for the Creamery butter industry might show a marked, but largely artificial, decline and the statistics for the Natural cheese industry a corresponding increase as a result of this transfer of the establishment's data

In order to minimize such distortions from one report period to the next and at the same time compile current industry data, some modification was made for the annual surveys in the general rule of classifying establishments on the basis of principal products reported in the current period. Small establishments (generally defined as those with fewer than 100 employees) were retained in the industry in which they were classified in the 1947 Census. Large establishments (generally those with 100 or more employees) that showed a change in their major activity from the previous period were also retained in their 1947 classifications unless the shift in major activity was significant. ''Significant" changes were defined by a rule that tested the effect of changing the industry code on both historical comparability and current industry levels of activity. This rule gave equal importance or "weight" to the effect of each such establishment on the estimates of current year industry levels and on the industry trends between years. Where changing the code minimized this joint effect, the current code was assigned; where the 1947 code minimized this joint effect, the 1947 code was retained. This procedure resulted in greater comparability of 1947, 1949, and 1950 Census Bureau general statistics.

In 1949, approximately 1,000 "large" establishments reported primary activities different from 1947, but in only two-thirds of these cases were the changes determined to be "significant". The industry classifications for these establishments were changed accordingly in 1949 and the reported data tabulated in the appropriate current industries. The remaining one-third of these establishments were retained in their 1947 Census industries. In 1950, about 800 establishments reported major activity changes from 1949 (or from 1947, if the establishments were not included in the 1949 Annual Survey). In about half of these cases the changes were sufficient to warrant changing the 1947 industry code.

7. Coordination of Census-BOASI Employment Data a. Differences in 1947 employment totals:

In the 1947 Census of Manufactures, published United States totals of the number of manufacturing establishments and of employment were significantly lower than the corresponding 1947 Old-Age and Survivors Insurance program totals, or those of other Government series. Much of the difference could be accounted for by (1) differences in scope (i.e., the Census figures did not include data for central administrative offices and auxiliary establishments of manufacturing concerns, logging camps, small sawmills, small "custom" producers, and certain other groups of establishments), (2) undercoverage in the Census, esti

ufacturing employment, and (3) classification of some establishments as manufacturing by BOASI and as as nonmanufacturing by Census because of differences in types of activities reported to the two agencies. In addition, some of the differen ce could be ascribed to the fact that BOASI employment totals (actually an approximate figure for many establishments) covered the mid-March pay roll, while Census employment represented an average of 12 monthly figures. There were, however, apparent discrepancies between the two employment series at the State and county levels, as well as by industry classification, which could not be accounted for by these differences of scope or reporting period.

Both Bureaus agreed that there was need to coordinate their statistical programs in order to provide the public with comparable and consistent employment series. It was believed, too, that if such coordination could be accomplished while at the same time preserving the confidential nature of the data supplied by employers to the two Bureaus, an important step would have been taken towards the goal of a more unified and improved Federal statistical program. The first coordination efforts by the two Bureaus were directed primarily toward resolving differences in 1947 industry codes assigned to single-unit establishments. In addition, comparison of the 1947 statistical data for individual multi-unit concerns showed that the two Bureaus often received different establishment reports from some of the largest multi-unit companies.

b. Coordination in the 1949 Annual Survey:

It was realized that the achievement of a high degree of coordination in manufacturing statistics would require a case-by-case comparison of the total number of employees reported and of the industry and geographic codes assigned. Such a project was undertaken in connection with the 1949 Annual Survey of Manufactures and involved about 30, 000 of the largest establishments covered in this survey.

Employees of the Bureau of Old-Age and Survivors Insurance assigned to this work were sworn in as Census Special Agents, and thus became subject to the Census laws protecting the confidentiality of reported data. Similarly, to retain the confidentiality of the OASI records, selected Census employees were authorized to work on the OASI file. These persons compared the information reported to the two agencies for most companies included in the annual survey sample. Individual annual survey reports were matched to OASI records developed from the Social Security tax returns. This matching opera

industry codes and location codes, and significant differences in employment. (For a detailed description of this matching operation, see "Explanation of Data," Section IV, First Quarter 1949 County Business Patterns, published jointly by the Bureau of the Census and the Bureau of Old- Age and Survivors Insurance.)

c. Coordination in the 1950 Annual Survey:

In the 1950 Annual Survey, a combination matching procedure and notification-of-code- change system was again established by the two Bureaus whereby coordination was maintained in the cases already covered in 1947 and 1949 and was extended to the few hundred important "unresolved" 1949 cases, as well as those cases that became problems for the first time in 1950. (A detailed description of the coordination work done during the 1950 Survey will be included in the First Quarter 1950 County Business Patterns publication to be released in 1952.)

d. Results of Coordination

Tables 2 and 3 of Chapter I, "Summary Statistics," in this publication summarize the results of such coordination by relating the 1949 and 1950 Annual Survey employment estimates with totals compiled from 1949 reports under the OASI program. As the tabulation of 1950 OASI data has not yet been completed, only an approximated 1950 total manufacturing employment figure (based on the 1950 Census Bureau employment estimates for operating manufacturing establishments and the 1949 OASI figures on administrative and auxiliary employment at separate establishments) can be shown in these two tables. It is unlikely, however, that changes in central administrative office and auxiliary employment between 1949 and 1950 were large enough to seriously affect the total manufacturing employment figure.

As the comparative data in these tables show, some differences still remain in the series produced by the two Bureaus, but these have been reduced considerably since 1947 and may be attributed in part to (1) sampling variation; (2) differences in scope--sometimes legal and sometimes administrative in nature; and (3) the fact that coordination was not complete at the time of publication. An additional source of difference between the published Census and OASI data may have arisen from the use of a "derived" BOASI average monthly employment figure for comparison with the Census data which represent an average of the 4 months: March, May, August, and November. Since OASI compiles only mid-March employment data, they were adjusted to comparable average employment totals by use of seasonal ratios computed from Bureau of Labor Statistics monthly employment estimates for these same

8. Employment and Pay Roll Data Collected in the Annual Surveys

a. Coverage

As described in a previous section, annual surveys covered operating manufacturing establishments only, and omitted separate administrative offices and auxiliaries of operating establishments. Operating manufacturing establishments, however, account for approximately 98 percent of total manufacturing employment, and separate central administrative offices and auxiliaries, for about 2 percent. Both of these latter groups, it was noted above, are covered under the OASI program. Because the Annual Survey of Manufactures employment figures for operating establishments have been closely coordinated with the Old-Age and Survivors Insurance program records for accuracy and consistency in classification and coverage, it is possible to relate the annual survey employment data for operating establishments to other published total manufacturing employment figures.

b. Definition of "All Employees":

"All employees" comprise all full-time and part-time persons on the pay rolls of operating manufacturing establishments who worked or received compensation for any part of the four selected pay periods covered during 1949 and 1950, including persons on paid sick leave, paid holidays and paid vacations. It is equivalent to the sum of the three employee classes for which statistics were collected in the 1947 Census, 1947 Census, namely, "Production and related workers, "Forceaccount construction workers, and Administrative, sales, supervisory, technical, office and all other personnel. "

c. Definition of "Production and Related Workers":

"Production and related workers" comprise working foremen and all non-supervisory workers (including leadmen and trainees) engaged in fabricating, handling, packing, warehousing, shipping, maintenance, repair, janitorial and watchman services, product development, auxiliary production for plant's own use (e. g., power plant), record keeping, and other services closely associated with these production operations of the establishment.

d. Definition of "All Other Employees':

"All other employees' comprise nonproduction personnel of operating manufacturing establishments, including those engaged in the following activities: executive, purchasing, finance,