Preliminary results from BEA's latest benchmark survey of U.S. direct investment abroad (USDIA), covering 1994, when viewed with the results from earlier surveys, suggest a high degree of continuity in the patterns of operations of U.S. multinational companies (MNC's). U.S.-MNC operations abroad remain concentrated in a number of large and wealthy economies; the 10 largest high-income economies accounted for just under 70 percent of total production by majority-owned foreign affiliates (MOFA's) of U.S. companies in 1994. Both in these economies and in most other economies that host operations by U.S. MNC'sincluding the rapidly growing economies in the emerging areas of Asia, Latin America, and Eastern Europeproduction by foreign affiliates was predominantly for sale in local markets rather than for export back to the United States. Thus, as in earlier years, the location of overseas production by U.S. MNC's appears to have been determined more by access to markets than by access to low-wage labor or to natural resources. Although foreign production and employment by U.S. MNC's was substantial, their operations remained centered in the United States, where about three-fourths of the worldwide employment and production of U.S. MNC's was located in 1994 (table 1).
Some change was evident in U.S.-parent companies' share of the U.S. economy: They accounted for about one-fourth of U.S. gross domestic product (GDP) in 1994, down from about one-third in 1982 (an earlier benchmark year). The decline largely reflected the concentration of U.S. parents in slower growing sectors of the economy, such as "petroleum extraction and refining" and manufacturing. As might be expected, the share of foreign affiliates in their host economies was much smaller, but it too declined. MOFA's, on average, accounted for 3 percent of GDP in a group of 48 important host economies in 1994, down from 4 percent in 1982. The decrease was most pronounced in countries where MOFA production had been relatively concentrated in the petroleum industry; in that industry, growth in the value of production by MOFA's tended to be constrained by declining oil prices and by the increasing role in the industry by some host governments.
The 1994 survey results indicate that U.S. MNC's continue to play a key, though somewhat diminished, role in U.S. international trade in goods. MNC-associated exports of goods represented 66 percent of total U.S. exports of goods in 1994, down from 77 percent in 1982./1/ MNC-associated imports of goods represented 38 percent of total U.S. imports of goods in 1994, down from 50 percent in 1982./2/
Despite the importance of MNC's in U.S. international trade, local sales, not export sales, were the primary means through which U.S. MNC's serviced both their domestic and foreign customers. The U.S. market was served primarily by U.S. parents, and foreign markets were served primarily by MOFA's. Sales by U.S. parents accounted for 96 percent of sales by U.S. MNC's to unaffiliated U.S. customers in 1994, while sales by MOFA's accounted for 72 percent of sales by MNC's to unaffiliated foreign customers./3/
The benchmark survey.Benchmark surveys are the foundation of BEA's data collection system for USDIA. They are now conducted every 5 years and are both more comprehensive in coverage and more detailed in terms of the items collected than the quarterly and annual sample surveys of USDIA that are also conducted by BEA. Benchmark surveys collect data both on the transactions and positions between U.S. parent companies and their foreign affiliates, which enter the U.S. international transactions accounts and international investment position, and on the overall operations of parents and affiliates. Over the years, the data that are collectedparticularly the operations datahave changed or expanded in response to changing needs and circumstances./4/
The processing of the 1994 benchmark survey is still under way, but enough data have now been tabulated to allow BEA to update its regular annual series of estimates on the operations of nonbank U.S. MNC's./5/ These estimates, which are provided in this article, are preliminary; revised estimates, together with estimates on the operations of bank MNC's and on the transactions and positions between parents and affiliates, will be available next fall, when the final results of the benchmark survey are published./6/
Organization of the article.The remainder of this article comprises three parts and an appendix. The first part discusses the 199394 changes in employment by U.S. parents and their foreign affiliates. The second part provides a profile of U.S. MNC's, including the distribution of affiliates by the percentage of U.S. parents' ownership and the distribution of the worldwide production of goods and services by U.S. MNC's by area, by industry, and by size of business enterprise. The third part highlights selected aspects of U.S.-MNC operations for which data are collected in more detail in benchmark survey years. The appendix discusses the coverage and methodology of the benchmark survey, the changes in the presentation of the results, and the use of the benchmark survey data to help refine and evaluate other BEA estimates of USDIA.
The change in employment of U.S. MNC's in 1994 can be estimated as the net changes in employment that result from changes in existing operations, the acquisition and establishment of affiliates, the sale and liquidation of affiliates, and benchmark revisions (table 2)./7/ Because the benchmark revisions accounted for a large part of the year-to-year change, it was necessary to remove the effects of these revisions from the 1994 benchmark survey results before the results could be compared with the 1993 annual survey estimates./8/ Based on consistent 1993 and 1994 estimates, U.S.-parent employment was virtually unchanged in 199394, as it was, on average, during 198994; foreign-affiliate employment decreased 2 percent, compared with a 1-percent average annual rate of growth during 198994.
In 1994, 917 affiliates were established or acquired by U.S. MNC's; these affiliates had combined employment of 182,900 (table 3)./9/ As in recent years, Europe was the most popular location for new affiliates. New European affiliates accounted for 47 percent of all new affiliates and for 40 percent of their employment. The size, affluence, and integration of the European market are probably the main attractions for U.S. direct investment in the area.
"Asia and Pacific" and "Latin America and Other Western Hemisphere" were the next most popular areas for new investments. The popularity of these areas may primarily reflect the attraction of emerging markets and the new investment opportunities created by the economic liberalizationssuch as the privatization of State-owned monopolies and reduced local-content requirementsin some host countries. Some MNC's may have relocated production for the U.S. market from the United States to low-wage countries in these areas in an effort to reduce labor costs, but it is unlikely that this occurrence was widespread: The share of sales to local customers by MOFA's as a percentage of total sales by MOFA's in these regions was above the average for all countries.
Manufacturing continued to be the most popular industry for new investments in 1994. Affiliates in manufacturing accounted for 33 percent of the number, and for 52 percent of the employment, of all new affiliates.
This section provides a profile of U.S. MNC's in 1994: It includes the distribution of foreign affiliate employment by area and by industry, the U.S. parents' percentage ownership of affiliates, U.S. parents' shares of private-U.S.-business gross product by industry, the MOFA shares of host-country gross product by area and of worldwide MNC operations by industry, and the extent to which gross product is concentrated among the largest parents and MOFA's. Changes in some of these characteristics since the 1982 and 1989 benchmark surveys are also examined.
The broadest perspective on the foreign operations of U.S. MNC's is provided by the data for all foreign affiliates, which cover all foreign business enterprises owned 10 percent or more by a U.S. company. At this level of ownership, a U.S. company is presumed to have a lasting interest in, and a degree of influence over the management of, the affiliates.
Nonbank foreign affiliates employed 7.0 million workers in 1994 (table 4). By area, affiliates in Europe, with 2.8 million employees, accounted for the largest percentage of total affiliate employment. Over two-thirds of European-affiliate employment was in the three largest European economiesthe United Kingdom, France, and Germany. Affiliate employment in Eastern Europe, at 122,000, remained small relative to the European-affiliate total, but it was up from only 600 in 1989 (the last benchmark survey year). Affiliates in Asia and Pacific, with 1.5 million employees, accounted for the next largest percentage of total affiliate employment; Japan, Australia, and Thailand together accounted for 44 percent of affiliate employment in the area. Affiliates in Latin America and Other Western Hemisphere employed 1.5 million workers; affiliates in the two largest economies in the areaBrazil and Mexicoaccounted for 70 percent of the area total. Affiliates in Canada employed 887,000 workers.
By industry, affiliates in manufacturing, with 4.1 million employees, accounted for the largest percentage of total affiliate employment. Within manufacturing, employment was fairly evenly spread among food, chemicals, electrical equipment, industrial machinery and equipment, and transportation equipment; employment in primary and fabricated metal manufacturing was relatively low. "Other industries," with 1.1 million employees, was the next largest major industry; within this category, retail trade and communication accounted for more than half of the total. "Services" affiliates employed 0.7 million workers, two-thirds of whom were employed by affiliates in business services (such as security, building maintenance, and personnel supply services).
Consistent with the "internalization" theory of the origin of MNC's, which suggests that MNC's tend to have certain firm-specific advantages that must be protected by a high degree of control over operations, U.S. direct investors own 100 percent of most of their foreign affiliates./10/ In 1994, 80 percent of all foreign affiliates were wholly owned, and 89 percent of all affiliates were MOFA's (table 5)./11/
U.S. direct investors held 49 to 50 percent of the shares in 6 percent of all affiliates. This level of ownership may allow U.S. direct investors to achieve economies of scale or to widen their market access with little or no need for capital from internal sources (through mergers, for example); it may also reflect host-government requirements that local owners must have the controlling interest.
U.S. direct investors held less than 49 percent of the shares in 6 percent of all affiliates. This level of ownership may allow U.S. direct investors to share knowledge or to facilitate trade with a foreign business without the need to control the management of that business; it may also reflect host-government restrictions or other structural barriers.
MOFA's.Taken together, the operations of U.S. parents and their MOFA's represent the worldwide operations of the U.S. MNC that are unambiguously controlled by the U.S. parent./12/ Gross product is the preferred summary measure of U.S.-MNC operations, and it is only available for both U.S. parents and MOFA's (but not for other affiliates)./13/ Because of this data constraint, and to distinguish unambiguous control within the MNC, the remainder of this profile of MNC operations is limited to an examination of gross product for U.S. parents, for MOFA's, and for parents and MOFA's combined.
U.S. MNC's produced $1.7 trillion of goods and services, as measured by gross product, in 1994 (table 6). By industry of parent, manufacturing accounted for 54 percent of the combined production of U.S. parents and MOFA's; "other industries" (primarily communication and retail trade), for 21 percent; and petroleum, for 11 percent./14/
U.S.-parent share of private-U.S.-business GDP.The gross product of U.S. parents accounted for 26 percent, or $1.3 trillion, of the GDP of all private U.S. businesses in 1994, about the same share as in 1989 but well below the 33-percent share in 1982 (table 7). The decline since 1982 mainly reflected the concentration of U.S. parents in slower growing segments of the economy, such as "petroleum extraction and refining" and manufacturing.
By industry, the shares accounted for by U.S.-parent gross product varied widely./15/ Parents in petroleum extraction and refining accounted for 97 percent of total U.S. GDP in that industry. The parents' share of private-U.S.-business GDP in manufacturing was 59 percent; in services, 8 percent; and in all other industries combined, 17 percent. The very high share of parents in petroleum extraction and refining reflects the domination of the industry by a small number of very large producers with highly integrated global operations. The high share of parents in manufacturing partly reflects their possession of the firm-specific advantages that enable them to serve foreign markets via direct investment.
The low share of parents in services reflects a variety of factors. U.S. direct investment in some service industries is inhibited by institutional factors in some host countries; for example, U.S. direct investment in health care services may be constrained, or even precluded, in countries where the government plays a prominent role in the delivery of health care. Service industries that are characterized by small-scale production may lack the firm-specific advantages that often provide the basis for direct investment in other industries.
This section examines the distribution of MOFA gross product by country, the MOFA share of total GDP of their host countries, and the MOFA share of gross product (and other selected data) of the worldwide MNC's.
MOFA gross product by country.Most of the production of goods and services by MOFA's, as measured by gross product, occurred in the largest high-income foreign countries in 1994 (table 8). MOFA's in the 10 largest high-income foreign economies accounted for just under 70 percent of total MOFA production./16/ In addition, 70 percent of the sales by MOFA's in these countries were local, as were a majority of the sales by MOFA's in most other countries that host operations by U.S. MNC's. These findings suggest that access to markets has been the dominant motivation for U.S. direct investment abroad.
MOFA share of host-country GDP.In 1994, the gross product of MOFA's accounted for 6 percent or more of the GDP in five of the host countries shown in table 9: Ireland (11.9 percent), Canada (8.6 percent), Singapore (8.2 percent), Honduras (7.0 percent), and Costa Rica (6.1 percent); the MOFA share of GDP in the United Kingdom was just under 6 percent (5.9 percent)./17/ By comparison, for U.S. affiliates of foreign companies, no single foreign country accounted for more than 1 percent of U.S. GDP in 1994; all U.S. affiliates combined accounted for only 6 percent./18/
Four of the countries with relatively high MOFA shares of host-country GDPthe United Kingdom, Canada, Singapore, and Irelandeach possess some or all of the following attractions to U.S. direct investment: (1) A common language with that of the United States, (2) marketing and legal systems that are similar to those in the United States, (3) geographic proximity to the United States, (4) political stability, and (5) relatively low corporate tax rates. The comparatively high MOFA shares of GDP in Costa Rica and Honduras partly reflect the important role that U.S.-owned agricultural production plays in those countries' small and relatively undiversified economies.
The MOFA share of host-country GDP was less than 1 percent in seven of the countries shown in table 9: South Africa, Turkey, Japan, the Republic of Korea, Saudi Arabia, China, and India. The low shares in most of these countries probably reflect past or present, and formal or informal, barriers to investment. The low share in South Africa partly reflects the negative U.S. reaction to the former system of apartheid, which led many U.S. companies to disinvest in that country.
The unweighted average of the MOFA shares of host-country GDP (for the countries shown in table 9) decreased from 3.9 percent in 1982 to 3.2 percent in 1989 and 2.9 percent in 1994. The decrease over the 12-year period, however, was not widespread and probably resulted more from factors specific to the petroleum industry than from a general decrease in the contribution of MOFA's to host-country production. Some of the largest decreases were in countries where MOFA production was relatively concentrated in petroleum production (such as the United Kingdom, Indonesia, the United Arab Emirates, and Saudi Arabia). The decreases in these countries reflected a combination of industry-specific factors./19/
MOFA operations in a worldwide MNC context.Data for U.S. parents and MOFA's combined can be used to gauge shifts in the location of the worldwide operations of U.S. MNC's. For example, changes in the MOFA share of worldwide U.S.-MNC gross product, capital expenditures, and employment reflect shifts in the location of the production and productive resources of U.S. MNC's./20/ Changes in the MOFA share of worldwide U.S.-MNC profit-type-return reflect shifts in the source of U.S.-MNC profits from current production./21/
At the all-industries level, the distribution of U.S.-MNC production and employment between the United States and abroad changed little from 1982 to 1994 (table 10). The MOFA share of worldwide MNC gross product edged up slightly, from 22 percent to 23 percent, and the MOFA share of worldwide MNC employment edged up from 21 percent to 23 percent. The MOFA share of capital expenditures rose substantiallyfrom 18 percent to 24 percentpartly because MNC's in the capital-intensive petroleum industry showed a more pronounced shift towards overseas operations than MNC's in other industries. The MOFA share of worldwide MNC profit-type return decreased to 27 percent in 1994 from 31 percent in 1982; the decrease probably reflects changes in economic conditions here and abroad that were relatively less favorable to USDIA in 1994 than in 1982./22/
By industry, the changes in the MOFA share of U.S.-MNC production and employment were more pronounced. In petroleum, the MOFA share of MNC gross product increased from 37 percent to 46 percent, and the MOFA share of MNC employment increased from 23 percent to 27 percent. The faster growth in overseas production and employment than in domestic production and employment reflected the fall in oil prices in 198286. As a result of the falling prices, many oil projects in the United States became unprofitable, because U.S. oil deposits were relatively expensive to develop. In response, U.S. oil companies spent a greater share of their exploration-and-development budgets on overseas projects.
In manufacturing, the MOFA share of MNC gross product increased from 22 percent in 1982 to 26 percent in 1994, and the MOFA share of MNC employment increased from 26 percent to 30 percent. The growth in the MOFA shares partly reflected the increased globalization of economic activity that occurred during this period, when both production abroad by U.S. MNC's and production in the United States by foreign-based MNC's were expanding. Production abroad by U.S. MNC's may have been stimulated by structural economic changes that created new market opportunities in host countriessuch as the enlargement and further integration of the European Union and the economic liberalizations in Latin America and in Eastern Europe. In addition, the growth may reflect global-sourcing strategies of U.S. MNC's; for example, purchases from MOFA's by U.S. parents in manufacturing as a percentage of the parents' total sales rose from 3 percent in 1982 to 5 percent in 1994.
The production of goods and services, both in the United States and abroad, by U.S. MNC's is concentrated among a small number of very large companies. In 1994, the 20 largest nonbank parents (out of a total of 2,658 nonbank parents) accounted for about a quarter of total U.S.-parent gross product, and the affiliates of these parents accounted for about a third of total MOFA gross product (table 11).
The concentrations were especially high in the petroleum industry and in the manufacturing industries of transportation equipment, food products, industrial machinery and equipment, and electronic and other electric equipment. For some of these industries, the high concentrations partly reflect scale economies or other barriers to entry that allow a small number of companies to dominate production; for example, in integrated petroleum production, refining, and marketing, large fixed capital costs must be incurred to reach a profitable scale of operation.
In this section, selected aspects of MNC operations for which data are collected in more detail in benchmark survey years are analyzed. The section begins by analyzing U.S.-MNC trade in goods and MOFA trade in goods by destination or origin, by product, and by intended use. It then examines the channels for delivering goods and services, the sales of goods and services by destination and by affiliation, and the research and development activities of MOFA's and their U.S. parents.
U.S.-MNC-associated U.S. trade in goods consists of (1) intra-MNC trade (trade between U.S. parents and their foreign affiliates) and (2) MNC trade with others (trade between U.S. parents and foreigners other than their foreign affiliates and trade between foreign affiliates and U.S. persons other than their U.S. parents).
U.S. MNC's account for significant shares of both total U.S. exports and total U.S. imports of goods. MNC-associated exports of goods were $337.0 billion or 66 percent of total U.S. exports in 1994 (table 12). MNC-associated imports of goods were $251.3 billion or 38 percent of total U.S. imports. These large shares reflect the significant presence that U.S. MNC's have in the U.S. economy and the fact that U.S. parents tend to be the most globally oriented U.S. firms.
Though large in 1994, the MNC-associated share of total U.S. exports of goods was even larger77 percentin 1982. The share decreased mainly because of the relatively slow growth in exports by U.S. parents to foreigners other than their foreign affiliates. Similarly, the MNC-associated share of total U.S. imports of goods declined from 50 percent in 1982. The share decreased mainly because of the relatively slow growth in imports by U.S. parents from foreigners other than their foreign affiliates.
From 1982 to 1994, intra-MNC exports of goods grew faster than total U.S. exports of goods, and their share of the total increased from 22 percent in 1982 to 25 percent in 1994. Intra-MNC imports of goods grew at about the same rate as total U.S. imports of goods, and their share of the total was 17 percent in both 1982 and 1994.
In 1994, MNC trade with others accounted for 62 percent, or $208.1 billion, of total MNC-associated exports. Most of the MNC trade with others88 percentconsisted of exports shipped by U.S. parents to "other foreigners." Intra-MNC trade accounted for 38 percent, or $129.0 billion, of total MNC-associated exports. Almost all of the intra-MNC exports97 percentwere shipped to MOFA's.
In 1994, MNC trade with others accounted for 56 percent, or $141.7 billion, of total MNC-associated imports. Most of the MNC trade with others86 percentconsisted of imports shipped by "other foreigners" to U.S. parents. Intra-MNC imports accounted for 44 percent, or $109.7 billion, of total MNC-associated imports. Most of the intra-MNC imports94 percentwere shipped by MOFA's.
Exports to MOFA's.In 1994, U.S. exports of goods shipped to MOFA's, at $147.8 billion, were 44 percent of total MNC-associated exports (table 13)./23/ Most of the exports to MOFA's$125.4 billion or 85 percentwere shipped by U.S. parents.
By area of destination, exports to MOFA's in Canada were largest, at $52.8 billion. Exports to MOFA's in Mexico were the next largest, at $14.8 billion. Most of the exports shipped to MOFA's in Canada and Mexico consisted of motor vehicles and parts and reflected the high degree of integrated production operations between U.S. parents and their Canadian and Mexican affiliates. Exports to European MOFA's were $41.9 billion. Within Europe, exports to MOFA's in the United Kingdom were largest, at $12.1 billion.
By product, more than two-thirds of total U.S. exports shipped to MOFA's consisted of machinery and of road vehicles and parts./24/ Exports of chemicals and of "other manufactures" were also sizable.
By intended use, exports for further manufacture accounted for 56 percent of total exports to MOFA's, and exports for resale without further manufacture accounted for 42 percent. Most of the remaining exports were of capital equipment.
Imports from MOFA's.In 1994, U.S. imports of goods shipped by MOFA's, at $118.2 billion, were 47 percent of total MNC-associated imports. Most of the imports from MOFA's$103.5 billion or 88 percentwere shipped to U.S. parents.
By area of origin, Canada accounted for 44 percent, the largest share of any major area, of imports from MOFA's. Mexico had the next largest share for a single country; nearly all of the imports were shipped to U.S. parents. Imports from Asia and Pacific, particularly Singapore, and from Europe, particularly the United Kingdom, were also sizable.
By product, the majority of imports shipped by MOFA's consisted of road vehicles and parts and of machinery. Imports of "other manufactures" and of petroleum and products were also substantial.
Despite the importance of MNC's in U.S. international trade, local sales, not export sales, were the primary means through which U.S. MNC's serviced their customers, both domestic and foreign. The U.S. market was served primarily by U.S. parents, and foreign markets were served primarily by MOFA's. Sales by U.S. parents accounted for 96 percent of sales by U.S. MNC's to unaffiliated U.S. customers./25/ Sales by MOFA's accounted for 72 percent of sales by MNC's to unaffiliated foreign customers./26/
The limited reliance on exports and imports reflects several factors. A local presence makes it easier to accommodate special requirements of the local markets. In addition, many sales are not feasible through international trade, because of trade barriers, transportation costs, or production-cost differentials. Finally, sales of many services (such as lodging) by their very nature require a local presence.
U.S. parents.Total sales by U.S. parents were $3,957 billion in 1994 (table 14). Of the total, sales of goods accounted for 69 percent and sales of services for 27 percent./27/ Of the total sales of goods by U.S. parents, 87 percent were to U.S. persons and 13 percent were to foreigners. More than half of the sales to foreigners were to unaffiliated foreign customers. Of total sales of services by U.S. parents, almost all were to customers in the United States.
By industry, sales of goods by parents were largest in retail trade, in motor vehicles and equipment, and in petroleum and coal products. Sales of services were largest in insurance, communications, and transportation.
MOFA's.Total sales by MOFA's were $1,432 billion. Sales of goods accounted for 86 percent, and sales of services accounted for 12 percent./28/ Most of the sales of both goods and services were to unaffiliated foreign customers in the affiliate's country of location.
By industry, MOFA sales of goods were largest in wholesale trade (particularly in durable goods), in petroleum, and in motor vehicles and equipment manufacturing. Sales of services were largest in insurance and in business servicesparticularly computer processing and data preparation services.
By area, affiliates in Europe accounted for about half of total MOFA sales of both goods and services. Within Europe, MOFA's in the United Kingdom and Germany had the largest sales of both goods and services. Outside Europe, MOFA's in Canada and Japan had the largest sales of both goods and services.
By destination, 67 percent of total sales by MOFA's were to customers in the affiliate's country of location, 23 percent were to customers in foreign countries other than the host country, and 10 percent were to customers in the United States (table 15). Most of the sales to foreign countries other than the host country were to other foreign affiliates of the same U.S. parent.
Local sales accounted for more than two-thirds of total sales by MOFA's in the United Kingdom, Canada, Germany, and Japan.
Sales to customers in foreign countries other than the host country accounted for more than half of total sales by MOFA's in Switzerland, Ireland, and Belgiumall countries that are important distribution centers for the European market. Thus, these sales probably reflected sales to neighboring countries within Europe.
Sales to customers in foreign countries other than the host country accounted for about a third of total sales by MOFA's in Singapore, Hong Kong, and Indonesia. For Singapore and Indonesia, these sales were concentrated in the petroleum industry and reflected the exports of petroleum products to a variety of destinations worldwide. For Hong Kong, the sales were concentrated in wholesale trade and reflected the country's extensive distribution infrastructure that is used to ship goods produced throughout the area.
Sales to the United States accounted for a relatively large share of total sales by MOFA's in Canada and Mexico (mostly in transportation equipment manufacturing), by MOFA's in Singapore (mostly in computer equipment manufacturing), by MOFA's in Nigeria (mostly in petroleum extraction), and by MOFA's in Hong Kong (mostly in wholesale trade).
The 1994 benchmark survey provides data on expenditures on research and development (R&D) performed and funded by U.S. parents and MOFA's and on the R&D employment (the number of scientists, engineers, and other employees engaged in R&D) of parents and MOFA's. This analysis focuses on the R&D performed by U.S. parents and by MOFA's, whether the R&D was financed by the parents or the MOFA's or by others. These data are comparable with the data on R&D performed by all U.S. companies from the National Science Foundation.
In 1994, private expenditures on R&D performed by U.S. parents totaled $79.6 billion or 87 percent of total private expenditures for R&D by MNC's; expenditures by MOFA's were $12.1 billion or 13 percent of the MNC total (table 16). The R&D employment of U.S. parents was 591,000 or 86 percent of total R&D employment by MNC's; employment of MOFA's was 92,000 or 14 percent of the MNC total.
U.S. parents.Total expenditures for R&D performed by U.S. parents were $91.1 billion or 76 percent of total expenditures for R&D performed by all U.S. businesses (table 16). R&D employment of parents was 591,000 or 77 percent of total R&D employment of all U.S. businesses (table 17)./29/
Of the total expenditures for R&D performed by parents, 85 percent was financed by the parents, and most of the remainder was financed by the Federal Government. U.S. parents accounted for 82 percent of the privately funded R&D performed by all U.S. businesses and for more than half of the federally funded R&D. U.S. parents' large share of total U.S. R&D expenditures reflects both their large size and their concentration in R&D-performing industries.
In this article, two measures of the R&D intensity of R&D-performing U.S. parents and of all R&D-performing U.S. companies are usedprivately funded expenditures for R&D as a percentage of sales and R&D employment as a percentage of total employment (table 17)./30/
Based on the expenditures measure, the R&D intensity of R&D-performing parents for all industries combined was identical to that of all R&D-performing U.S. companies; based on the employment measure, the R&D intensity of parents was slightly higher than that of all U.S. companies.
The finding that U.S. parents generally have an R&D intensity similar to that of all U.S. companies is not surprising, given that many of the largest U.S. firms are parents and that parents account for more than four-fifths of the privately funded R&D performed in the United States.
MOFA's.Based on the expenditures measure, the R&D intensity of MOFA's was highest 8 percentin audio, video, and communications equipment manufacturing. Based on the employment measure, the R&D intensity of MOFA's was highest14 percentin computer and data processing services and next highest13 percentin "other transportation equipment" manufacturing (table 17).
Reflecting the tendency of MNC's to perform most of their worldwide R&D in their home country, MOFA's had lower R&D intensity than their U.S. parents in 19 of the 26 industries with comparable data on R&D expenditures. MOFA's R&D intensity was significantly lower in manufacturing of computer and office equipment, of instruments, and of drugs and in computer and data processing services.
The 1994 benchmark survey covered virtually the entire universe of U.S. direct investment abroad in terms of its dollar value. The preliminary results from the survey are based on reported or estimated data for 2,658 nonbank U.S. parent companies and for 21,300 nonbank foreign affiliates. The survey covered all foreign affiliates that had assets, sales, or net income of $3 million or more and their U.S. parents./31/
Three related types of data were collected: (1) Financial and operating data for foreign affiliates, (2) financial and operating data for U.S. parents, and (3) direct-investment-position and balance-of-payments data.
The financial and operating data, which are the subject of this article, include balance sheets; income statements; employment and compensation of employees; U.S. trade in goods; sales by type and destination; research and development expenditures and employment; and external financial positions. The financial and operating data have also been used to compute the estimates of gross product presented in this article.
The direct-investment-position and balance-of-payments data cover financial positions and transactions between U.S. parents and their foreign affiliates. The balance-of-payments data include direct investment capital flows between U.S. parents and their foreign affiliates, income earned by U.S. parents on their direct investments, and royalty and license fees and other private services transactions between parents and affiliates. These data are not shown in this article but will appear in the publication presenting the final results of the survey next fall.
The data collected in the 1994 benchmark survey will provide the basis for further evaluation and refinement of other BEA estimates of U.S. direct investment abroad. For the financial and operating data, the benchmark survey data will allow BEA to improve its estimates for 1995 forward, both by providing the basis for more accurate estimates for affiliates too small to be reported on the annual sample survey and by identifying new reporters that will provide data in the annual survey. For the balance-of-payments and direct-investment-position data, the survey will provide a basis for revising the estimates derived from BEA's quarterly survey of U.S. direct investment abroad for 1994 forward; the revised estimates are scheduled for publication in July 1998. For both types of data, BEA will evaluate the estimates for 199093 to determine whether the incorporation of information obtained in the 1994 benchmark survey would significantly improve their accuracy.
Methodology.The concepts and definitions underlying the 1994 benchmark survey are essentially the same as those underlying the previous benchmark survey, which are described in U.S. Direct Investment Abroad: 1989 Benchmark Survey, Final Results. The methodology for the 1994 survey will be published with the final survey results.
To produce these preliminary results of the 1994 benchmark survey, BEA prepared estimates for the reports that were not received or were not yet processed and for the items that were not reported or were reported incorrectly. The degree of estimation varies from item to item; in some cases, reporters had difficulty supplying the required information because the data were not easily accessible or were unavailable from their financial-accounting records. In particular, data on trade and employment are subject to a higher degree of estimation than other items. The release of one itemcompensation per hour of production workers of majority-owned manufacturing affiliatesis being delayed until the release of the final benchmark data because estimates for this item are not yet available.
In the 1994 benchmark survey, a long form that requested information in considerable detail was filed for affiliates with assets, sales, or net income greater than $50 million. The most detail was obtained for majority-owned nonbank affiliates. In order to reduce the reporting burden, a short form that requested less detail was used for smaller affiliates./32/ For these affiliates, BEA has estimated the items that appear only on the long form, so that the results can be published in the same detail for all affiliates regardless of size.
Changes in industry presentation.The country detail in this article is identical to that in the article on the 1989 benchmark survey and in the articles on the intervening annual surveys. However, three changes have affected the industry detail. First, beginning with the publication of the preliminary 1994 benchmark survey results, the data for nonbank U.S. parents and foreign affiliates exclude savings institutions and credit unions. The change in coverage reflects the reclassification of savings institutions and credit unions from the "finance, except banking" industry (which is covered by the nonbank data) to the industry "depository institutions" (which will replace the industry "banking" in the publication of the final 1994 benchmark results). This change will not materially affect the comparisons of the data for 1993 with the data for 1994, because in 1993, only one U.S. parent and no foreign affiliates were classified as a savings institution or a credit union.
Second, beginning with the preliminary 1994 benchmark survey results, the "communication and public utilities" group was disaggregated, and the "metal mining" and "nonmetallic minerals mining" groups were aggregated, in the industry table stub. Third, beginning with the revised 1993 annual estimates, the names of two industry groups were changed; the group "machinery, except electrical" is now called "industrial machinery and equipment," and the group "electric and electronic equipment" is now called "electronic and other electric equipment."
Benchmark revisions.Both in the 1994 benchmark survey and in the preceding benchmark survey covering 1989, data were required to be filed for all foreign affiliates with assets, sales, or net income of at least $3 million and for their U.S. parents.
In the intervening annual sample surveys covering 199093, data were required to be filed only for foreign affiliates with assets, sales, or net income of at least $15 million and for their U.S. parents. Estimates for 199093 on the operations of "small" affiliates with assets, sales, or net income of $3$15 million and of the parents of only small affiliates were derived by extrapolating forward the data from the 1989 benchmark survey.
When the 1994 benchmark survey forms were received, many new small affiliates and some parents of only small affiliates were identified and were added to the universe. Conversely, other small affiliates that had been carried forward since the last benchmark survey were discovered to have been sold or liquidated since the 1989 benchmark survey; they (and some of their parents) were subtracted from the universe. The net result of these additions and subtractions is shown in table 2 as "benchmark revisions."
Box: General Notes to Tables
Box: Data Availability
1. MNC-associated exports of goods consist of exports shipped by U.S. parents to their foreign affiliates, exports shipped by U.S. parents to foreigners other than their foreign affiliates, and exports shipped to foreign affiliates by U.S. persons other than their U.S. parents.
2. MNC-associated imports of goods consist of imports shipped to U.S. parents by their foreign affiliates, imports shipped to U.S. parents by foreigners other than their foreign affiliates, and imports shipped by foreign affiliates to U.S. persons other than their U.S. parents.
3. Unaffiliated customers are those that are not members of the same MNC.
4. Earlier benchmark surveys covered 1929, 1936, 1940, 1943, 1950, 1957, 1966, 1977, 1982, and 1989. The 1943 survey was conducted by the Treasury Department. The 1950 survey was the first to collect operating data. For a discussion of the evolution of BEA's data collection system in response to changing circumstances, see Betty L. Barker, "Investment Statistics for a Global Economy," in Accuracy, Timeliness, and Relevance of Economic Statistics, edited by Zoltan Kenessey (Voorburg, The Netherlands: Editions Voorburg, forthcoming in 1997).
5. The 1993 annual survey results appeared in "U.S. Multinational Companies: Operations in 1993," SURVEY OF CURRENT BUSINESS 75 (June 1995): 3151.
6. For additional information on the benchmark survey and its linkages to other BEA data on USDIA, see the appendix "The Benchmark Survey."
7. Employment usually provides a more accurate indication of real economic activity than do assets or sales because changes in employment are not directly affected by valuation changes (such as those caused by inflation and by exchange-rate fluctuations). However, employment does not account for differences in productivity over time or across industries. Gross producta measure that can capture differences in productivityis not used in this section of the article (or in other discussions of the operations of all affiliates), because it is unavailable for affiliates that are not majority owned.
8. For a description of the sources of the benchmark revisions, see the appendix.
9. Because of the lower reporting threshold of the benchmark survey, the total number, assets, sales, and employment of newly acquired or established affiliates in 1994 are not comparable with the estimates of the same items for such affiliates in 1993 and other nonbenchmark years (see the appendix). Excluding the affiliates that would have been exempt from reporting on the annual surveys, there were 450 newly acquired or established affiliates in 1994, and they had combined assets of $61,625 million, sales of $19,336 million, and employment of 151,700.
10. These firm-specific advantages, such as superior production or marketing techniques, allow MNC's to overcome the various barriers to investing abroad, such as foreign languages and unfamiliar business environments. However, these advantages may only be successfully realized internationally through affiliates in which the U.S. parent has clear and effective control; otherwise, the parent's control over its proprietary intangible assets, such as product designs or quality control, may be compromised.
For an elaboration of this theory and for other theories of the origin of MNC's, see J. David Richardson, "Multinational Companies: Descriptions and Dimensions," in Understanding International Economics, Theory and Practice (Boston: Little, Brown, and Company, 1980).
11. A U.S. parent may have a direct ownership share in a given foreign affiliate, an indirect ownership share through another of its foreign affiliates, or a combination of direct and indirect ownership shares. The total U.S.-parent ownership share shown in table 5 is the sum of the direct and indirect ownership shares.
12. A U.S. parent represents the consolidation of the majority-owned domestic operations of a U.S. MNC; see the box "Key Terms."
A small percentage of MOFA's are majority owned by a group of U.S. parents in which none of the parents has a majority stake. The group usually influences or controls the management of the affiliate in a manner comparable to that of a single parent with the same total ownership interest. Most of these jointly owned MOFA's are in the petroleum industry, where parents sometimes pool their resources in order to raise capital or to mitigate risk.
13. Some of the data that are needed to compute gross product for less-than-majority-owned affiliates are not collected in the benchmark survey in order to reduce the reporting burden on survey respondents. It is generally much more difficult for survey respondents (that is, U.S. parent companies) to obtain information on minority-owned affiliates because of their lack of control over the affiliates' operations and their inability to require the majority owners to provide the information.
14. When a MOFA is classified by the industry of its U.S. parent, its gross product is assigned to the primary industry of the parent, which may differ from the primary industry of the MOFA. Placing each U.S. parent and all of its MOFA's in the same industry allows comparisons of the domestic and foreign operations of U.S. MNC's at the industry level.
15. At the all-industries level, the estimates of U.S.-parent gross product are generally consistent with the estimates of all-private-U.S.-business GDP in the national income and product accounts. For individual industries, however, inconsistencies may result from differences in the basis for the industrial distribution of the estimates and in the components of gross product included. (The latest estimates of U.S. GDP by industry appeared in "Improved Estimates of Gross Product by Industry, 195994," SURVEY 76 (August 1996): 133155.)
The GDP of all private U.S. businesses is distributed among industries on the basis of the principal product or service of each establishment (factory, mine, store, or office), whereas U.S.-parent gross product is distributed on an enterprise, or company, basis in which each U.S. parent is classified in the principal industry of all its establishments combined. Because the establishments of a large company may be classified in different industries, the distribution of data by industry of establishment can differ significantly from that by industry of enterprise, particularly if the data are highly disaggregated.
The comparability of U.S.-parent gross product and all-U.S. GDP by industry is also limited because U.S.-parent gross product includes, and all-U.S. GDP by industry excludes, subsidies received. (However, subsidies received by U.S. parents are believed to be small.) In addition, U.S.-parent gross product excludes, and all-U.S. GDP by industry includes, business transfer payments. In this article, U.S.-parent gross product as a share of the GDP of all private U.S. businesses is computed only at the highly aggregated level shown in table 7.
16. The World Bank ranks world economies by size on the basis of GDP and classifies them as low- or high-income on the basis of per capita gross national product. In 1994, the 10 largest high-income foreign economies were Japan, Germany, France, Italy, the United Kingdom, Canada, Spain, Australia, the Netherlands, and Switzerland. These rankings are based on data from the World Bank, World Development Report, 1996 (Oxford University Press, 1996).
17. The MOFA shares may be somewhat understated because the host-country GDP data used to compute the shares have not been adjusted to exclude banking, government, and other segments of the economy in which nonbank MOFA's cannot, or do not, invest. The comparability of host-country GDP to MOFA gross product may also be affected by coverage problems or by the use of statistical methods and definitions that differ from those used in deriving the gross product for MOFA's or that differ from one country to another.
18. The share of U.S. GDP accounted for by U.S. affiliates of foreign companies is not strictly comparable with the share of host-country GDP accounted for by MOFA's, because the share of U.S. GDP includes all affiliates, not just those that are majority owned, and because the denominator of that share is adjusted to exclude banking and other industries in which nonbank U.S. affiliates cannot, or do not, invest. For a discussion of the share of U.S. GDP accounted for by U.S. affiliates of foreign companies, see "Foreign Direct Investment in the United States: New Investment in 1995 and Affiliate Operations in 1994," SURVEY 76 (July 1996): 108.
19. First, some host governments sought to increase their own participation (or that of State-owned firms) in petroleum operations in their countries. Second, oil prices began to decrease gradually in 1982 and then fell sharply in 1986, which diminished the (nominal) value of production in petroleum-related industries. The drop in oil prices lowered the value of both MOFA and host-country gross product, but it tended to lower the value of MOFA gross product more than the value of host-country GDP because MOFA's tend to be more highly concentrated in petroleum than their host economies.
20. Changes in the MOFA share of capital expenditures should be interpreted cautiously because of the cyclical nature of capital spending.
21. Profit-type return is an economic-accounting measure of the profits from current production. Unlike net income, it excludes nonoperating items, such as special charges and capital gains and losses, and it excludes income from equity investments.
22. The U.S. economy was in recession in 1982, whereas the economies in Europe were still growing. In 1994, the situation had reversed; real U.S. GDP increased 3.5 percent, compared with only a 2.3-percent increase in the real GDP of the European countries of the Organisation for Economic Co-Operation and Development.
23. In table 13, the line "all areas, all products" for total exports shipped to MOFA's is the sum of lines 3 and 9 for 1994 in table 12, and that line for total imports shipped by MOFA's is the sum of lines 13 and 19 for 1994.
24. The product categories used in the benchmark survey are based on the Standard International Trade Classification (SITC) and are summarized in BEA's Guide to Industry and Foreign Trade Classifications for International Surveys.
25. For this comparison, the sales by U.S. parents have been reduced by the value of the inputs received from MOFA's because most of the goods sent by MOFA's to U.S. parents are probably either intermediate products or final products for resale and, thus, are ultimately embodied in parents' sales. Any capital equipment that may be included among these goods should be excluded from this reduction (because this equipment is not directly embodied in parents' sales), but it could not be excluded, because the data are unavailable. However, it is unlikely that capital equipment accounts for a significant portion of the goods shipped to U.S. parents by MOFA's: Capital equipment shipped by U.S. parents to MOFA's accounted for only 2 percent of the total value of U.S. parents' exports of goods to MOFA's in 1994.
26. For this comparison, the sales by MOFA's excludes the value of the goods (other than capital equipment) exported to MOFA's by U.S. parents.
27. The remaining 4 percent was accounted for by the investment income of parents in the finance and insurance industries; these parents include investment income in sales because it is a primary activity of the company.
The data on the investment income of U.S. parents and MOFA's in finance and insurance is collected in the direct investment surveys. In these surveys, all sales are classified as sales of goods, as sales of services, or as investment income. Sales of services are those characteristic of the following industries: Industries in the "services" division of the Standard Industrial Classification; finance (except depository institutions), insurance, and real estate; agricultural, mining, and petroleum services; and transportation, communication and public utilities. The exclusion of depository institutions reflects their exclusion from the data series generally, not a judgment that they do not belong to a services industry.
Note that a U.S. parent or a MOFA that is not classified in one of these industries can nonetheless have sales of services; for example, U.S. parents in manufacturing had $89.0 billion in sales of services in 1994. Conversely, a parent or MOFA in a services industry can have sales of goods, but MOFA's in services had only $7.3 billion in sales of goods out of total sales of $1,232.5 billion.
28. The remaining 2 percent was accounted for by the investment income of affiliates in finance and insurance.
29. The 1994 estimates for all-U.S.-business R&D are based on the data from the National Science Foundation (NSF), Research and Development in Industry: 1993, NSF 96304 (Arlington, VA, 1996). The data cover all U.S. businesses, including depository institutions.
30. The comparisons between the R&D intensity measures for the U.S. parents and MOFA's and those for all U.S. companies are approximate, because the data for parents and MOFA's are from the 1994 benchmark survey, while the data for all R&D-performing U.S. companies are estimates based on a sample survey of industrial firms in 1993 (and the 1993 sample is a subset of a large probability sample selected for 1992). The measures at the industry level may also differ because parents and MOFA's are classified by industry of sales, whereas in the surveys conducted by the Census Bureau for NSF, R&D-performing U.S. companies are classified by payroll.
31. In order to reduce reporting burden on survey respondents, the rules for the survey exempted foreign affiliates for which assets, sales, and net income were each less than $3 million. In claiming exemption for affiliates, reporters were required to supply values for the three itemsassets, sales, and net incomeon which the claim was based. The data for the exempt affiliates have not yet been tabulated for the 1994 benchmark survey, but in previous benchmark surveys, exempt affiliates have accounted for 1 percent or less of the universe value, so that coverage in terms of value is assumed to be virtually complete.
32. Copies of the long and short forms will be included in the publication containing the final results of the survey.