1961 and subsequent months are based on the new weights. (See table D-3, pp. 236-237 of this issue.)

Effect of the Weight Change

The effect of the weight revision on the relative importance of major groups, subgroups, and product classes of commodities in the index is shown in table 1. The relative importance of an item for any specified period is its basic value weight (1958 value in this case) adjusted for the percentage change in price from the weight date to the specified date, with the result expressed as a percentage of the total adjusted value for all commodities. Changes in relative importance result from (1) weight revisions such as the one just completed, (2) minor interim weight adjustments to account for the addition or deletion of items, and (3) different rates of price change among various commodities. Thus, relative importance figures are distinguishable from weights because they change from month to month, whereas weights, by definition, are fixed for long periods. However, relative importances are in fact the weights implicit in the month-to-month changes in the index numbers."

The relative importance figures in the first column of table 1 are the 1958 weight values adjusted for price changes between the reference date (1958) and the month of introduction (December 1960) and thus represent the effective weights of the groups in the WPI from December 1960 forward. The data in the second column of the table show, for the same date, the relative importance of the various groups of commodities in December 1960 under the 1954 weighting structure. Since both 1954 and 1958 data have been adjusted for price movements to December 1960, differences in the two sets of figures reflect only changes in weight assignments and in the sample of items priced.

Comparison of the relative importance figures for December 1957 and December 1960 based on 1954 weights shows the effect of price change (and minor sample revisions) on the relative importance of items. For example, between December 1957 and December 1960, when the all commodities index advanced 0.8 percent, eight of the major com

modity group indexes advanced more in price and therefore gained in relative importance.

In general, the results of the recent weight revision represent the effect of updating the index to take account of changes in production and marketing patterns between 1954 and 1958. Thus the increase in relative importance of the fuels group from 7.65 to 7.87 percent is due principally to the large increase in production of gas and electricity which took place between 1954 and 1958. In other groups, the weight revision reflects not only shifting market patterns, but also improvements in coverage of the data used in developing the weights and changes in imputation patterns. Some of the more important reasons for the changes in relative importance in December 1960 are as follows:

Farm Products. The larger increase in farm output than in industrial production between 1954 and 1958 accounts for almost all of the rise in the relative importance of the farm products group from 10.16 to 10.59 percent of the total WPI. The remainder is due to addition to the 1958 weights of the value of baby chicks produced in commercial hatcheries.

Processed Foods. Weights in the processed foods group were increased by the addition of processed poultry and frozen fruits and vegetables produced in establishments not classified in manufacturing industries.

Textile Products and Apparel. The increase in relative importance of textile products was due largely to the inclusion of the value of converters' shipments of finished goods in the weight structure for the first time.

Chemicals and Allied Products. Between 1954 and 1958, production of chemicals and related products increased 29.8 percent compared with 8.9 percent for total industrial production. Increases in the value of shipments ranged from 35 percent to more than 50 percent for commodities such as soaps and synthetic detergents, photographic

See formula (2), footnote 6.

supplies, plastic materials, pharmaceuticals, and toilet preparations. As a result, the relative importance of the chemicals and allied products group has increased from 5.78 to 6.64 percent.

Pulp, Paper, and Allied Products. The small drop in the relative importance of the pulp, paper, and allied products group which resulted from the introduction of 1958 weights was due almost entirely to further exclusions of interplant transfers and direct sales to consumers by the printing and publishing industry.

Metals and Metal Products. The introduction of the 1958 weights for metals and metal products reduced the relative importance of that group, mainly because production of iron and steel decreased between 1954 and 1958. In addition, production of fabricated metal products rose considerably less than total industrial production.

Machinery and Motive Products. The machinery group experienced the greatest change in weight, falling from 19.65 to 17.57 percent of all commodities. A large part of the decrease occurred because, between 1954 and 1958, production of machinery and related products increased relatively less than total industrial output-only 2.5 percent compared with 8.9 percent. Many items such as machine tools, fans and blowers, and motors and generators actually decreased in volume. Production of nonelectrical machinery dropped 4.2 percent. About one-eighth of the total decrease in the weight assigned to this group resulted from a change in imputation patterns. A sizable value for the instruments and related products industry, formerly assigned to the machinery group as a whole, was assigned elsewhere, mainly to the miscellaneous products group. Also, the weight structure now excludes, for the first time, the values of rebuilt machinery and of railroad cars made in railroad car shops-the former being considered as repair work and the latter, as captive production and thus an interplant transfer.

Miscellaneous Products. The introduction of the 1958 weights brought an increase in the relative importance of miscellaneous products, since the production of such items as toys, sporting goods, and phonograph records had risen at a higher rate

than industrial output as a whole and because of the assignment here of values of some instruments and related products. In addition, the weigh for manufactured animal feeds was increased br the inclusion of prepared animal feeds produced in nonmanufacturing establishments.

Other Commodity Groups. Changes in relative importance in the remaining major commodity groups were less than 0.5 percentage points and arose mainly from shifts in production and marketing patterns between 1954 and 1958. As mentioned previously, large increases i gas and electricity sales raised the weight for the fuels and related products and power group. Declining output of lumber and various household appliances reduced the weight for the lumber and wood products and for the furniture and other household durables groups. Higher production of cement, clay products, and concrete and plaster products resulted in greater relative importance for the nonmetallic mineral products group. In the hides, skins, leather, and leather products group, a less-than-average increase in volume of leather and leather products caused a decline despite the addition to the weights of raw furs and of sheepskins sold for pulled wool. The drop in weight in the rubber and rubber products group resulted from a decline in the value of imports and from the exclusion of interplant transfers for crude natural rubber.

Revision of Commodity and Reporter Sample

In January 1961, 290 commodities were added to the WPI sample and 78 items were dropped. In addition, 554 individual reporter series were added, 538 from company reporters and 16 from trade publications. Nearly 100 new reporters strengthened reports for existing items; the rest were used to construct new commodity series.

Increases in commodity coverage in three majori groups-chemicals and allied products, metals and metal products, and machinery and motive products-account for over 80 percent of the! new items (and nearly all of the net increase). These groups, plus furniture and other household) durables, also gained nearly 80 percent of the' additional individual reporter series.

In the chemicals group, the principal increase in coverage occurred among pharmaceutical prep

arations. The sample of commodities in this grouping was completely reappraised, and the index was subdivided by end use. This will facilitate comparison with price trends at retail as shown in the Consumer Price Index.

In the machinery group, the expansion of coverage provides a new grouping for special industry machinery and equipment, with indexes for food products machinery, textile machinery and equipment, woodworking machinery, and printing trades machinery and equipment. In addition, in October 1960, seven types of imported and domestic compact passenger automobiles were introduced because of their increasing importance in the United States domestic car market.

In the metals and metal products group, items were added to the nonferrous and the fabricated metal products subgroups and a new index was introduced for lighting fixtures.

Other improvements include strengthening of the furniture and other household durables group with the addition of over 50 new company reporter series and introduction of an index for porch and lawn furniture. Moreover, cotton broadwoven goods in the textile products group were regrouped to show finished goods and grey goods separately.

periodic weight revisions. (The base period has also been changed several times, but these changes have not affected the weighting structure of the index and hence are not discussed here.)

Calculation. From 1921 until the major weight revision in 1952, articles properly falling under more than one classification were included under each of the classifications. Thus, structural steel, nails, and certain other metal products used in building were placed in both the building materials and the metals and metal products groups. Foods produced on the farm which reach the consumer practically unchanged, such as potatoes, eggs, and milk, were included among both farm products and foods. In computing the all-commodities index, however, such articles were counted only once. This system, unlike today's, made it impossible for the users of the WPI to construct their own special groupings by using detailed price indexes and relative importance figures. Also, the original WPI, which covered the years 1890-99, was compiled as an unweighted. index of price relatives, but the number of series used for each commodity was roughly proportional to its importance in the market. This system was eliminated in 1914, when the index was recomputed retroactively with 1909 weights.

Another change in calculation methods was made in the major revision of 1952, when value weights were substituted for quantity estimates but without changing the basic formula. A chaining procedure was adopted whereby monthto-month price relatives are weighted with values, rather than the absolute prices weighted by physical quantities. In addition, once-a-month pricing replaced the use of arithmetic averages of 4 or 5 weeks' prices each month, after comparisons revealed little difference in movement.

Scope. In the two most recent revisions, attention was also given to the definition of the universe. In order to avoid an overlap of the CPI and WPI universes, the concept was clarified to exclude primary market sales directly to consumers. As indicated, the present weight universe is the value of commercial shipments of all goods produced by manufacturing, mining, agricultural, forestry, and fishing industries, in the United States, as well as commercial imports. It includes shipments originating in the United States, whether for domestic

or foreign use; imports for resale in U.S. primary markets; sales of gas and electricity to industrial and commercial users, including sales of electricity by Federal, State, and local governments; sales of scrap and waste materials for industrial use; and sales to the Federal Government of types of goods which are also sold to civilian customers, such as food and office supplies. Also included are sales through factory-owned retail outlets where data are lacking to permit the separation of these sales from other company shipments.

Excluded from the weight universe are the values of interplant transfers; goods produced and consumed within the same establishment; sales to the Federal Government of specialized products, such as military aircraft and ships; sales of goods by Federal, State, and local governments, except sales of electricity; direct sales from producers to household consumers (e.g., sales of electricity to residential users, and bread and milk sold on retail delivery routes); all services, whether transportation, distribution, personal, business, etc.,8 including services performed by establishments primarily engaged in producing goods (e.g., repair work and processing, on a commission basis, of goods owned by others); and such individual items as works of art or race horses.

Weight Revisions. New weights have been introduced into the WPI numerous times since the U.S. Department of Labor first published an index of wholesale prices covering the years 1890-99. Often, the new weights were applied retroactively and the index for earlier years was revised. Three principal retroactive revisions were as follows:

1. In 1921, the index incorporated weights from the 1919 Census of Manufactures. Computations were carried back to 1890 and 1909 weights (introduced in an earlier retroactive revision) were replaced.

2. In 1927, a decision was made to change weights each 2 years (in that period a Census of Manufactures was collected biennially). The index was revised back to 1913.

3. In 1952, there was a comprehensive revision of weights, sample, and methodology and the WPI was recalculated back to 1947. Value weights, based on the 1947 industrial censuses, replaced quantity weights; and the present system of imputation was introduced, wherein

each commodity series in the index represents a class of items whose prices are presumed to move similarly.

Weights have been changed three times since 1952. In January 1955, adjustments were made to aline relative importance of major commodity groups with 1952-53 values. Value weights from the 1954 and 1958 industrial censuses were introduced in January 1958 and January 1961.

Changes in the relative importance of major groups, 1890-1960, and in the number of commodities are shown in table 2. These figures refer to the WPI as it is now published from 1890 to date and do not show the effect of those weight changes which were later replaced by the retroactive revisions described above.

Relative importance data are shown before and after each weight change. For each set of weights, the relative importance for the first time period shown reflects change brought about by the introduction of the new weights; the relative importance for the other time period reflects the effect of price change from the introduction of new weights to the specified revision date as well as minor changes due to introducing or dropping items.

Weights in the WPI reflect in broad outline changes which have taken place in the marketing economy of the United States over the years. The decreasing importance of farm products is portrayed strikingly. The relative importance of this group fell from 29.04 percent of the total index in 1890 to 13.67 in 1933, then rose to 20.83 in 1947, as farm prices rose at a rate greater than industrial prices during intervening years. From 1947 to 1960, the newly defined farm products group again fell in relative importance. It can be observed that, although farm products gained relative importance several times owing to greater than average price increases between weight revisions, only two revisions increased their weights in comparison with those at the time of the preceding revision.

Direct sales of other commodities may be included in the weight universe if separate data for commercial shipments are not available.

Electricity and gas, although deemed services for some purposes, are included in the WPI because they are considered to be functionally similar to and competitive with fuels such as coal and oil.

This early index was an unweighted average of 99 price relatives. This index represented an updating and revision by Roland P. Falkner of the University of Pennsylvania, of indexes compiled in accordance with a U.S. Senate resolution of 1891 which led to the publication of data on wholesale prices for 1840-91 in the so-called "Aldrich Reports" of the Senate Committee on Finance.