Foreign Direct Investment: Review of Commerce Department Reports and
Data-Sharing Activities (Letter Report, 09/29/95, GAO/GGD-95-242).

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-95-242
     TITLE:  Foreign Direct Investment: Review of Commerce Department 
             Reports and Data-Sharing Activities
      DATE:  09/29/95
   SUBJECT:  Foreign investments in US
             Interagency relations
             Reporting requirements
             Agency reports
             Data collection operations
             Federal records management
             Surveys
             Investment planning
             Economic analysis
IDENTIFIER:  Dept. of Commerce Survey of Current Business
             Dept. of Commerce Standard Industrial Classification System
             
**************************************************************************
* This file contains an ASCII representation of the text of a GAO        *
* report.  Delineations within the text indicating chapter titles,       *
* headings, and bullets are preserved.  Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are           *
* identified by double and single lines.  The numbers on the right end   *
* of these lines indicate the position of each of the subsections in the *
* document outline.  These numbers do NOT correspond with the page       *
* numbers of the printed product.                                        *
*                                                                        *
* No attempt has been made to display graphic images, although figure    *
* captions are reproduced. Tables are included, but may not resemble     *
* those in the printed version.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


Cover
================================================================ COVER


Report to Congressional Requesters

September 1995

FOREIGN DIRECT INVESTMENT - REVIEW
OF COMMERCE DEPARTMENT REPORTS AND
DATA-SHARING ACTIVITIES

GAO/GGD-95-242

Foreign Direct Investment

(280116)


Abbreviations
=============================================================== ABBREV

  ASM - Annual Survey of Manufacturers
  BEA - Bureau of Economic Analysis
  BLS - Bureau of Labor Statistics
  EIN - employer identification number
  FDI - foreign direct investment
  FDIUS - foreign direct investment in the United States
  GDP - gross domestic product
  IID - International Investment Division
  IRS - Internal Revenue Service
  ISI - International Surveys Industry
  ITA - International Trade Administration
  OGC - Office of the General Counsel
  OMB - Office of Management and Budget
  R&D - research and development
  SSEL - Standard Statistical Establishment List
  SIC - Standard Industrial Classification

Letter
=============================================================== LETTER


B-259956

September 29, 1995

The Honorable Larry Pressler
Chairman, Committee on Commerce, Science, and Transportation
United States Senate

The Honorable Thomas J.  Bliley, Jr.
Chairman, Committee on Commerce

The Honorable Bill Archer
Chairman, Committee on Ways and Means

The Honorable Benjamin A.  Gilman
Chairman, Committee on International Relations
House of Representatives

The Honorable Connie Mack
Chairman, Joint Economic Committee
Congress of the United States

This is our final report responding to a requirement under the
Foreign Direct Investment and International Financial Data
Improvements Act of 1990 (Public Law 101-533) that we analyze the
Secretary of Commerce's first three annual reports on foreign direct
investment in the United States (FDIUS) and review federal government
efforts to improve the quality of foreign direct investment (FDI)
data.\1 Specifically, our objectives were to (1) assess the extent to
which Commerce's second and third reports-- issued in 1993 and
1995--fulfilled the requirements of the 1990 act and addressed the
recommendations in our 1992 report;\2 (2) review the process by which
federal agencies collect FDI data; (3) review the status and
processes of the data exchanges, or links, initiated by the 1990 act
between the Commerce Department's Bureau of Economic Analysis (BEA)
and its Bureau of the Census and between BEA and the Labor
Department's Bureau of Labor Statistics (BLS); and (4) evaluate the
extent to which implementation of the act has brought about the
intended improvements in public information on FDI in the United
States. 


--------------------
\1 FDI is the ownership by a foreign person or business of 10 percent
or more of the voting equity of a firm located in the United States. 

\2 We reviewed Commerce's 1991 FDIUS report in Foreign Direct
Investment:  Assessment of Commerce's Annual Report and Data
Improvement Efforts (GAO/NSIAD-92-107, Mar.  18, 1992). 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Commerce has issued three FDIUS reports--in August 1991, June 1993,
and January 1995.  Commerce has approached each of the three FDIUS
reports differently, in response to changing public concerns about
FDIUS and limited staff and budget resources, according to Commerce
officials.  Together, the 1993 and 1995 reports included discussion
of all the data requirements of the 1990 act for which data exists,\3
and responded to the recommendations in our 1992 report.  In
addition, based on factors such as organizational structure of the
reports and individual chapters, sufficiency of evidence for
principal findings, coverage of specific industry sectors, coverage
of countries with the largest shares of direct investment in the
United States, and use of relevant outside economic studies, we found
that the two reports adequately present the Commerce Department's
analysis and findings. 

Overall, the Commerce reports' analyses and conclusions relating to
the effects of FDIUS on the U.S.  economy were thorough and
reasonable.  There were a few exceptions where Commerce's conclusions
were more definitive than warranted by the evidence.  For example, in
some cases, Commerce characterized FDI as having a "positive impact"
on the U.S.  economy without providing sufficient support for this
statement.  (See app.  I for a more detailed discussion of factors
that, in our view, limited some aspects of Commerce's analysis of the
possible effects of FDIUS on the U.S.  economy.)

The Commerce Department is the principal source of U.S.  government
data on FDIUS.  Commerce's BEA obtains information on FDIUS through
four survey questionnaires that require U.S.  affiliates of foreign
firms\4 to report on a wide range of financial and operating data. 
(See app.  II for a discussion of BEA surveys.) To ensure compliance
with FDIUS survey reporting requirements, BEA has strengthened survey
procedures and increased the number of staff devoted to survey
follow-up. 

The BEA-Census and the BEA-BLS data-sharing efforts, initiated by the
1990 act, have generated data on U.S.  affiliates of foreign firms at
a greater level of industry specificity than was previously
available.  The data have enabled Commerce to provide a richer
description of U.S.  affiliates' activities and to draw more
meaningful comparisons between their operations and those of other
U.S.  firms without imposing any additional burdens on survey
respondents.  However, certain restrictions and other factors related
to the protection of business-confidential data continue to limit
more extensive data sharing among U.S.  government agencies.  (See
app.  III for a more detailed discussion of the BEA, Census, and BLS
data-sharing projects.)

The Commerce Department's FDIUS reports and the data-sharing
activities between BEA, Census, and BLS have largely fulfilled the
purpose of the 1990 act by improving the quality and quantity of
federal government data on FDIUS.  As a result, both government
officials and analysts in the research and academic communities have
access to FDIUS data that was previously unavailable.\5


--------------------
\3 The 1990 act requires that the Commerce reports "compare business
enterprises controlled by foreign persons with other business
enterprises in the United States with respect to employment, market
share, value added, productivity, research and development (R&D),
exports, imports, profitability, taxes paid, and investment
incentives and services provided by state and local governments
(including quasi-governmental entities)."

\4 Commerce uses this term to refer to U.S.  businesses with
investment resulting in the foreign ownership of 10 percent or more
equity interest. 

\5 Information available to government agencies and the public is
subject to the confidentiality requirements that apply to proprietary
business data. 


   BACKGROUND
------------------------------------------------------------ Letter :2

To facilitate the public debate on FDIUS issues by improving existing
government information, Congress enacted the Foreign Direct
Investment and International Financial Data Improvements Act of 1990. 
This act required the Secretary of Commerce to submit an annual
report addressing the history, scope, trends, and market
concentrations of FDIUS, as well as its effects on the U.S.  economy. 
In addition, the act provided for an exchange of
business-confidential data between the Bureau of the Census and BEA
and authorized BLS to have access to selected business-confidential
BEA data.\6 The purpose of this data sharing, as specified by the
1990 act, is to improve the quality of U.S.  government data on FDIUS
and to enhance analysts' ability to assess the impact of that
investment on the U.S.  economy.  BLS gives BEA access to publicly
available macro-level, or aggregated, data on foreign-owned
establishments generated from the BEA-BLS data link project. 

BEA data on foreign investment are collected on a consolidated firm
or "enterprise" basis and reported under the industry category of the
firm's primary business\7 and then linked with "establishment" or
plant-level data collected by Census and by BLS.  Since Census and
BLS data are collected on an "establishment basis"--i.e., from
individual commercial plants--the data are more likely to correlate
to specific industry sectors.  However, Census and BLS data do not
identify foreign ownership.  Linking BEA's enterprise data with
Census' and BLS' establishment data enables Commerce to report on the
operations of U.S.  affiliates of foreign firms in over 800
individual industries at the establishment level, as opposed to only
135 industries at the enterprise level.  The establishment industry
categories are disaggregated according to the Standard Industrial
Classification (SIC) system.\8 See figure 1 for an illustration of
how one industry category within the manufacturing sector is
disaggregated at the 2-, 3-, and 4-digit SIC levels. 

   Figure 1:  Examples of SIC
   Disaggregation of One Industry
   Category

   (See figure in printed
   edition.)

Source:  Standard Industrial Classification Manual, 1987.  Office of
Management and Budget (Washington, D.C.:  U.S.  Government Printing
Office). 

In addition to its three reports, Commerce published data from the
first phase of the BEA-Census data exchange effort in June 1992. 
Commerce also published data from the BEA-Census data exchange effort
in 1993, and again in 1994.  BLS has published 1989, 1990, and 1991
data from the BEA-BLS data-sharing efforts in July 1992, October
1992, October 1993, and December 1994, respectively. 


--------------------
\6 In this report, we use the terms "business-confidential" or
"micro-level" data to refer to data that identify individual
companies or individual plants.  We use the term "macro-level" data
to refer to aggregated, publicly available data that do not identify
individual companies or individual plants. 

\7 For example, a U.S.  affiliate whose primary business is chemicals
but that has substantial petroleum operations would be categorized
entirely as a chemicals investment. 

\8 SIC is the statistical classification standard underlying all
establishment-based (plant-level) federal economic statistics
classified by industry.  The classification covers the entire field
of economic activities and defines industries in accordance with the
composition and structure of the economy. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :3

The Foreign Direct Investment and International Financial Data
Improvements Act of 1990 directs us to analyze and report on
Commerce's first three annual reports on FDIUS and review government
efforts to improve the quality of FDIUS data.  To assess how well
Commerce fulfilled the reporting requirements of the 1990 act, we
reviewed the 1993 and 1995 reports, with specific attention to
Commerce's coverage of the data requirements of the act, and to the
overall quality of Commerce's analysis of the potential effects of
FDI on the U.S.  economy.  In addition, we evaluated the extent to
which the 1993 and 1995 reports responded to the recommendations in
our 1992 report. 

We used standard economic principles in our review and evaluation of
the Commerce reports, with special attention to the chapters relating
to the implications of FDIUS for U.S.  trade, technology transfer,
tax payment, employment, and banking issues.  We relied on internal
economists as well as an outside economist with expertise in FDIUS
issues to carry out this evaluation.  We also considered such factors
as organizational structure, sufficiency of evidence for principal
findings, coverage of the data requirements of the 1990 act, coverage
of specific industry sectors, coverage of major investing countries,
and use of relevant outside studies.  We interviewed officials from
BEA, Census, and BLS, as well as several outside experts, in the
course of our review. 

We obtained written comments on a draft of this report from the
Secretary of Commerce.  They are discussed on page 15 and presented
in their entirety in appendix IV.  We also discussed the results of
our work with program officials in BLS and incorporated their
suggestions where appropriate. 

We performed our review in Washington, D.C., from January 1995 to
August 1995 in accordance with generally accepted government auditing
standards.  See appendix V for a more detailed description of our
objectives, scope, and methodology. 


   COMMERCE'S FDIUS REPORTS
   FULFILLED REQUIREMENTS OF THE
   1990 ACT
------------------------------------------------------------ Letter :4

We found that, taken together, Commerce's 1993 and 1995 FDIUS reports
largely fulfilled the requirements of the 1990 act and addressed the
recommendations in our 1992 report on Commerce's 1991 FDIUS report. 
In an effort to address changing public concerns about FDIUS and
conserve agency resources, Commerce took a different approach to each
of its three reports, according to Commerce officials.  The reports
included discussion of all the data requirements in the act for which
data existed, such as comparing U.S.  affiliates of foreign firms'
operations to those of other U.S.  companies with respect to
employment, exports and imports, and research and development (R&D)
spending.  With few exceptions, the two reports adequately presented
Commerce's analysis and findings regarding publicly debated FDIUS
issues. 


      COMMERCE'S FDIUS REPORTS
      RESPONDED TO PUBLIC CONCERNS
---------------------------------------------------------- Letter :4.1

Commerce has approached each of the three FDIUS reports differently
in terms of organization and content.  Commerce officials said these
differences reflected the changing nature of public concerns about
FDIUS and resource considerations within the agency.  The August 1991
report highlighted the growth and characteristics of FDIUS in five
industry sectors, including electronics, automotives (including
automobile parts and components), steel, chemicals, and banking.  It
provided a description of the initial BEA-Census data link effort,
which was not yet complete at the time of the report's publication. 

In our 1992 review of Commerce's 1991 FDIUS report, we recommended
that Commerce's subsequent FDIUS reports (1) provide an analysis that
clearly distinguishes between costs and benefits derived from FDI and
those derived from all foreign investment in the United States, (2)
make greater use of available government studies and private sector
data, and (3) provide more focused analyses of publicly debated
questions regarding the effects of FDI in the U.S.  economy.  We
subsequently determined that Commerce had adequately addressed our
recommendations in its June 1993 report. 

Commerce's June 1993 FDIUS report was organized by general issues of
public policy concern rather than by industry sector.  The report
contained analyses of the implications of FDIUS for U.S.  merchandise
trade patterns, technology development and transfer, and corporate
tax payment.  It also presented a more detailed description of the
operations of foreign-owned firms\9 in the United States, as well as
the results of the first phase of the data link project, based on
data obtained through 1987 BEA and Census surveys.  Further, the
report included an extensive literature survey on the economic issues
relating to FDIUS, such as technology transfer, exports and imports,
and employment effects. 

The January 1995 report also highlighted general FDIUS issues of
public policy concern rather than specific industry sectors.  It
included new data obtained through the BEA-BLS data link on
occupational employment patterns in foreign-owned manufacturing
establishments, and further analysis of the role of U.S.  affiliates
in U.S.  merchandise trade.  Commerce also reported the results of
the comprehensive 1992 BEA benchmark survey\10 of FDIUS and of
ongoing BEA-Census and BEA-BLS data link projects.  The primary
factor that distinguished it from the 1993 report was that, with the
exception of the introduction and chapter 6, all of the chapters of
the 1995 report were reproductions of articles previously published
in BEA's monthly Survey of Current Business or contained data
previously released in BLS publications.  Commerce officials told us
they believed this approach was a better use of limited staff and
budget resources, given the cyclical nature of public concerns about
FDIUS. 


--------------------
\9 A foreign-owned firm is a business in the United States in which
there is sufficient foreign investment to be classified as foreign
direct investment, as defined by Commerce. 

\10 This survey is normally conducted every 5 years. 


      COMMERCE COVERED THE DATA
      REQUIREMENTS OF THE 1990 ACT
---------------------------------------------------------- Letter :4.2

Together, Commerce's 1993 and 1995 FDIUS reports covered all of the
data requirements specified by the 1990 act for which data existed
and presented Commerce's analysis and findings in a comprehensive
manner.  Specifically, Commerce presented extensive data on the
history, scope, trends, and market concentrations of FDIUS.  It also
compared the operations of U.S.  affiliates of foreign firms with
those of other business enterprises in the United States with respect
to employment, value added, productivity, R&D spending, exports and
imports, profitability, taxes paid, and market share. 

The market share information was limited by Commerce's data
aggregation and confidentiality requirements.  To the extent possible
with existing data, Commerce reported on U.S.  affiliates of foreign
firms' market concentration in various U.S.  industries.  Commerce
relied primarily on sales data to estimate U.S.  affiliates' market
share, but also examined U.S.  affiliates' share of U.S.  gross
domestic product (GDP) and employment.  However, if Congress is
concerned about the amount of foreign control exercised in specific
product sectors, market sales data at a less aggregated level would
be needed.\11 According to Commerce officials, presenting more
detailed sector data in these reports would likely compromise the
confidentiality requirements of data collection agencies (see
discussion in app.  III). 

The one item called for in the law (section 3(c)(1)) but not
addressed in the reports was information about investment incentives
and services provided by state and local governments, including
quasi-government entities.  According to a BEA official, BEA attempts
to collect this type of data through its survey of U.S.  business
enterprises newly acquired or established by foreign investors. 
However, a BEA official said that in many cases the data BEA receives
are not complete.  Therefore, the reliability of these data is
questionable, and they are not published in Commerce's FDIUS reports. 
These data are, however, publicly available upon request with a
disclaimer from Commerce about their reliability. 


--------------------
\11 Detailed data on U.S.  affiliates at the 4-digit SIC level appear
in the Census Bureau's Foreign Direct Investment in the United
States:  Establishment Data for 1987 (Washington, D.C.:  June 1992). 
In some cases, information is suppressed to avoid disclosure of data
on individual companies. 


      COMMERCE'S CONCLUSIONS
      SOMETIMES MORE DEFINITIVE
      THAN WARRANTED BY THE
      EVIDENCE
---------------------------------------------------------- Letter :4.3

While our economic review showed Commerce's analysis to be adequate,
in some instances its interpretation of the effects of FDI in the
U.S.  economy were overly definitive.  The conclusion in chapter 8 of
Commerce's 1995 report, with regard to the occupational employment
patterns of foreign-owned manufacturing establishments, illustrates
this problem.  In the conclusion, the report stated that "on balance,
foreign investment in high skill industries has a positive impact on
the U.S.  manufacturing labor market." However, the statistical data
presented showed that foreign and U.S.-owned firms were actually
similar in the occupational distribution of their employees.  A
similar problem appears in Commerce's discussion of technology
transfer issues in chapter 6 of the 1993 report (see app.  I). 

To reach a more definitive conclusion on the "positive impact" of
FDIUS, the analysis would require a comparison of the observed
scenario to the scenario that would have occurred in the absence of
FDI, sometimes called the 'counterfactual' scenario.  While it is not
possible to state with absolute certainty what would have happened
(the counterfactual), this approach often highlights important
assumptions about the cause and effect relationships between various
factors.  In some cases, Commerce did not formally include such
scenarios in its analysis (for more details, see app.  I). 


      LINKED DATA PRESENTED IN
      COMPARATIVE ANALYSES OF U.S. 
      AFFILIATES AND U.S.-OWNED
      FIRMS
---------------------------------------------------------- Letter :4.4

The 1995 report shifted emphasis away from the economic effects of
FDI toward a general comparison of the operational behaviors of U.S. 
affiliates of foreign firms to those of U.S.-owned firms.  Commerce
used linked data to examine several characteristics of firms,
including plant scale, plant and equipment expenditure, R&D spending,
capital intensity, skill level, wage compensation, and labor
productivity.  These analyses are helpful in identifying the
potential effects of FDI and in determining the industry sectors that
have attracted the most foreign investment. 


      SOME COMPLEX QUESTIONS
      REMAIN DIFFICULT TO ANSWER
---------------------------------------------------------- Letter :4.5

Commerce's ability to perform statistical analyses on FDIUS-related
questions is currently limited by the level at which available data
are aggregated.  In its analyses, Commerce presently uses the 3-digit
SIC-level data, and where possible, the 4-digit SIC level data.  Data
at the 4-digit SIC level are sufficiently detailed to address some
issues, such as the role of U.S.  affiliates of foreign firms in U.S. 
employment and GDP, but other issues related to market control and
technology transfer could be more effectively addressed using more
narrowly defined industry categories. 

Nevertheless, because some FDIUS questions are so complex, definitive
conclusions would be hard to draw even if less aggregated data were
available.  For example, it would be difficult to determine
empirically whether foreign firms invest in the United States with
the intent of acquiring U.S.  technology.  A fuller understanding of
the technology strategies employed by foreign investors in the United
States would require continued research and debate.  Determining the
effects of FDI on U.S.  imports and exports and on federal tax
revenues would also be empirically difficult in some cases. 


   COMMERCE IS THE PRINCIPAL
   SOURCE OF FEDERAL GOVERNMENT
   DATA ON FDIUS
------------------------------------------------------------ Letter :5

Within the U.S.  government, the Commerce Department is the principal
source of U.S.  government data on FDIUS.  BEA collects FDIUS data
directly from U.S.  businesses through surveys, while the
International Trade Administration (ITA) obtains its data primarily
from news accounts of FDIUS transactions, according to Commerce.  In
addition, the Census Bureau within Commerce collects detailed
information on the operations of nearly all U.S.  businesses, both
foreign and domestically owned.  However, Census does not have
systems established specifically to track FDIUS. 

Many other federal government entities collect data on foreign
investment incidental to their overall missions.  The Treasury
Department is primarily responsible for collecting data on portfolio
foreign investment, which includes bonds and other debt instruments
as well as equity interest of less than 10 percent.  The Departments
of Agriculture, Energy, and Defense monitor certain aspects of
foreign investment related to their particular industries.\12


--------------------
\12 For more information on federal data collection on foreign
investment in the United States, see Foreign Investment:  Federal
Data Collection on Foreign Investment in the United States
(GAO/NSIAD-90-25BR, Oct.  3, 1989). 


      BEA HAS ESTABLISHED STEPS TO
      ENSURE COMPLIANCE WITH ITS
      FDIUS SURVEYS
---------------------------------------------------------- Letter :5.1

BEA obtains information on FDIUS through four survey questionnaires
that cover a wide range of financial and operating data for U.S. 
affiliates of foreign firms.  Data reported by survey respondents are
classified according to BEA's International Surveys Industry (ISI)
classification system, which is based roughly on SIC categories.\13
Beginning in 1990, BEA established steps to ensure compliance with
its FDIUS surveys by strengthening survey follow-up procedures and
increasing the number of staff devoted to survey follow-up. 


--------------------
\13 The ISI coding system is an aggregation of detailed codes
contained in the 1987 SIC manual. 


         FDIUS SURVEYS
-------------------------------------------------------- Letter :5.1.1

BEA's FDIUS surveys require qualifying companies to disclose
financial and operational data to BEA in accordance with the
International Investment and Trade in Services Survey Act (Public Law
94-472, 22 U.S.C.  3101-3108, Oct.  11, 1976, as amended).\14 The
individual responses are considered business proprietary information,
and only aggregated data are publicly released.  These surveys cover
such topics as balance of payments flows, U.S.  business enterprises
acquired or established by FDI, and the operations of U.S. 
affiliates of foreign firms. 

The ISI classification system that BEA uses in collecting data on
U.S.  affiliates of foreign firms is roughly based on the SIC system
at the 3-digit level.  To facilitate survey responses, the ISI system
combines certain SIC industry categories based on typical company
structures of U.S.  affiliates.  According to BEA officials, the ISI
classifications correlate more closely with the organizational
arrangement of U.S.  affiliates than does the SIC system, which is
designed for classifying individual establishments within an
enterprise. 


--------------------
\14 Under the International Investment and Trade in Services Survey
Act, U.S.  businesses are required to report to the U.S.  government
if FDI results in ownership or control of 10 percent or more equity
interest in the U.S.  business.  Failure to file the required forms
is punishable by a civil penalty of not less than $2,500 and not more
than $25,000 (22 U.S.C.  3105 (a)), or by a criminal penalty not to
exceed $10,000 or imprisonment for not more than 1 year, or both (22
U.S.C.  3105(c)). 


         EFFORTS TO ENSURE SURVEY
         COMPLIANCE
-------------------------------------------------------- Letter :5.1.2

In response to reduced compliance with reporting requirements among
large company\15 reporters in the 1987 benchmark and 1988 annual
FDIUS surveys, BEA has instituted efforts to ensure U.S.  affiliates
of foreign firms' compliance with its benchmark and annual FDIUS
survey reporting requirements, according to a BEA official.  By the
end of November 1989--6 months after the May 31 reporting
deadline--BEA had received 68 percent of the large company reports in
the 1988 annual survey, compared with 84 percent received by the same
time in the 1987 survey and 92 percent in the 1986 survey.  BEA
officials told us that one of the factors that may have contributed
to this decline in compliance was the rapid (39 percent) growth in
the numbers of qualified large companies to which BEA sent surveys
between the survey covering 1986 and the survey covering 1988.\16
They said that BEA's survey follow-up procedures and staff resources
at the time were not sufficient to manage the growing volume of
potential reporters. 

Beginning with the annual survey covering the year 1989, BEA's
International Investment Division (IID), together with Commerce's
Office of the General Counsel (OGC), undertook a concerted effort to
tighten procedures and ensure U.S.  affiliates' compliance with the
1989 survey and subsequent surveys.  For example, "repeat offenders"
(those large companies that were late in reporting in both the 1988
survey and the 1989 survey) were sent a letter from Commerce's OGC in
place of IID's standard follow-up letter.  In addition, IID and OGC
accelerated their telephone follow-up for late reporters.  Further,
Commerce carried out standard compliance procedures earlier in the
processing cycle compared with previous years.  A BEA official also
told us that in fiscal year 1991 Congress appropriated increased
funding to BEA for survey compliance efforts.  As a result, BEA now
has three full-time staff devoted primarily to FDIUS survey follow-up
efforts.  This official explained that, prior to the 1991 funding
increase, each survey editor was expected to conduct his own
follow-up work. 

In early 1990, BEA developed indicators to measure one key element of
compliance--the timeliness of reporting by large company respondents. 
The indicators show (1) the cumulative number of reports received by
BEA, on a monthly basis, over the 11-month period following the
annual survey mailing in March, and (2) the cumulative dollar value
of the assets associated with those reporting companies, for the same
period.  Since BEA began implementing steps to address the reduced
compliance with the 1987 and 1988 surveys, the timeliness of
reporting on subsequent surveys has returned to acceptable levels,
according to BEA officials.  For example, the percentage of reports
received within 6 months of the May 31 reporting deadline increased
from 68 percent for the 1988 survey, to 92 percent for the 1989
survey (for which BEA first tightened its compliance procedures), and
to 96 percent for the 1993 survey; while the cumulative value of the
assets associated with those reporting companies increased from 69
percent for the 1988 survey, to 92 percent for the 1989 survey, and
to 98 percent for the 1993 survey.\17 Based on these data, BEA
officials believe that their efforts to maintain high compliance
rates have had a positive, measurable impact on the timeliness of
reporting by large companies. 

BEA has established other systems to improve its surveys and data
management processes, a BEA official told us.  BEA has a continuous
process to improve the quality of its survey forms, which includes
proposing changes to the forms, soliciting feedback from survey users
and respondents through a series of meetings and discussions,
publishing a request for public comment on proposals, and finally,
submitting proposals to the Office of Management and Budget (OMB) for
formal review and clearance.  This official said that BEA has also
instituted an office-wide "best practices"\18 initiative to ensure
the accuracy of the data it produces and tabulates.  Formal "best
practices" standards are now part of each BEA staff member's work
plan. 


--------------------
\15 "Large companies," as defined by Commerce, are those with assets
or sales greater than $100 million.  Delays in reporting by these
companies have the greatest impact on the quality of BEA's FDIUS
estimates. 

\16 According to BEA documents, the number of qualified large
reporters to which BEA mailed surveys increased from 837 in the
survey covering 1986 to 1,160 in the survey covering 1988.  For the
survey covering 1993, the number of large company survey recipients
had increased to 1,841. 

\17 A BEA official told us that some of the perceived improvement in
compliance between 1989 and 1993 may be attributable to the use of
different methods for computing compliance rates.  The total number
of reports used to calculate compliance rates for the 1986-89 surveys
may have included enterprises that were later disqualified because
they were sold, liquidated, or consolidated into another report, or
for other reasons, while these enterprises were excluded from the
total number of qualified reporters for calculations made for the
1990-1993 surveys. 

\18 "Best practices" is a term that is commonly used in the field of
organizational quality management to refer to those practices
identified by an organization through experience as particularly
effective in achieving that organization's goals. 


   DATA-SHARING EFFORTS GENERATED
   NEW, DETAILED FDIUS DATA
------------------------------------------------------------ Letter :6

The BEA-Census and BEA-BLS data link projects, initiated under the
1990 act, have greatly improved the amount and quality of data
available about FDIUS.  The data have enabled Commerce to produce
more detailed analyses of FDIUS and to draw more meaningful
comparisons between the activities of U.S.  affiliates of foreign
firms and those of U.S.  firms than previous data allowed.  For
example, by comparing the market and employment shares of
foreign-owned establishments with U.S.  establishments, Commerce has
been able to respond to concerns about the possibility that foreign
investors might be acquiring a disproportionate level of ownership in
certain U.S.  industries. 

Thus far, the data link project between BEA and Census has generated
data covering the number, employment, payroll, and shipments or sales
of foreign-owned establishments in 1987\19 and foreign-owned
manufacturing establishments' operations in 1988, 1989, 1990, and
1991.\20 Commerce's FDIUS reports have included the results of the
data links for 1987, 1989, and 1990.  The BEA-BLS data link project
has generated data on the employment and wages of foreign-owned
establishments in all industries in 1989 through 1991, as well as the
occupational employment of foreign-owned manufacturing establishments
in 1989, which was included in Commerce's 1995 report. 

Data provided by Commerce and BLS officials show that the data link
projects have been carried out at an average annual cost of about
$1.6 million.  According to BEA officials, although BEA does not have
a separate budget line item for the BEA-Census and BEA-BLS data link
projects and does not separately track costs incurred on these
projects, BEA officials estimate that BEA's average annual cost of
carrying out the projects was about $1 million for 1991 through 1995. 
Of this amount, an average of $300,000 per year was paid by BEA to
Census, to reimburse Census for its costs associated with the
project.  The average annual budget for BLS to perform the BEA-BLS
data link project was slightly less than $600,000 between 1991 and
1995.\21 A major achievement of the two data link projects is that
they have produced significant and extensive new data without causing
any increase in companies' reporting burdens. 

According to Commerce officials, several opportunities exist to
improve FDIUS data sharing.  These include expanding the BEA-Census
data link project to include other data items and attempting to
resolve differences between BLS' and Census' establishment databases. 
However, resource constraints, as well as other factors related to
the protection of business confidential data, may limit the agencies'
ability to pursue such activities.  (See app.  III for a discussion
of these factors.)


--------------------
\19 This includes both manufacturing and nonmanufacturing
establishments. 

\20 Commerce's coverage of foreign-owned manufacturing
establishments' operations for these years include six additional
data items.  See appendix III for more detail. 

\21 The average annual budget for BLS is based on an average for
1991, 1992, 1994, and 1995--the 4 years that Congress appropriated
funding for the BEA-BLS data link project.  In 1993 Congress did not
appropriate funding for the BEA-BLS data link project. 


   IMPLEMENTATION OF THE 1990 ACT
   HAS IMPROVED U.S.  GOVERNMENT
   DATA ON FDIUS
------------------------------------------------------------ Letter :7

Based on our review of the Commerce Department's FDIUS reports and
data exchange activities, we found that the implementation of the
1990 act has improved the quantity and quality of U.S.  government
FDIUS data to a great extent.  The data link operations mandated by
the law produced significant improvements in publicly available FDIUS
data, according to BEA, Census, and BLS officials.  This was done at
an average annual cost of about $1.6 million.  These officials told
us they believe the benefits of the data link have been well worth
the investment.  The new data are available to the public through
several means, including regularly published Commerce Department and
BLS reports, Commerce's National Trade Data Bank,\22 and annually
produced computer disks that can be purchased from Commerce and BLS. 

The Commerce Department reports mandated by the law have provided a
regular venue for disseminating new FDIUS data and current analysis
of publicly debated questions relating to the effects of FDIUS on the
U.S.  economy.  With each publication, the Commerce reports'
coverage, analysis, and organization have provided a growing body of
quality information on FDIUS.  The most recent report, issued in
1995, presented a large amount of data on the characteristics of U.S. 
affiliates of foreign firms, including extensive use of tables and
graphics.  It also included the new data obtained from BEA's 1992
benchmark survey, the BEA-Census data link, and the BEA-BLS data
link.  Overall, it provided useful information for further analysis
by Commerce and other analysts about the potential economic effects
of FDI on the U.S.  economy. 

In our view, compiling previously published articles and data is a
reasonable approach to fulfilling the reporting requirements of the
1990 act in a period of government budgetary constraint and when
FDIUS issues have been extensively covered in BEA's Survey of Current
Business and in periodic joint BEA-Census and BLS publications. 


--------------------
\22 Commerce's National Trade Data Bank was established in 1988 to
provide public access, including electronic access, to the export
promotion and international economic data of 15 federal agencies. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :8

We received written comments on a draft of this report from the
Secretary of Commerce.  These comments were of a technical nature,
and we have incorporated changes in the report where appropriate.  A
copy of the Secretary's comments is presented in appendix IV.  We
also discussed the draft report with program officials in BLS and
incorporated their suggestions where appropriate. 


---------------------------------------------------------- Letter :8.1

We are providing copies of this report to the Secretary of Commerce
and other interested parties.  We will make copies available to other
parties upon request.  Major contributors to this report are listed
in appendix VI.  If you have any questions concerning this report,
please contact me on (202) 512-4812. 

Allan I.  Mendelowitz, Managing Director
International Trade, Finance, and Competitiveness Issues


FACTORS THAT LIMITED SOME ASPECTS
OF THE COMMERCE DEPARTMENT'S
ANALYSIS
=========================================================== Appendix I

Although the findings presented in Commerce's 1993 and 1995 reports
were generally reasonable and credible, we found several factors that
limited some aspects of the reports' analysis.  In some cases,
Commerce did not clearly acknowledge that firm conclusions could not
be drawn without the use of "counterfactual scenarios" to account for
economic conditions in the absence of foreign direct investment
(FDI).  We also found that Commerce's statements regarding the
positive impact of U.S.  affiliates of foreign firms' research and
development (R&D) spending did not acknowledge the possibility that
technological developments resulting from R&D do not necessarily
benefit the U.S.  economy.  Finally, in one case, we noted statements
in the 1993 report that seemed contradictory. 


   CERTAIN CONCLUSIONS REGARDING
   FDI WERE OVERLY DEFINITIVE
--------------------------------------------------------- Appendix I:1

In the 1993 report, Commerce sometimes reached conclusions about the
effects of FDI on the U.S.  economy without acknowledging possible
"counterfactual scenarios," i.e., what would have happened in the
absence of FDIUS.  Such scenarios are often used in discussions of
the effects of policy changes.  The difference between the observed
scenario--when FDI is present--and the counterfactual scenario--when
FDI is not--would constitute the effects of FDI.  While it is not
possible to state with any certainty what would have happened in the
absence of FDI, the counterfactual approach can highlight important
assumptions about the cause and effect relationships between various
economic factors. 

One example of where a counterfactual scenario would have been useful
is in chapter 8 of Commerce's 1995 report, which addresses the
occupational employment patterns of foreign-owned manufacturing
establishments.  In the conclusion of this chapter, the author stated
that "on balance, foreign investment in high skill industries has a
positive impact on the U.S.  manufacturing labor market." However,
the statistical data presented showed that foreign and U.S.-owned
firms were actually similar in the occupational distribution of their
employees.  Without knowledge of U.S.  labor market conditions in the
absence of FDI, one cannot draw definitive conclusions about the
positive impact of FDI on the U.S.  labor market. 

An example of how this discussion of counterfactual scenarios can be
used effectively appeared in chapter 6 of the 1995 report.  The
author pointed out that the trade deficits of U.S.  affiliates of
foreign firms amounted to more than half of the total amount of the
U.S.  merchandise trade deficit\1 in recent years, and that most of
the U.S.  affiliates' deficit was accounted for by wholesale trade
affiliates rather than manufacturing affiliates.  The author
concluded that the overall "effect" of those wholesale trade
affiliates on trade flows was unclear:  on the one hand, many of
their imports probably would have been brought into the country by
unaffiliated U.S.  wholesalers, even in the absence of U.S. 
affiliates; on the other hand, U.S.  affiliates may have allowed
foreign parent companies to expand their exports to the United
States.  The author's discussion of scenarios that might have
occurred in the absence of FDI improves our understanding of the
possible effects of FDI on the U.S.  economy. 


--------------------
\1 The U.S.  merchandise trade deficit is the amount by which U.S. 
merchandise imports exceed merchandise exports. 


   TECHNOLOGY ISSUES WERE
   ADDRESSED, BUT CONCLUSIONS
   ABOUT R&D SPENDING WERE OVERLY
   DEFINITIVE
--------------------------------------------------------- Appendix I:2

Commerce's analyses of the implications of FDIUS for the development
and transfer of U.S.  technology included an extensive amount of
relevant data.  Particularly useful were Commerce's analyses of the
market concentration of U.S.  affiliates of foreign firms in
high-technology sectors, and U.S.  affiliates' royalty and licensing
fee payments to foreign parent companies. 

We found that Commerce's conclusions about recent patterns of R&D
spending by U.S.  affiliates of foreign firms were overly definitive. 
Commerce concluded that "U.S.  affiliates have contributed to U.S. 
technological development, dramatically increasing their R&D spending
in the United States over the past ten years." In our review of the
economic literature on the motives of multinationals' FDI, we found
that higher R&D spending by U.S.  affiliates does not necessarily
lead to a higher technology development in the U.S.  economy. 
Sometimes foreign firms locate in the United States simply to monitor
the technology developments of other firms in this country.\2 Even if
R&D funds are dedicated to technology development, there is no
guarantee that such spending will ultimately benefit the U.S. 
economy. 

In one instance in the 1993 report, the authors made two statements
that seemed to be contradictory.  On one hand, Commerce presented
evidence to suggest that some foreign firms have used their
affiliates to gain U.S.-developed "critical technologies"\3 and to
displace U.S.  firms.  Commerce's evidence was based on several case
studies of certain high-technology industries conducted by experts
within and outside the Commerce Department.  On the other hand,
Commerce concluded from its own systematic data analysis that there
was little evidence that foreign acquisitions of small, U.S. 
high-technology firms had resulted in large scale technology transfer
abroad.  Rather, Commerce said that the data suggest U.S.  affiliates
of foreign firms were contributing positively to U.S.  R&D investment
and technological development, and that the R&D spending patterns of
U.S.  affiliates were similar to those of domestic firms.  In our
view, the evidence Commerce cited to support its broad statement that
U.S.  affiliates have contributed to U.S.  technological development
was not sufficiently strong to support the overall conclusion,
because Commerce's analysis did not include discussion of possible
counterfactual scenarios. 

Due to the complexity of the technology issues and the limitations of
the SIC data classification system, some questions cannot be
conclusively answered at this time.  For example, Commerce's effort
to evaluate FDI's presence in sectors that engaged substantially in
the development of critical technologies was hampered by the
aggregated level of available data.  To describe the activities of
companies involved in the production of critical technologies, the
data would have to be significantly disaggregated--beyond the 3- or
4-digit Standard Industrial Classification (SIC) code levels.\4

Neither the 4-digit SIC level nor the "DOC-3"\5 data developed by
Commerce is sufficiently detailed by industry to address questions
about the activities of U.S.  affiliates of foreign firms in U.S. 
critical technology sectors.  Commerce used a modified version of the
DOC-3 definition in its analysis.  Based roughly on 3-digit SIC
codes, this definition includes only broad industry groups such as
"industrial chemicals and synthetics," "computers and office
machines," "electronic components," "instruments and related
products," and "other transportation equipment." Some of the products
included in the definition are actually low-technology products.  For
example, the "computers and office machines" category includes such
products as scales, balances, cash registers, and adding machines;
and "other transportation equipment" includes ship and boat building
and railroad equipment.  Similar limitations exist with the 4-digit
SIC data.  For example, the 4-digit "electrical machinery, equipment,
and supplies, not elsewhere classified" (SIC 3629) category includes
both high-technology items, such as "atom smashers" (particle
accelerators) and "cyclotrons," and low-technology items, such as
"Christmas tree lighting sets."


--------------------
\2 Richard E.  Caves' analysis of Japanese investment shows that an
important motive for Japan's FDI is to acquire technology.  See
"Japanese Investment in the United States:  Lessons for the Economic
Analysis of Foreign Investment," The World Economy, Vol.  16, Number
3 (May 1993). 

\3 Commerce defines critical technologies as those that generate
future innovations in a wide range of goods and services. 
Specifically, critical technologies satisfy long-term national
security, economic, or scientific objectives, such as a strong
national defense, improved economic competitiveness, a rising
standard of living, improved public health, and energy independence. 

\4 According to Commerce, the SIC system is too broad to capture the
existence, the extent, and the frequency with which any given
critical technology is embodied in a product and thus reflected in an
industry's output or trade data. 

\5 Commerce developed the DOC-3 definition classification scheme
specifically for studying U.S.  industries that use advanced
technologies.  DOC-3 classifies industries by their total R&D
spending, including R&D spending for both the final product and for
related inputs. 


SURVEYS CONDUCTED BY THE BUREAU OF
ECONOMIC ANALYSIS (BEA)
========================================================== Appendix II

BEA collects data on foreign direct investment in the United States
(FDIUS) through four survey questionnaires that require U.S. 
affiliates of foreign firms to disclose a broad range of financial
and operating data.  The most comprehensive of these surveys is BEA's
benchmark survey, which is required by law to be conducted every 5
years.  The other three FDIUS surveys collect data on the status of
newly acquired or established U.S.  affiliates, the current
operations of U.S.  affiliates, and on balance of payments flows
between U.S.  affiliates and their foreign parents. 


   BENCHMARK SURVEY
-------------------------------------------------------- Appendix II:1

The International Investment and Trade in Services Survey Act (P.L. 
94-472, 22 U.S.C.  3101 to 3108, as amended), requires BEA to conduct
the benchmark survey of FDIUS (or census) at least once every 5
years.  The most comprehensive of the BEA surveys, it collects both
financial and operating data and balance of payments data for the
entire universe of U.S.  affiliates of foreign firms with more than
$1 million in total assets, sales, or net income during the benchmark
year.  It includes balance sheets and income statements; measures of
employment and employee compensation; sales of goods and services;
property, plant, and equipment; merchandise trade; research and
development expenditures; and, for selected items, data broken down
by state.  Although it is normally conducted every 5 years, the 1987
benchmark survey was conducted after a 7-year interval in order to
coincide with the Census Bureau's quinquennial economic census.  The
purpose of this adjustment was to facilitate the link between BEA's
enterprise data and the Census Bureau's establishment data, and to
enhance their analytical usefulness, according to Commerce. 


   OPERATIONS OF U.S.  AFFILIATES
   OF FOREIGN FIRMS
-------------------------------------------------------- Appendix II:2

BEA's annual sample survey of FDIUS collects data on the overall
operations of nonbank\1 U.S.  affiliates of foreign companies.  This
survey provides annual updates of the financial and operating data
collected in BEA's benchmark surveys.  A key measure is the value of
total assets of U.S.  affiliates at year end.  The annual and the
benchmark surveys are the only BEA sources of foreign investment data
by state.\2 Data from the annual FDIUS survey have been available
since 1977. 


--------------------
\1 Nonbank U.S.  affiliates include all companies except those
classified as banks, savings and loan institutions, credit unions, or
other business entities that accept deposits. 

\2 Some state-level data are also available from the BEA-Census and
BEA-BLS data links. 


   SURVEY OF U.S.  BUSINESSES
   ACQUIRED OR ESTABLISHED BY
   FOREIGN DIRECT INVESTORS
-------------------------------------------------------- Appendix II:3

Data collected by BEA's survey of U.S.  business enterprises acquired
or established by foreign direct investors\3 is complied on an annual
basis.  This data series covers new direct investments and collects
data on the associated transactions only for the year in which the
new investments were made and includes all financing, including local
borrowing in the United States.  Data have been available since 1979. 
An adjunct form is to be filed by persons who act as intermediaries,
such as attorneys or accountants, for new direct investment
transactions and is used only to obtain the names and addresses of
the principals to the transactions so that the primary form can be
mailed to the appropriate person. 


--------------------
\3 This form is required for new investment transactions in which a
foreign person, or a U.S.  affiliate of a foreign person, acquires at
least a 10-percent ownership interest, provided the total cost of the
investment is at least $1 million or involves the acquisition of 200
or more acres of land. 


   FOREIGN DIRECT INVESTMENT
   POSITION AND BALANCE OF
   PAYMENTS FLOWS
-------------------------------------------------------- Appendix II:4

BEA's survey on the U.S.  foreign direct investment position and
balance of payments flows is a quarterly sample survey that collects
information on transactions between U.S.  affiliates of foreign firms
and their foreign parent companies for inclusion in the U.S.  balance
of payments accounts, the national income and product accounts, and
in calculating the inward FDI and international investment position
of the United States.\4 The purpose of this survey is to monitor
capital flows, income, fees and royalties, and other services
transactions between foreign parent companies and their U.S. 
affiliates.  Data from this survey have been available since 1950. 


--------------------
\4 The FDI position of the United States is the cumulative value of
foreign parents' investment in their U.S.  affiliates.  It includes
only equity and debt funding provided by foreign parents, not funds
provided by other foreign firms or by U.S.  firms. 


BEA, CENSUS, AND BUREAU OF LABOR
STATISTICS (BLS) DATA-SHARING
PROJECTS
========================================================= Appendix III

The Foreign Direct Investment and International Financial Data
Improvements Act of 1990 (Public Law 101-533) authorized BEA to share
business-confidential data on FDIUS with Census and BLS, and Census
to share business-confidential data with BEA in order to improve the
quantity and quality of data on FDIUS.\1 In accordance with the 1990
act, BEA enterprise data has been linked with Census and BLS
establishment data and has generated more detailed information on the
characteristics and operations of U.S.  affiliates of foreign firms
in the United States than was previously available.  BEA, Census, and
BLS officials said there were opportunities for further collaboration
to improve the quality and quantity of data available on FDIUS, but
certain resource limitations and other factors related to protecting
business- confidential information may inhibit their fulfillment. 


--------------------
\1 The provisions providing for the exchange of data between BEA and
Census and for BLS access to BEA data are as follows:  section 5(a)
of Public Law 101-533 amended title 13 U.S.C., to allow BEA to obtain
access to Census' data; section 6(d) amended the International
Investment and Trade in Services Survey Act (Public Law 94-472, Oct. 
11, 1976, 22 U.S.C.  3104), to allow Census and BLS to obtain access
to BEA's data. 


   BEA-CENSUS DATA LINK
------------------------------------------------------- Appendix III:1

The BEA-Census data link project involves linking BEA's
business-confidential enterprise data on U.S.  affiliates of foreign
firms--collected at the 3-digit International Surveys Industry (ISI)
code level--with Census' business-confidential establishment data
collected at the 4-digit SIC level. 

Thus far, the data link project between BEA and Census has generated
data covering foreign-owned U.S.  establishments for 1987-91.  For
1987, both manufacturing and nonmanufacturing establishments were
covered and data were provided on the number, employment, payroll,
and shipments or sales of the foreign-owned establishments.  For
1988-91, only manufacturing establishments were covered, but more
data items were obtained--including data on the number, value added,
shipments, employment, total employee compensation, employee
benefits, hourly wage rates of production workers, cost of materials
and energy used, inventories by state of fabrication, and
expenditures on new plant and equipment of foreign-owned
establishments.  The data were obtained by matching enterprise data
collected in BEA's 1987 Benchmark and 1988-91 Annual Surveys of
Foreign Direct Investment in the United States to establishment data
from Census' 1987 Economic Censuses, 1987-91 Report of Organization
surveys, and 1988-91 Annual Survey of Manufacturers (ASM), as well as
establishment data Census obtains from administrative or other
statistical agencies.  The Census establishments that linked to BEA's
enterprises in the most recent BEA-Census data link (1991) accounted
for 98 percent of the employment by foreign-owned manufacturing firms
in the United States. 

The BEA-Census data link is a technically complex process requiring
both automated and manual procedures.\2 The following is a simplified
explanation of how the BEA-Census link is conducted.  Figure III.1
illustrates the process at a simplified level. 

   Figure III.1:  Simplified
   Illustration of the Process for
   Linking BEA Enterprise Data
   With Census and BLS
   Establishment Data

   (See figure in printed
   edition.)

   \ a Verifying matched cases
   includes developing and
   checking preliminary data
   tables to verify the accuracy
   of the linked data.

   (See figure in printed
   edition.)

   Source:  GAO analysis of data
   link process.

   (See figure in printed
   edition.)


--------------------
\2 The scope of linking and reconciling the data is reflected in the
numbers involved for an upcoming data link:  1992 BEA files covering
approximately 12,000 enterprises and 39,000 employer identification
numbers (EIN) are being matched against Census files covering 9.5
million establishments.  Preliminary indications are that about
105,000 of these establishments were at least 10-percent foreign
owned. 


      COMPUTERIZED LINK
----------------------------------------------------- Appendix III:1.1

The BEA-Census data link project begins when BEA sends Census a
computer tape containing micro-level data on foreign-owned
enterprises.  The data tape contains key information about the
enterprise, such as its name, address, and EIN.\3 The tape also
includes other descriptive items for the enterprise, such as the
number of its employees and its sales in dollars. 

Census then attempts to match by computer BEA's enterprise EINs with
EINs listed in Census' Standard Statistical Establishment List
(SSEL), a computerized list covering all U.S.  companies and their
establishments--about 9.5 million single and multi-unit companies.\4
The computerized EIN matching operation has three possible outcomes: 
(1) an enterprise links to one or more of Census' establishments, (2)
two or more enterprises link to one or more establishments,\5 or (3)
an enterprise does not link to any of Census' establishments.  For
those enterprises that do not link (outcome 3), Census and BEA
conduct further research.  For those enterprises that do link
(outcome 1 or 2), Census and BEA verify the accuracy of the matched
cases. 


--------------------
\3 The IRS assigns all enterprises an EIN for tax-related purposes. 
EINs are reported on both BEA and Census surveys; all companies are
required to use their EINs when filing federal and state payroll and
income taxes. 

\4 When a company consists of only one establishment, the company is
referred to as a single-unit company.  When a company has more than
one establishment, the company is referred to as a multi-unit
company. 

\5 This can occur when establishments have multiple foreign owners. 


      IDENTIFYING CASES THAT DID
      NOT LINK
----------------------------------------------------- Appendix III:1.2

Once the computerized link has been completed, Census must identify
those cases in which BEA's enterprise did not link to any of Census'
establishments.  This nonlinkage may have occurred because the
original EIN that BEA provided to Census for that enterprise on the
data tape was incorrect--perhaps the enterprise reported the EIN
incorrectly to BEA.  Whatever the reason, Census tries to identify
the correct EIN by researching the enterprise on the SSEL.  If Census
is unable to identify the correct EIN, Census forwards the case to
BEA for further research. 

The research at BEA often entails checking historical or archived
information in various BEA files to ensure that the enterprise is a
valid U.S.  affiliate of a foreign firm, i.e., the enterprise is at
least 10 percent foreign owned and the EIN for that enterprise in
BEA's files is valid.  If the enterprise is not a valid U.S. 
affiliate, the enterprise is eliminated from inclusion in the data
link.  If the enterprise is a valid U.S.  affiliate, BEA obtains the
necessary information to allow the enterprise to be matched correctly
to Census' establishments.  BEA then sends this information back to
Census.  Generally, the research at BEA on unmatched cases is carried
on concurrently as the project moves into the reconciliation phase. 
For the data link covering 1992, 243 cases were referred to BEA for
further research.  Depending on the research required, each case
referred can take up to 15 days to research, according to BEA
officials. 


      RECONCILIATION PROCESS
----------------------------------------------------- Appendix III:1.3

At this point the cases that did link--those in which a BEA
enterprise linked to at least one Census establishment--must be
reconciled.  A BEA official--who has been sworn in as a Census
agent--works with Census to help evaluate whether those cases that
linked were correctly matched and to reconcile them if they were not. 
The reconciliation process is very time-consuming and intensive.  For
example, for the data link covering 1992, about 1,700 cases were
reconciled because of data discrepancies.  According to BEA and
Census officials, the reconciliation process generally takes about 10
weeks to complete.  The process requires Census and BEA to compare
the employment count for a given BEA enterprise with the aggregate
employment count for the Census establishments that were linked to
the BEA enterprise.  If there is a large difference between the BEA
and Census employment counts--generally over 100 employees--Census
and BEA officials must research each case further.  To do so, Census
and BEA officials compare the data provided by BEA on each
enterprise's name, address or location, and employment with Census'
SSEL data. 

In general, Census and the BEA official are able to resolve
discrepancies between BEA and Census data by further researching the
cases, according to BEA and Census officials.  However, when the
discrepancy for some linked cases cannot be resolved, those cases
must be returned to BEA for further research.  The research conducted
on mismatched enterprises is similar to that previously described for
unmatched enterprises, except that BEA may contact the enterprise
directly to assure an accurate link; research may sometimes take up
to 90 days to complete. 

Once Census and BEA have reconciled and correctly matched BEA's
foreign-owned enterprises with Census' establishments, all of the
linked cases must be reverified.  Census and BEA officials again work
together to verify their judgement that linked cases have been
reconciled and correctly matched.\6 BEA also verifies the accuracy of
the linked data by generating preliminary tables to check both the
consistency with other data on FDIUS (both from BEA and from other
sources) and the internal consistency within the preliminary tables
themselves.  Developing and checking the tables usually takes BEA
about 2 weeks.  The data link is complete once the individual linked
cases and the table data have been verified.\7


--------------------
\6 However, BEA officials may not see cases that were based on data
Census obtained from IRS and therefore are unable to verify these
cases.  Restrictions on BEA access to IRS data are set forth in the
IRS regulations (26 CFR 301.6103(j)(1)-1) that implement the IRS
statutes. 

\7 For the 1988-91 data link projects using data from the ASM, an
additional step was necessary once the link with establishment data
from the SSEL was completed, but before tables were developed for
publication.  Data were extracted from the ASM for those
establishments that linked and reported on the ASM.  Since the ASM is
a sample survey, data for establishments that did link but did not
report on the ASM were imputed using industry averages for the
relevant data elements. 


      DISCLOSURE AVOIDANCE REVIEW
      AND TABLE DEVELOPMENT
      FOLLOWING DATA LINK
----------------------------------------------------- Appendix III:1.4

Developing and publishing tables covering the data generated by the
data link on an annual basis is also a joint Census-BEA project.  For
example, tables generated from the 1987 data link project provided
over 600 pages of data tables on FDIUS disaggregated by industry,
country, and state.  BEA designs and writes the computer programs for
the tables and generates the data with assistance from Census. 
Census performs disclosure avoidance review on each table to ensure
that no confidential data are disclosed;\9 in many cases, data in
tables must be suppressed before the tables can be published.  BEA
also checks the tables for their accuracy by comparing the table data
with other data on FDIUS.  The process to design the tables, generate
the data, and perform the necessary disclosure avoidance review, and
check the tables can take as long as 7 months.  According to Commerce
officials, the need to suppress certain data elements so as not to
compromise the confidentiality of the data is one of the problems
that the agencies face in making the more detailed, 4-digit SIC level
data available, because it limits the amount of data that can be
published. 


--------------------
\8 Disclosure avoidance review is the process of suppressing data
from publications so as to avoid disclosing confidential data. 

\9 Public Law 101-533 restricts the agencies from disclosing any data
that may reveal the identity of an individual respondent, be it an
enterprise or an establishment. 


   BEA-BLS DATA LINK
------------------------------------------------------- Appendix III:2

Like the BEA-Census data link project, the BEA-BLS data link project
was developed to obtain data on U.S.  affiliates of foreign firms at
a greater level of industry detail than is available from the BEA
data alone.  The BEA-BLS data link project has generated data on the
fourth-quarter employment and wages of foreign-owned establishments
in 1989, 1990, and 1991, as well as the occupational employment of
foreign-owned manufacturing establishments in 1989.  Data for the
first set of data link projects covering employment and wages were
derived by matching BEA's enterprise data from its Annual Survey of
Foreign Direct Investment in the United States with BLS'
establishment data from its Covered Employment and Wages (ES-202)
Program--covering approximately 6.5 million U.S.  establishments.\10
Data for the data link covering occupational employment were derived
by matching the linked manufacturing establishment data from 1989
with other 1989 establishment data from BLS' Occupational Employment
Statistics Survey.\11

The BEA-BLS data link process is very similar to that of the
BEA-Census data link project.  However, BLS reconciles and verifies
the BEA-BLS data link on its own, with input from BEA.  As with the
BEA-Census data link project, the BEA-BLS data link project is
designed to link BEA's 3-digit ISI data on enterprises with BLS'
4-digit SIC data on establishments.  The data link project begins
with BEA sending its data tape containing business-confidential
enterprise data and key identifiers (such as the enterprises' EIN,
name, address, employment, etc.) to BLS (see fig.  III.1).  BLS then
is to perform a computerized link of BEA's data with BLS'
business-confidential establishment data. 

Once BLS has generated a computerized link of the two data sets, it
attempts to verify and reconcile discrepancies in the data.  In
general, BLS will try to resolve these discrepancies on its own,
often using secondary sources such as the Directory of Corporate
Affiliations or Moody's Industrial Manual to help explain why
mismatches may have occurred and to identify cases that should have
matched.  However, when discrepancies cannot be easily explained, BLS
sometimes sends questions about unmatched or mismatched cases to BEA
for further research. 

BLS then is to verify that all cases included in the link have been
matched correctly and develop tables for data publication.  The
linked establishments from the 1991 employment and wages BEA-BLS data
link accounted for about 99 percent of the employment by all U.S. 
affiliates of foreign firms.  Like Census, BLS must perform
disclosure avoidance review on each table before the data are
published. 


--------------------
\10 The ES-202 is a federal-state cooperative program whereby data
are primarily derived from the quarterly tax reports submitted to
state employment security agencies by employers subject to state
unemployment insurance laws and federal agencies subject to the
Unemployment Compensation for Federal Employees program.  The states
provide the data directly to BLS. 

\11 The Occupational Employment Statistics Survey is a periodic mail
sample survey that state employment security agencies conduct of
nonfarm establishments to obtain wage and salary employment by
occupation.  These data are used to estimate total employment by
occupation for the nation, each state, and selected areas within
states. 


   OPPORTUNITIES EXIST TO EXPAND
   DATA LINK, BUT RESOURCES ARE
   LIMITED
------------------------------------------------------- Appendix III:3

According to Commerce officials, several opportunities exist for
improving FDIUS data by expanding the BEA-Census data link project. 
Specifically, opportunities exist to link BEA data with Census
product-level and export data, as well as with Census longitudinal\12
data on manufacturing establishments' operations.  However, budget
and resource constraints may limit the agencies' ability to pursue
these projects. 

According to BEA officials, the two agencies are currently evaluating
the possibility of linking BEA's enterprise data with product- and
product class-level data obtained by Census through its economic
censuses and ASM survey.\13 A link with product-level data would
enable Commerce to provide data on specific products or product
classes produced by foreign-owned establishments at a much greater
level of detail than either BEA's 3-digit ISI industry data or the
4-digit SIC industry data currently produced under the data link
project.  According to agency officials, one potential problem with
linking these highly detailed product-level data is that much of the
data would be business-confidential and would need to be suppressed. 
However, the product data would enable BEA to study with greater
accuracy and precision issues such as whether U.S.  affiliates of
foreign firms are targeting high-technology industries. 

Another opportunity exists for a data link with Census' exporter
database.  Census developed this database by matching information on
exports from the U.S.  Customs Service with individual establishments
listed on Census' SSEL register.  For 1987, Census was able to
attribute approximately 60 to 70 percent of all U.S.  exports to
establishments on Census' register.  Census is now constructing a
1992 exporter database that could potentially be linked to BEA's
enterprise data.  Such a link would generate much more detailed,
precise data on exports by type of product than BEA has because U.S. 
exports are reported to Customs according to a 10-digit schedule that
classifies commodities.\14 The more detailed data could help shed
light on how U.S.  affiliates of foreign firms contribute to U.S. 
exports. 

Commerce officials also told us they anticipate developing a link
between Census' Center for Economic Studies longitudinal database and
BEA data at some point in the future.  Such a link would allow
Commerce to analyze individual manufacturing establishments'
operations over time.  For example, Commerce could study changes in
establishments' employment, value added, shipments, etc., once the
establishments become foreign owned. 

Commerce officials stated that budget and staff constraints, as well
as the unavailability of funding, may limit the agencies' ability to
pursue these additional data link projects.  For example, BEA
officials emphasized that, to date, no funding has been allocated to
pursue a data link to Census' export or longitudinal databases. 


--------------------
\12 A longitudinal study would be one that follows the growth and
change of an individual or group over a period of years. 

\13 These data are based on Census' Numerical List of Manufactured
and Mineral Products, a classification system for those principal
products and services of the manufacturing and mining industries in
the United States.  The classification system is a 7-digit system
representing Census' extension of the 4-digit SIC system; the
economic censuses collect product data at the 7-digit level, while
ASM collects product class data at the 5-digit level. 

\14 The 10-digit schedule is known as "Schedule B," Statistical
Classification of Domestic and Foreign Commodities Exported from the
United States.  It is an extension of the 6-digit Harmonized
Commodity and Coding System, the internationally recognized
classification system for commodities. 


   DATA-SHARING RESTRICTIONS LIMIT
   EXCHANGE AND USE OF DATA BY
   AGENCIES
------------------------------------------------------- Appendix III:4

According to Commerce and Labor officials, certain restrictions and
other factors continue to limit the extent to which federal agencies
exchange and use FDIUS data on an ongoing basis.  Specifically,
restrictions on the use and disclosure of confidential data obtained
from IRS limit reconciliation and analysis of data generated from the
BEA-Census data link project.  In addition, various factors restrict
BLS' data sharing with BEA.  While Public Law 101-533 provides a
mechanism for agencies to resolve data access issues, BEA and BLS
have not used this mechanism to resolve issues related to BEA access
to BLS' business-confidential establishment data. 


      RESTRICTIONS ON IRS DATA
      SHARING
----------------------------------------------------- Appendix III:4.1

Although both Census and BEA are permitted to request and obtain
confidential information directly from IRS, restrictions on the use
and disclosure of such data prevent either BEA or Census from sharing
the data with each other.  According to BEA officials, these
restrictions prevent BEA from comparing, analyzing, or verifying data
in its own databases with data on individual establishments that
Census obtains from IRS.  Section 401 (a) of title 13 U.S.C.  states
that Census may share with BEA only data collected directly from
respondents by the Census Bureau itself.\15 In addition, IRS
regulations (26 CFR 301.6103(j)(1)-1), which describe the projects
for which access to IRS data is permitted, do not specifically
mention the BEA-Census FDI link project.  IRS has stated that BEA may
not have access to the IRS data on the FDI data link files until it
has revised its regulations to specifically mention the data link
project, according to BEA officials.  Therefore, Census cannot
disclose to BEA any data Census has obtained directly from IRS until
Census has verified the data through its own surveys. 

According to BEA officials, Census and IRS are currently developing
an agreement to modify IRS' implementing regulations for title 26 of
the United States Code so that BEA staff who are sworn Census
employees may be granted access to IRS data contained in Census'
files.\16 The agencies do not anticipate that this action will
require any legislative changes in either title 13 or title 26 of the
United States Code. 


--------------------
\15 Title 13 of the United States Code, among other things, governs
the collection and publication of statistics that the Census Bureau
gathers. 

\16 Title 26 of the United States Code is the U.S.  tax code
provision that governs the disclosure of IRS taxpayer information. 


      RESTRICTIONS ON BLS DATA
      SHARING
----------------------------------------------------- Appendix III:4.2

BEA would like access to BLS' business-confidential establishment
data to evaluate differences between the BLS' and Census'
establishment data bases.\17 However, according to BLS officials, BLS
has pledged not to disclose any of the business-confidential
employment and wage data it obtains from the states under cooperative
agreements.  BLS officials told us that BLS would have to get
permission from each of the states before any such data could be
released to BEA, or any other agency.  Although Public Law 101-533
neither prohibits nor requires BEA access to BLS'
business-confidential establishment data, section 8(e)(2) of Public
Law 101-533 states that the Director of the Office of Management and
Budget shall be responsible for resolving questions on access to
information with regard to any exchange of information between BEA
and BLS.  At this time, the agencies have not requested mediation on
these issues from OMB. 



(See figure in printed edition.)Appendix IV

--------------------
\17 Such differences may be due to several factors, including
differences related to how SIC industry codes are assigned to
establishments, and when the SIC codes are assigned and updated. 


COMMENTS FROM THE DEPARTMENT OF
COMMERCE
========================================================= Appendix III


OBJECTIVES, SCOPE, AND METHODOLOGY
=========================================================== Appendix V

The Foreign Direct Investment and International Financial Data
Improvements Act of 1990 directs us to analyze and report on
Commerce's first three annual reports on FDIUS and review government
efforts to improve the quality of FDIUS data.  Specifically, our
objectives were to (1) assess the extent to which Commerce's second
and third reports--issued in 1993 and 1995--fulfilled the
requirements of the 1990 act and addressed the recommendations in our
1992 review;\1 (2) review the process by which federal agencies
collect FDI data; (3) review the status and processes of the data
exchanges, or links, initiated by the 1990 act between the Commerce
Department's Bureau of Economic Analysis (BEA) and its Bureau of the
Census and between BEA and the Labor Department's Bureau of Labor
Statistics (BLS); and (4) evaluate the extent to which implementation
of the act has brought about the intended improvements in public
information on FDI in the United States. 

To assess how well Commerce fulfilled the reporting requirements of
the 1990 act, we reviewed the 1993 and 1995 reports, with specific
attention to Commerce's coverage of the data requirements of the act,
and to the overall quality of Commerce's analysis of the potential
effects of FDI on the U.S.  economy.  In addition, we evaluated the
extent to which the 1993 and 1995 reports responded to the
recommendations in our 1992 report. 

We used standard economic principles in our review and evaluation of
the Commerce reports, with special attention to the chapters relating
to the implications of FDIUS for U.S.  trade, technology transfer,
tax payment, employment, and banking issues.  We relied on internal
economists as well as an outside economist with expertise in FDIUS
issues to carry out this evaluation.  We also consulted Commerce
officials frequently in the conduct of our review to ensure
consideration of their views in our findings. 

In evaluating Commerce's reports, we considered the following
factors: 

  Organizational structure:  We considered whether the organizational
     structure of the reports as a whole and individual chapters (1)
     facilitated discussion of key FDIUS issues, (2) presented
     principal findings in a logical, consistent format, and (3) used
     tables and graphics effectively to highlight the trends in FDIUS
     and describe the characteristics of U.S.  affiliates of
     foreign-owned firms. 

  Sufficiency of evidence for principal findings:  To evaluate the
     sufficiency of Commerce's support for its principal findings, we
     considered whether the reports (1) presented convincing evidence
     to establish causal relationships, (2) identified limitations in
     the data available or used, (3) used appropriate analytical
     techniques to address specific questions, and (4) qualified
     conclusions where appropriate. 

  Coverage of the data requirements of the 1990 act:  We reviewed the
     reports to determine the extent to which they included
     discussion of the data requirements of the 1990 act.  To the
     extent of available data, the act requires Commerce to compare
     business enterprises controlled by foreign persons with other
     business enterprises in the United States with respect to
     employment, market share, value added, productivity, research
     and development, exports, imports, profitability, taxes paid,
     and investment incentives and services provided by state and
     local governments, including quasi-government entities. 

  Coverage of specific industry sectors:  We assessed the extent to
     which the reports included discussion of most of the major
     industry sectors identified in the SIC system at the 2-digit and
     3-digit levels,\2 as well as specific industries with higher
     levels of foreign direct investment and/or those that involve
     the use or production of advanced technologies.  Where
     appropriate, we also evaluated Commerce's presentation of the
     4-digit SIC data made available through the BEA-Census and
     BEA-BLS data links. 

  Coverage of major investing countries:  We determined the extent to
     which the reports included coverage of the countries with the
     highest shares of direct investment in the United States, which
     included Japan, United Kingdom, Canada, Germany, France, and
     Switzerland in 1993. 

  Use of relevant outside studies:  We evaluated the extent to which
     the reports included reference to current FDIUS publications by
     major academic or research institutions and to economists with
     recognized expertise in FDI issues. 

To identify and obtain information on significant FDI research and
policy developments, we reviewed current literature on FDIUS and
attended conferences where researchers presented the results of
recent FDIUS studies.  In addition, we consulted with outside experts
in government and the research communities to obtain their
perspectives on the Commerce reports and on our principal findings. 
To obtain information on federal government FDI data collection
activities, we interviewed officials from BEA, Census, and BLS, and
obtained documents outlining their data collection processes, as well
as current examples of relevant survey questionnaires.  We also
consulted past GAO and Commerce reports which discussed federal
government FDI data collection efforts outside of the Departments of
Commerce and Labor. 

To review the status and processes of the interagency data exchanges
required by the 1990 act, we interviewed officials with
responsibility for such activities in BEA, Census, and BLS.  These
officials provided us with detailed verbal and documentary
descriptions of the steps required to perform the data exchanges.  In
addition, in June 1995 we observed a demonstration of the data link
reconciliation process at the Census Bureau. 

To evaluate the extent to which the implementation of the 1990 act
has led to improvements in FDIUS data, we considered factors such as
the contribution of the BEA-Census and BEA-BLS data exchange
programs, the overall quality and coverage of the Commerce Department
reports since 1991, and Commerce's changing approach to fulfilling
its reporting requirements under the act. 

In addition to our usual quality assurance procedures, we requested
an outside research economist with expertise in FDIUS issues to
review a draft of the report and provide comments.  We have
incorporated his suggestions where appropriate. 

We performed our review in Washington, D.C., from January 1995 to
August 1995 in accordance with generally accepted government auditing
standards. 


--------------------
\1 We reviewed Commerce's 1991 FDIUS report in Foreign Direct
Investment:  Assessment of Commerce's Annual Report and Data
Improvement Efforts (GAO/NSIAD-92-107, Mar.  18, 1992). 

\2 The 2-digit SIC industry groups include agriculture, forestry, and
fishing; mining; construction; manufacturing; transportation,
communications, electric, gas, and sanitary services; wholesale
trade; retail trade; finance, insurance, and real estate; services;
and public administration.  The 3-digit SIC industry groups are
further disaggregations of the 2-digit groups; there are over 400
such groups. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================== Appendix VI

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C. 

Curtis F.  Turnbow, Assistant Director
Sara B.  Denman, Senior Evaluator
Carolyn M.  Black-Bagdoyan, Evaluator
Jane-yu H.  Li, Senior Economist
Martin de Alteriis, Social Science Analyst
Elizabeth J.  Sirois, Adviser
Rona Mendelsohn, Evaluator (Communications Analyst)

OFFICE OF THE CHIEF ECONOMIST,
WASHINGTON, D.C. 

Loren Yager, Assistant Director

OFFICE OF GENERAL COUNSEL,
WASHINGTON, D.C. 

Herbert I.  Dunn, Senior Attorney
