professional, and technical activities, sales, sales-delivery (e.g., routemen), advertising, credit, collection, and installation and servicing of own products, routine office function, factory supervision (above the working foremen level), and force-account construction employees.

e. Monthly vs. "Average" Employment Totals:

The annual survey report forms requested total and production-worker employment of each operating manufacturing establishment for the pay. roll period ended nearest the middle of each quarter of 1949 and 1950, except that the middle of March was selected for the first quarter to facilitate the comparison of Census and Bureau of OldAge and Survivors Insurance employment statistics (as described in Section 7 above). An average employment figure was calculated tor both "All employees" and "production workers" from the four reported monthly figures. The 1947 Census averages, on the other hand, were based on 12 monthly figures.

f. Definition of "Wages and Salaries":

"Total wages and salaries for the year, reported for all employees in operating manufacturing establishments, are defined as the gross earnings of such employees, including commissions, dismissal pay, nonproduction bonuses, vacation and sick-leave pay, and compensation in kind, and prior to such deductions as employees' Social Security contributions, withholding taxes, group insurance, union dues, and savings bonds. This definition of wages and salaries, recommended to all Federal statistical agencies by the Bureau of the Budget, is exactly the same as that used in the 1947 Census.

g. Definition of "Production Worker ManHours":

"Man-hours worked" is a comprehensive measure of employment that includes all "Production and related worker" plant hours worked or paid for, except hours paid for vacations, holidays or sick leave, when the employee was not at the plant. Actual overtime hours are include rather than straight-time equivalent hours.

As in the 1947 Census, man-hours were requested in 1949 and 1950 by quarter (either 13 weeks or 3 calendar months), because the majority of manufacturers can report such information most readily on this basis.

9. Material Costs, Inventories, and Shipments Data a. Definition of "Total Cost of Materials, etc."

The term Cost" refers to direct charges actually paid or payable (after discounts and in

tainers and supplies, (2) fuel, and (3) purchased electric energy, actually consumed during the entire year. Manufacturers were asked to include cost of materials or fuels consumed regardless of whether purchased by the individual establishment from other concerns, transferred to it from other establishments of the same company, or withdrawn from inventory during the year. Such cost items were to be included regardless of whether the materials were processed by the reporting establishment or by others for its account (contract work). Manufacturers were also instructed to exclude from this total all materials sold in the same form as purchased, but to include such costs in a separate merchandising" question (Item 9). (See Section 9 f. below.)

Separate entries were required for each of items (1), (2), and (3) above, as well as for cost of work done for the reporting establishment on its own materials by other concerns (contract work). The sum of all of these items was deducted from value of shipments in order to obtain value added by manufacture. It should be noted that the cost

of fuel consumed (including anthracite and bituminous coal, natural and manufactured gas, fuel oil, liquefied petroleum gas, and purchased steam) did not include fuel when used as a material in the manufacturing process--for example, coal in the production of coke. Such costs were included with other materials cost under item (1) above.

b. Definition of "Inventories":

The values of beginning- and end-of-year inventories, respectively, were requested of all operating manufacturing establishments in the surveys, except those in the Printing and publishing industries group. As in the three most recent censuses, annual survey inventories data were collected under two major categories: (a) finished products and (b) materials, supplies, work in process, fuel and other inventories

c. Definition of "Value of Products Shipped":

The amounts under the heading of "value of products shipped", whether shown as industry totals or as class of products totals, represent the received or receivable net selling values, f.o.b. plant, after discounts and allowances, and excludManufacing freight charges and excise taxes. turers were asked to include in shipments all products made by or for each establishment from materials owned by it, whether sold or transferred to other establishments of the same company (i.e. other manufacturing plants, wholesale branches, central warehouses, etc.), and to exclude from shipments goods resold in the same condition as when purchased, such resales being segregated in a separate item (see Section 9f below). Products

by another establishment were included only in the shipments reported by the latter establishment, and the payments received by the former establishment for such work were reported by it in the shipments section of the form as "receipts for contract work.

d. Duplication in Cost of Materials and Value of Products Shipped Totals:

The aggregate of the cost of materials and value of shipments figures for industry groups and. for all manufacturing industries combined includes large amounts of duplication, owing to the use of the products of some industries as materials by other industries. With some important exceptions, such as "Steel works and rolling mills" and "Motor vehicles and parts", this duplication is not significant within individual industries but arises from the addition of totals of related industries representing successive stages in the production of finished manufactured products. Examples are the addition of flour mills to bakeries in the Food and kindred products group, and the addition of blast furnaces to steel mills in the Primary metal industries group. Estimates of the over-all extent of this duplication indicate that the value of manufactured products, exclusive of such duplication--namely, the value of "finished" manufactures--tends to approximate two-thirds of the total value of products reported in the annual surveys. Because of the unknown and varying amounts of duplication contained in cost of materials and value of products shipped, these figures are not shown for industry groups and States. They are also omitted for a small number of individual industries for which the proportion of duplication in value of shipments is believed to exceed 10 percent.

e. Value Added by Manufacture:

"Value added by manufacture" measures the approximate value created in the process of manufacture. It, therefore, provides the most satisfactory census measure of the relative importance of given industries for the United States as a whole or for geographic areas. Value added is calculated by subtracting the cost of materials, supplies, and containers, fuel, purchased electric energy, and contract work from the total value of shipments.

Value added by manufacture should not be confused with "income produced in manufacturing". The latter figure is compiled by the Office of Business Economics, Department of Commerce. The main difference between these two measures arises from the fact that income produced excludes, in addition to cost of materials, such additional costs as depreciation, the labor costs involved in maintenance and repairs, State and local taxes (other than corporate income taxes), advertising,

An analysis of entries in Item 9 (Sales and cost of products bought and resold without further manufacture) of the annual survey form, made for a sub-sample of the 1950 reports, indicates that receipts from merchandising resales by operating manufacturing plants amounted to roughly 2 to 3 percent of their receipts for manufacturing activities. Such information on merchandising transactions was obtained in the annual surveys primarily, however, to make certain that it would not be included in data on shipments of manufactured products, and also to help determine whether establishments engaged in a combination of manufacturing and nonmanufacturing activities properly fall within the scope of a manufactures survey. In those instances in which establishments reported a dollar volume of sales and receipts for nonmanufacturing operations (wholesale, retail, service, and trade) in excess of the value of their manufactured products, the industry classification was. resolved by reference to the 1947 Census of Manufactures reports and the Bureau of Old-Age and Survivors Insurance program records.

The areas in which a combination of manufacturing and nonmanufacturing operations were most frequently reported are as follows: petroleum bulk terminals and petroleum refining; lumber distribution and mill work; poultry dressing and fresh egg distribution; feed milling and distribution of hay, grain, etc.; ice manufacturing and fuel distribution; and meat processing and meat

distribution.

10. Expenditures for New Plant and Equipment

In the annual surveys, as in the 1947 Census, manufacturers were requested to report expenditures made during the specified year for permanent additions and major alterations that were charged to the fixed-asset accounts of the individual establishments operating in the report year and were of the type for which depreciation accounts are ordinarily maintained. Since these surveys are restricted to operating manufacturing establishments, the figures shown in this publication do not include capital expenditures during the specified year for (1) nonmanufacturing establishments of manufacturing companies; (2) plants under construction but not in operation; (3) plants idle throughout the year; (4) plant and equipment

Apart from the reporting and sampling errors inherent in both series, the differences between them arise principally from the following differences in coverage. The OBE-SEC estimates cover all of the above-mentioned types of capital outlays excluded from the census series except those for land and for plant and equipment leased to manufacturers. Also, in contrast to Census, these estimates exclude data for manufacturing establishments of companies engaged primarily in manufacturing activities. The first-named factor above--expenditures for nonmanufacturing establishments of manufacturing companies--is particularly important in the Petroleum refining industry, where company expenditures for developing new wells and for distribution facilities are normally much larger than their capital expenditures at refineries.

11. Metals Consumed

non

In addition to the total cost of materials

in the annual surveys, plants in the metal fabricating industries were also required to report their consumption of five types of metal mill shapes (plate, sheet, strip, rod, bar, wire, etc.) and castings: (1) iron, (2) carbon steel, (3) alloy steel, including stainless, (4) aluminum, and (5) copper and copper-base alloy. The consumption of each of these metals was divided between metals purchased (or received from other establishments of the same company), for which both quantity and cost data were collected, and metals produced and consumed in the same establishment (i.e., "captive" production), for which only quantity figures were obtained. The reported data covered consumption for all purposes i.e., for production of fabricated products; for maintenance, repair, and operating purposes; and for construction and alteration of plant facilities.

The metals were defined to cover only those processed to the mill or foundry level, excluding metals fabricated or otherwise processed beyond this level. Machined castings were considered, for this purpose, as processed beyond the foundry level. The consumption estimates shown in the tables thus represent only the direct use of the specified mill shapes and castings and not the total metal weight of the products of the given industry. For example, the steel consumption shown for the motor vehicles industry consists only of steel directly consumed in the form of rough castings, sheets, bars, wire, etc., and does not include the weight of metal in sub-assemblies, stampings, forgings, wire products, etc., produced in other industries.

Data were not collected on the consumption of forgings, and in order to account for the quantity of steel thus omitted, the consumption of steel (ingot, billet, and bars) by the forgings industry was obtained.

Except for one change, the definitions followed in the 1949 and 1950 Annual Surveys were identical. Whereas in 1949 insulated wire and cable was considered to be a fabricated product for which data were collected on the consumption of bare wire used in its manufacture, in 1950 insulated wire and cable was treated as a mill shape for which consumers were required to report the copper content of such wire and cable.

In making comparisons with statistics on

should be noted that the metals consumed data shown in the tables are limited to the major metalfabricating industries, which fall within the following broad industrial groupings: Fabricated metal products, Machinery (except electrical), Electrical machinery, and Transportation equipment. With the exception of the iron and steel forgings industry (3391), therefore, this report excludes the consumption in those primary metals industries (major industry group 33) also engaged in fabricating metal products, as well as in such other significant metal consuming areas as Miscellaenous manufactures and Furniture and fixtures (major group 25). Similarly, consumption in nonmanufacturing activities, such as construction, transportation, mining, agriculture, and public utilities, and shipments for export are not covered. For each metal shown, it is believed that the estimates of consumption in manufacturing represent over 70 percent of the total consumed in all uses.

12. Estimating Procedures Used in the 1949 and 1950 Annual Surveys

Most of the annual survey estimates were derived by combining the 1947 Census (universe base) totals, which were not subject to any sampling error, with sample estimates of the amount of change from 1947. The amount of change, or estimated difference, was computed by:

(1) Multiplying each sample establishment's 1947 Census and current figures by the reciprocal of the establishment's sampling rate;

(2) Summing the inflated establishment data for each year to summary (publication) levels; and

(3) Calculating the estimated increase or decrease from 1947 by subtracting the 1947 sample estimate from the current estimate for the identical item.

Estimated increases from 1947 were added to the corresponding 1947 Census universe totals, and estimated decreases were subtracted from the corresponding base totals, in deriving the final estimates.1

'The estimated difference may also be computed by subtracting the base period value from the current value individually for each sample establishment, weighting the individual differences, and summing algebrically. This procedure, involving individual differences, was used for the 1949 Annual Survey, in order to facilitate the sampling

Sample establishments that had zero values in either the base or current year also were accounted for in the estimated differences. New companies, for example, contributed zero to the sample estimates for 1947, although their current figures were included in the 1949 and 1950 inflated sample totals; while sample units from the 1947 Census list that went out of business before the survey contributed zero to the current inflated sample totals, although their 1947 figures were included in the sample estimates for that year. Similarly, establishments that were reclassified from one industry to another since 1947 in effect were tabulated as changes to zero for their original industries, and as changes from zero for their new industries. This was also true of establishments that moved from one State to another after 1947.

This difference estimating formula was used, with few exceptions, for all the general statistics estimates, for the product class estimates derived from the Annual Survey sample, and for most of the metals consumed estimates. It was not used for any estimates based on reports from logging or sawmill establishments (industries 2411 and 2421). Less important exceptions included estimates in scattered cells of the size distribution table, and a few of the metals consumed estimates. Neither was the difference formula used for the detailed statistics: expenditures for new plant and equipment, inventories, fuels consumed, and electric energy purchased. In all cases where this formula was not used, the estimates were developed by simply inflating the current figures for the sample establishments and summing the inflated figures.

The standard errors of estimates derived by the difference formula tend to be appreciably smaller than the standard errors of estimates derived by the simpler formula (i. e., based on current data alone). The gain in efficiency realized is strongly affected by the correlation between base year and current year values for identical sample units. Relatively higher costs in relation to the expected benefits led to the choice of the simpler formula where it was used.

Various approximate formulas were used to estimate the standard errors of the estimates, as described in Appendix B. Generally, these approximations tended to overstate the errors. Moreover, the published standard errors were rounded upward in all cases (i.e., to the next highest integer for standard errors under 10 percent; standard errors of 11 to 15 percent were

13. Qualifications of the Published Data

All tables published in this report contain columns showing the sampling variation associated with each estimated total. In making any study of the historical trend of the 1947, 1949, and 1950 figures, it is especially important to take account of the effect of such sampling variation on the apparent increase or decrease shown between years.

The approximate sampling variation of the estimated change between 1947 and 1949 or between 1947 and 1950 is solely a function of the 1949 (or 1950) standard error, and can be obtained by multiplying the 1949 (or 1950) estimate by its percentage standard error. It should be noted that the results of these calculations are expressed as absolute amounts, such as number of employees, dollar values, etc., rather than as percentages. The sampling variation of the estimated change between 1949 and 1950 is more difficult to measure, but it is approximately equal to the larger of these two amounts.1

1A more conservative estimate of the sampling variation of the estimated change between 1949 and 1950, which generally will be an upper bound, can be obtained by multiplying each of the estimates by its percentage standard error, squaring each product, adding the squares, and extracting the square root of the sum of squares.

In addition to such sampling errors, individual figures may be subject to biases arising from procedural errors or response errors which are not reflected in the "standard errors" and which were not detected in the editing of the report forms or in the review of the detailed machine tabulations from which the published tables were prepared. Among such errors are those resulting from improper coverage and sample weighting of establishments (i.e., sample control errors), incorrect reporting of data by respondents, and clerical and machine tabulation errors at the Census Bureau. Estimates which were subject to large uncorrected errors of this type were generally suppressed, along with other suspect estimates, when the final publication totals were systematically reviewed both for reasonableness and for consistency with other Census series and related data. Other specific qualifications of the published data are usually indicated in the appropriate headnotes and footnotes to the tables.

14. Disclosure of Data for Individual Companies

The Bureau of the Census is prohibited by law from publishing any statistics that disclose information reported by individual companies. In the 1949 and 1950 Annual Surveys, the same practices used in the 1947 Census were followed in analyzing the published data for disclosures. (See "General Explanations, " 1947 Census of Manufactures, Volume I, Section 15, or Volumes II or III, Section 14.)