U.S. affiliates of foreign companies: 1987 benchmark survey results.

Howenstine, Ned G.

NONBANK U.S. affiliates of foreign companies continued to expand in 1987, according to preliminary results of BEA!s latest benchmark survey of foreign direct investment in the United States.1 Affiliates' employment, assets, and sales increased, primarily as a result of acquisitions of U.S. companies by foreign investors.

Benchmark surveys are BEA!s most comprehensive surveys-both in terms of companies covered and information gathered. The 1987 survey updates universe estimates derived from other BEA surveys and provides information not available in those surveys. The last benchmark survey covered 1980.

The following are highlights from the 1987 benchmark survey:

* Employment by U.S. afffliates increased 222,000 in 1987, to 3,160,000

The rate of increase- 8 percent-was slightly above the 6- percent average annual rate of increase for 1980-87.

* The share of all-U.S.-business employment accounted for by affiliates was 3.6 percent in 1987, up sightly from 3.5 percent in 1986. In 1980, the share was 2.7 percent.

* Manufacturing accounted for nearly one-half of affiliate employment in 1987 (chart 7). Retail trade (18 percent), wholesale trade (10 percent), and services (9 percent) accounted for the next largest shares. Although large, the share of affiliates in manufacturing was down from 1980, when it was 54 percent. The share of affiliates in services, in contrast, doubled from 4 percent to 8 percent. Shares also increased in retail trade and finance (except banking).

* Affiliates with ultimate beneficial owners (UBO's) in the United Kingdom and Canada accounted for the largest shares of total affiliate employment percent 20 and 19 percent, respectively (chart 9). These shares were significantly larger than the next largest shares, which were the 12-percent share for affiliates with UBO's in West Germany and the 9-percent share for affiliates with UBO's in Japan. Since 1980, shares of affiliates with UBO's in Canada, Japan, and Australia increased, while those of affiliates with UBO's in all other individual countries shown in the chart declined (chart 10).

* Among U.S. regions, almost onehalf of total affiliate employment was

in the Southeast (789,000) and the Mideast (735,000). Among States, California had the most affiliate employees (324,000), followed by New York (300,000), Texas (208,000), and New Jersey (169,000).

* Total assets of U.S. affiliates were $926 billion in 1987, up $88 billion from 1986. As in earlier years, most of the increase resulted from acquisitions of U.S. companies by foreign investors. The 11-percent increase in 1987 was well below the 18-percent average annual rate of increase for 1980- 87. The comparatively slow growth in 1987 partly reflected sales by several foreign owners of interests in nonbank finance affiliates that had large assets. A South African parent sold its minority interest in a particularly large affiliate. Primarily as a result of this sale, total assets of affiliates with LTBO's in South Africa declined over $70 billion.

9 By industry of affiliate, the largest increases in assets in 1987 were in

manufacturing ($26 billion), insurance ($20 billion), and nonbank finance ($19 billion)(tables 2 and 3).

* By country of UBO, affiliates with UBO's in Japan had a particularly large increase in assets-$98 billion (tables 4 and 5). As a result, total assets of these affiliates reached $196 billion at yearend 1987 and were larger than those for any other country. Prior to 1987, Japan ranked third in terms of assets, after the United Kingdom and Canada. The sharp rise in Japanese owned affiliates' assets in 1987 was concentrated in finance (except banking). In that industry, assets increased $82 billion, to $119 billion, largely because of Japanese investors' acquisitions of minority interests in two large U.S. companies.

* The gross book value of U.S. affiliates' property, plant, and equipment (PPE) increased $26 billion, to $346 billion. Of the total, PPE used for manufacturing accounted for $125 billion, commercial property for $91 billion, and all other uses for $130 billion. Affiliates' commercial property grew at a faster average annual rate over the 1980-87 period than their total PPE percent, 18 compared with 15 percent. As a result, the share of total PPE accounted for by commercial property increased from 23 percent in 1980 to 26 percent in 1987. At the same time, the

share of PPE used for manufacturing declined from 39 percent to 36 percent.

* Sales by U.S. affiliates increased $59 billion, to $731 billion. This 9- percent increase equaled the average annual growth rate for 1980-87. Of the 1987 total, sales of goods accounted for $622 billion, sales of services for $91 billion, and investment income for $19 billion. By industry, sales were

largest in wholesale trade and manufacturing. country of UBO, af- ffliates with UBO's in Japan and the United Kingdom had the largest sales.

* Net income of U.S. affiliates increased from $2 billion in 1986 to $10 billion in 1987. A shift from large capital losses to capital gains accounted for a substantial portion of the increase. Net income before capital gains or losses increased from $6 billion to $8 billion. In 1986, capital losses occurred because petroleum affiliates lowered the book value of their oil reserves and because chemicals and machinery manufacturing affiliates lowered the book value of various assets.

* U.S. merchandise exports shipped by affiliates decreased $2 billion in 1987, to $48 billion, and U.S. merchandise imports shipped to affiliates increased $15 billion, to $141 billion. Exports were $4 billion lower than in 1980 and imports were $65 billion higher. In 1987, affiliates in wholesale trade accounted for 61 percent of exports and 75 percent of imports. By country of destination or origin, trade with Japan accounted for over 41 percent of affiliates' exports and for nearly 50 percent of their imports.

* The benchmark survey covered 8,260 fully consolidated U.S. businesses; these businesses represented 21,895 individual U.S. companies.

The data in this article, like the estimates in earlier articles based on data from BEA's annual surveys of foreign direct investment, cover the financial structure and operations of non- bank U.S. affiliates. Data collected in the benchmark survey on banks and on transactions and positions between U.S. affiliates and their foreign parents will be available next year when final results of the benchmark survey are published.

The concepts and definitions underlying the 1987 benchmark survey data are essentially the same as those used for BEA's annual surveys and for the

1980 benchmark survey. For most items, the 1987 benchmark survey data are comparable to estimates derived from the annual surveys for earlier years; the few differences are discussed in the technical note. Changes in the presentation of the data and in

the coverage and methodology of the benchmark survey are also discussed in the technical note.

The remainder of this article first analyzes information from the benchmark survey that either has not been available before or has not been available since the previous benchmark survey in 1980, then reviews changes in affiliate employment, and finally discusses the share of the U.S. economy accounted for by U.S. affiliates in 1987. More detailed results of the benchmark survey and the annual surveys for prior years are available in separate publications (see page 140).

Expansion of Information

Some data items were collected for the first time in the 1987 benchmark survey. Other items were collected in previous benchmark surveys, but not in the annual surveys for nonbenchmark years. This section discusses some of the new information-that is, affiliates' sales of goods and of services, sales of services to U.S. persons and to affiliated or unaffiliated foreigners, and manufacturing employment by State. It also discusses information that has not been available since the 1980 benchmark survey-that is, employment by industry of sales, commercial property of affiliates, and U.S. merchandise trade by destination or origin and by product. Sales of services

The 1987 benchmark survey collected, for the first time, affiliates' sales (or gross operating revenues) disaggregated into goods, services, and investment income and sales of services further disaggregated into those to U.S. persons or to affiliated or unaffiliated foreigners." The data were col-

lected as part of a broader BEA data-improvement effort for services.

In 1987, of U.S. affiliates total sales of $731 billion, goods accounted for $622 billion, services for $91 billion, and investment income for $19 billion (table 6). Of total sales of services, $87 billion, or 96 percent, were to U.S. persons and $4 billion were to foreigners. Most sales of services to foreigners were either to the foreign parent group or to other' foreigners ($1 billion and $2 billion, respectively); sales to foreign affiliates of U.S. affiliates were small.

By industry, over one-third ($33 billion) of affiliates' total sales of services were by affiliates in insurance. Affiliates in "services' and finance (except banking) also had substantial sales of services-$17 billion and $14 billion,

respectively. In services," over onehalf ($9 billion) of the total was by affiliates in business services, particularly advertising.

By country of UBO, sales of services by affiliates with LTBO's in the United Kingdom, at $20 billion, and in Canada, at $18 billion, were largest almost twice those by affiliates with UBO's in Japan, which ranked third. For affiliates with UBO's in the United Kingdom, sales of services were largest in insurance and business services. For those with LTBO's in Canada, sales of services were largest in insurance. For affiliates with UBO's in Japan, they were largest in finance (except banking).

Employment by industry of sales

In most tables by industry in this article, classification is by "industry of affiliate." On this basis, the affiliate's primary industry-that is, the industry that accounts for the largest portion of the affiliate's sales-is determined, and all data are shown in that industry even if the affiliate has activities in secondary industries. For two items-

sales and employment-data from the benchmark survey are also available by "industry of sales." On this basis, sales and the associated employment in secondary industries are shown in those industries rather than in the affiliate's primary industry.6

Table 7 compares employment by industry of sales with employment by

industry of affiliate. Employment is higher by industry of sales than by industry of affiliate in retail trade, services, and "other" industries. In other" industries, most of the difference is in mining and transportation. Employment by industry of sales is significantly lower in petroleum, manufacturing, and wholesale trade, mainly because affiliates classified in these industries have substantial employment in other industries (such as retail trade and services). 7 Although employment was lower by industry of sales than by industry of affiliate for manufacturing as a whole, it was significantly higher in some industries within manufacturing. For example, in motor vehicle manufacturing, employment was 55,000 by industry of sales, but only 34,000 by industry of affiliate. It was higher by industry of sales because several affiliates classified in motor vehicle wholesale trade had motor vehicle manufacturing operations. These affiliates are classified in wholesale trade because most of their sales result from the wholesale distribution of imported cars rather than from sales of cars they manufactured in the United States. When classified by industry of affiliate, the manufacturing employees and the other employees of these affiliates all appear in wholesale trade. When classified by industry of sales, the manufacturing employees appear in manufacturing.

Table 8 shows employment by industry of sales cross-classified by country of UBO. For some countries, the industry distribution of employment in this table differs significantly from that in table 17, which shows employment by industry of affiliate cross-classified by country of UBO. For example, when classified by industry of sales, employment by affiliates with LTBO's in the United Kingdom is significantly lower in manufacturing and significantly higher in retail trade and services. In contrast, employment by af-

filiates with UBO's in Japan is higher in manufacturing and lower in wholesale trade when classified by industry of sales. Manufacturing employment by State

The benchmark survey, for the first time, collected a breakdown of affiliates' manufacturing employment by State." Manufacturing employees in a given State are employees on the payroll of manufacturing plants located in the State, including employees in central administrative offices and auxiliary units that primarily serve these plants.

Table 9 shows affiliates' manufacturing employment by State cross- classified by country of LTBO.' Total manufacturing employment was 1,233,000, of which one-third was accounted for by five States: California (120,000), Pennsylvania (82,000), New York (71,000), North Carolina (69,000), and Ohio (68,000).

By country of LTBO, manufacturing employment was largest for affiliates with UBO's in the United Kingdom (282,000), Canada (193,000), and Germany (164,000).

For each country, employment was concentrated in a few States.

Affiliates with

Japanese UBO's had the most concentrated employment-over one-half of their manufacturing employment was in five States (California, Michigan, Illinois, Ohio, and Tennessee). Affiliates with Canadian LTBO's had the least concentrated employment-one-third of their manufacturing employment was in five States (North Carolina, Pennsylvania, New York, Tennessee, and California).

Commercial property

The benchmark survey indicates that the value of U.S. commercial property owned by U.S. affiliates, a measure of foreign ownership of U.S. real estate, was $91 billion in 1987 (table

10). By State, almost 45 percent of the total was in three States- California ($17 billion), New York ($13 billion), and Texas ($10 billion). Affiliates' commercial property holdings in Texas were twice as large as those in Florida ($5 billion), which ranked next in size.",

By country of UBO, afffliates with UBO's in Canada had by far the largest holdings of commercial property-$23 billion. Affiliates with LTBO's in Japan and the United Kingdom followed, with $13 billion each.

Compared with other affiliates, the commercial property holdings of affiliates with LTBO's in Japan were particularly concentrated; three States (Cal- ifornia, New York, and Hawaii) accounted for over 78 percent of their holdings. By comparison, holdings of affiliates with UBO's in Canada and the United Kingdom were less concentrated; the three largest States California, New York, and Texas, in both cases) accounted for 43 percent and 36 percent, respectively, of these affiliates' holdings.

Merchandise trade by country of destination or origin and by product

In addition to data on U.S. affiliates' merchandise trade by industry of affiliate and country of UBO, the benchmark survey collected data on such trade by country of destination or origin and by product."

In 1987, U.S. affiliates had U.S. merchandise exports of $48 billion and U.S. merchandise imports of $141 billion (table 11). Most imports-69 percent-were goods for resale without further processing.

Affiliates' exports to Japan were over five times as large as those to any other country and accounted for 41 percent of all affiliate exports. Affiliates' imports from Japan were over four times as large as those from any other country and accounted for almost one-half of all affiliate imports. A large portion of this trade was accounted for by Japanese-owned wholesale trade affiliates, which primarily distribute products produced by others. Thus, a significant part of these affiliates' exports consists of products of other U.S. businesses that are purchased by

the affiliate and resold abroad; similarly, a significant part of their imports consists of products for resale in the United States without further processing.

After Japan, the next most important destinations of exports were Canada ($4 billion) and the United Kingdom ($3 billion); for imports, the next most important countries of origin were Germany ($16 billion) and Canada ($8 billion). For most countries, U.S. affiliates' imports from a country significantly exceeded their exports to that country.

By product, U.S. affiliates' exports of food were $10 billion, one-fifth of their total exports. Exports of chemicals and machinery were also large-$8 billion and $7 billion, respectively. For affiliates' imports, by far the largest categories were road vehicles and parts ($48 billion) and machinery ($34 billion).

Changes in Employment

This section discusses changes in affiliate operations in 1987 based on changes in employment. Employment

was chosen because changes in it are not directly affected by inflation and, thus, tend to correspond more closely than other available items to changes in real economic activity.

Employment by affiliates increased

222,000 in 1987, to 3,160,000. Increases were substantial in several industries, notably services (44,000), finance (except banking) (31,000), and, within manufacturing, in office and computing machines (20,000) and rubber products (15,000) (table 12). In each industry, the increase was largely the result of acquisitions of U.S. com-

panies by foreign investors. Acquisitions also added significant numbers of employees in retail trade. However, the overall increase for the industry was small (6,000 employees) because the industry classification of a major affiliate changed. Prior to 1987, the affiliate, which has operations in both retail trade and tobacco manufacturing, was classified in retail trade; in 1987, it sold a substantial part of its retail trade operations and, as a result, its classification shifted to tobacco manufacturing (included in "other" under other manufacturing" in the tables).

A number of other industries had significant shifts in employment because of changes in industry classification. Some of the changes resulted from restructuring of affiliates' operations. In some cases, the affiliates sold or discontinued businesses in industries that had accounted for the major part of their overall operations; in others, they acquired or expanded businesses in industries that had previously accounted for only a minor part of their operations. In a few cases, an affiliate both sold and acquired major businesses. Additional changes in industry classification resulted from BEA's review of the detailed information collected in the benchmark survey on the activities of affiliates.

Within manufacturing, the effects of changes in industry classification were particularly significant. For example, within food manufacturing, the increase in beverages (18,000) and the decrease in "other" (32,000) occurred largely because an affiliate's classification shifted from grain mill products (included in other' food manufacturing in the tables) to beverages. (The changes for the two industries were not completely offsetting because total employment of the affiliate dropped substantially between 1986 and 1987.) Similarly, within electrical machinery manufacturing, the increase in audio, video, and communications equipment and the decrease in electronic components largely reflected a change in classification of a major affiliate. The decrease in electronic components also reflected another affiliate's sale of a major subsidiary that resulted in the affiliate's industry classification changing to instruments. Finally, within chemicals, the increase in other" and the decrease in industrial chemicals partly reflected a shift in classification of a large chemicals affiliate that restructured its operations.

By country of UBO, increases in employment were largest for affiliates with UBO's in Japan (64,000), Germany (51,000), and the Netherlands (27,000).

Employment by affiliates with Japanese UBO's increased mainly in finance (except banking), services, and manufacturing. For affiliates with LTBO's in Germany, increases were mainly in manufacturing and retail trade. For affiliates with UBO's in the Netherlands, the increase was more than accounted for by an increase in retail trade. Among U.S. regions, the largest increase in affiliate employment was in the Mideast (68,000) (table 13). Among States, by far the largest increases were in California and New York and 35,000 32,000, respectively followed by Massachusetts and Pennsylvania, with 16,000 each.

Share of the U.S. Economy This section uses two measures employment and total assets-to discuss the share of the U.S. economy accounted for by U.S. affiliates. First,

the size of U.S. affiliates relative to the U.S. economy, in total and by industry, is discussed in terms of employment. The industry comparisons use data on affiliate employment classified y industry of sales. Industry of sales, rather than industry of affiliate, is used because it corresponds more closely to the by-industry- of-establishment classification that is used for the all-U.S.-business employment data. (Classification by industry of sales, however, is not identical to classification by industry of establishment. See footnote 6.) Second, affili-

ates' shares in manufacturing are discussed in terms of total assets. Assets are classified by industry of enterprise for both U.S. affiliates and all U.S. businesses.

In 1987, U.S. affiliate employment accounted for 3.6 percent of the 86,584,000 employees of all nonbank U.S. businesses (table 14). The affiliate share was up slightly from 1986, when it was 3.5 percent. The increase reflected the strong growth in affiliate employment in 1987, which, as discussed earlier, largely reflected acqui-

sitions of U.S. companies by foreign investors.

By industry, affiliate shares of U.S. employment were highest in mining (8.4 percent) and manufacturing (7.3 percent) and lowest in communication and public utilities and in construction (less than 1 percent in each). Within manufacturing, affiliate shares were highest in petroleum and coal products (39.5 percent) and chemicals (23.5 percent) and lowest in textile products and transportation equipment (3.2 percent in each).13 Within transportation equipment, the affiliate share was 6.5 percent in motor vehicles and less than 1 percent in other."

In petroleum and coal products, the affiliate share of the all-U.S.-business total is significantly overstated because of the different industry classifications used for the affiliate and all- U.S.-business employment data. Afiliate employment in this industry is largely accounted for by integrated petroleum companies that are involved in all phases of the petroleum industry, including the extraction and refining of crude oil and the marketing of gasoline and other petroleum products. In the direct investment data, when classified by industry of sales, all of the sales and employment of the integrated companies in any of these activities are included under petroleum and coal products manufacturing. In contrast, in the all-U.S.-business data, which are classified by industry of establishment, the employment of integrated companies is distributed among the activities of the company's individual establishments;

thus, only employees in the companies' manufacturing establishments are included in petroleum and coal products manufacturing.

A rough adjustment can be made to exclude affiliates' nonmanufacturing employees from this industry using data reported in the benchmark survey. After this adjustment, the affiliate share of all-U.S.-business employment in petroleum and coal products was about 19 percent.14

For manufacturing as a whole, U.S. affiliates' share of total assets was substantially higher than their share for employment-13.2 percent, compared

with 7.3 percent (table 15). Their share of assets was higher for two main reasons. First, affiliates are more concentrated than all U.S. businesses in industries, such as chemicals and petroleum and coal products, that have relatively low employmentto-assets ratios. Second, differences in valuation may cause affiliate shares based on total assets to be overstated. When a company is acquired-whether by foreign or U.S. buyers-its assets are often revalued to reflect the new, generally higher, value implicit in the acquisition price.

Because much of the growth in foreign direct investment in recent years has been through acquisitions, the portion of affiliates' assets that has been recently revalued is probably higher than that for all U.S. businesses. 16

Within manufacturing, the ranking of industries based on the size of the affiliates' shares of total assets was similar, but not identical, to that based on employment. The four industries with the largest affiliate shares based on total assets stone, clay, and glass; chemicals; primary metals; and petroleum and coal products-were also the top four based on employment; however, the order of the four industries was different for the two measures. Also, the two industries with the smallest affiliate shares based on total assets-textile products and transportation equipment-were also the industries with the smallest shares based on employment.

For petroleum and coal products manufacturing, the share based on total assets was significantly lower than

that based on employment. As noted earlier, differences in industry classification caused the affiliate share of employment to be overstated. However, after the adjustment of the employment data discussed earlier, the share based on assets and the share based on employment are very close-19 percent and 17 percent, respectively.

In several industries-particularly stone, clay, and glass; chemicals; primary metals; rubber and plastics products; printing and publishing; and fabricated metal products-the affiliate shares based on total assets were significantly higher than those based on employment. Asset shares were higher partly because of the differences in the valuation discussed earlier. Also, in some of these industries such as stone, clay, and glass- affiliates may be more diversified than other U.S. businesses. In such industries, an enterprise-based classification, like that used for the total assets comparisons, would result in higher affiliate shares than an activity based classification, like the industry-of-sales and industry-of-establishment classifications used for the employment comparisons.

For manufacturing as a whole, the affiliates' share of the total assets of

all U.S. businesses was higher in 1987 than in 1986-13.2 percent, compared with 12.5 percent. The increase reflected a sharp jump in the affiliate share in rubber and plastics products, a jump that, in turn, reflected the acquisition of a U.S. tire manufacturer by foreign investors. Affiliate shares also increased substantially in stone, clay, and glass products; food; instruments; and other' manufacturing.

Technical Note

The 1987 benchmark survey covered all U.S. affiliates of foreign direct investors (U.S. companies owned 10 percent or more by a foreign person) that had assets, sales, or net income of more than $1 million. For similar items, the benchmark survey data in this article are comparable to universe estimates for earlier years, which were derived from data reported in BEA's annual survey of foreign direct investment in the United States. The benchmark survey data indicate that, in general, the totals estimated for earlier years are not significantly overstated or understated.

The consistency of the annual survey estimates and the benchmark survey data reflects improvements in BEA's data collection system and estimation procedures. In particular, a mandatory BEA survey of new investments provides each year the information, especially on smaller investments, needed to update the U.S. affiliate universe.

Except as noted, the concepts and definitions underlying the 1987 data are essentially the same as those underlying BEA's 1980 benchmark survey, as described in Foreign Direct Investment in the United States, 1980. A full methodology of the 1987 survey will accompany the revised data to be published next year.

The preliminary benchmark survey results in this article include estimates of data for reports not received or processed in time for publication and for which BEA had a report for a prior year that could serve as a basis for estimation. The degree of estimation varies from item to item-for example, 9 percent of total assets, but 15 percent of employment, was estimated. Most of the estimation was for small affiliates. The estimation of missing data, which is a departure from the practice in previous benchmark surveys, permitted results to be published in 13 months, about one-half the time required for the last benchmark survey. The final results to be published next year will incorporate data from reports received or processed after publication of these preliminary results. Revisions are expected to be small overall; however, they could be sizable for some individual countries, industries, States, or for merchandise exports and imports by product. Revisions are most likely to be sizable in cells in which small affiliates predominate.

The number of U.S. affiliates included in the data for 1987 is significantly smaller than the number included in the estimates for the past few years. The benchmark survey indicated that a significant number of companies that were below the exemption level of BEA's annual survey, and for which BEA had been making estimates since the 1980 benchmark survey, have been sold, liquidated, or merged or consolidated with another U.S. affiliate since 1980. Thus, they have been eliminated from the number count. Most of these companies were small real estate affiliates. Although the number of U.S. affiliates for 1987 will probably remain smaller than estimated for earlier years, it will be revised up somewhat when final survey results are

published. The revision will reflect the addition of companies whose reports were received or processed too late and could not be estimated for these preliminary results.

The industry detail shown in this article differs from that in the articles presenting the annual survey results for earlier years, both because greater detail is provided and because the industry coding system for direct investment surveys has been changed to align it with the 1987 revised Standard Industrial Classification system. The added detail is mainly in services industries.

The country detail shown has been reorganized along geographic lines; economic or political groupings, such as the European Communities, are no longer shown in the body of the tables (they are shown as addenda in some tables). Also, in this article, unlike in the articles for earlier years, the industry classification of a U.S. affiliate's UBO, if a business enterprise, reflects the LTBO's worldwide consolidated activities rather than just the activities in the UBO's country of classification or the activity of the LTBO itself.

In the 1987 benchmark survey, a long form, requesting information in considerable detail, was filed by affiliates with assets, sales, or net income greater than $20 million. To minimize the burden on survey respondents, a less detailed short form was introduced for filing by smaller affiliates. For these affiliates, BEA has estimated the

items that appear only on the long form so that the published results are

presented in the same detail for all affiliates.