Improving Business Statistics Through Interagency Data Sharing: Summary of a Workshop (2006)

This page intially left blank

Robert P. Parker

Consultant on Federal Statistics

During the past 20 years, the National Research Council of the National Academies has prepared several reports on data sharing and related issues, such as privacy and confidentiality. However, as noted in the discussion here of each of these reports, the topics covered by the Workshop on the Benefits of Interagency Business Data Sharing differ from those in the previous reports. The workshop focused on data and information on businesses, not on individuals, and their use for statistical, not research, purposes. In addition, the workshop focused on sharing by the Bureau of Labor Statistics (BLS), the Bureau of Economic Analysis (BEA), and the Census Bureau; these three statistical agencies are authorized to share data under provision of Subpart B of the Confidential Information Protection and Statistical Efficiency Act (CIPSEA) as enacted in December 2002.

This chapter reviews the history of CIPSEA and of efforts by the Office of Management and Budget (OMB) for related legislation that would expand data sharing by changing the provisions of the Tax Reform Act of 1976, which allow the Internal Revenue Service (IRS) to share tax information with selected statistical agencies. CIPSEA provides statutory confidentiality protection for data collected by statistical agencies for statistical purposes (Subpart A), and it permits identifiable business records to be shared for statistical purposes by BLS, BEA, and the Census Bureau (Subpart B). CIPSEA, however, did not change the provisions of Internal Revenue Code Section 6103, which precludes these agencies from sharing tax return information for statistical purposes. Such additional sharing would allow the agencies to develop a single business register for all of their

economic surveys and programs and to use information from the register to improve the quality of their surveys. It would also create additional opportunities to create new statistical programs from existing data.

There have been efforts by several administrations to change Section 6103 to allow the sharing of basic identifying information from IRS records for statistical purposes. This chapter also reviews the positions on this change by the U.S. Government Accountability Office (GAO), the Department of the Treasury, the IRS, and the Joint Committee on Taxation (JCT), as well as the positions of businesses.

Some of the major ongoing data-sharing efforts by the three agencies are reviewed in this chapter, which covers efforts allowed under Section 6103, the International Investment and Trade in Services Act of 1990, and BEA, BLS, and the Census Bureau confidentiality restrictions.

In this chapter, “data sharing” is defined as the exchange of information collected from businesses and individuals or reported to the IRS in identifiable form for statistical purposes. For other key terms, the chapter follows the definitions in CIPSEA. Business data are “operating information and financial data and related information about businesses, tax exempt organizations, and government entities.” Identifiable form means information “that permits the identity of the respondent to whom the information applies to be reasonably inferred by either direct or indirect means.” Statistical purposes are “the description, estimation, or analysis of the characteristics of groups, without identifying the individuals or organizations that comprise such groups.” They also include methods and procedures related to the “collection, compilation, processing, or analysis” of data about these groups and the development of related “measurement methods, models, statistical classifications, or sampling frames.” Using definitions from the report of the Privacy Protection Study Commission (1977), the term “research” is used to refer to “any systematic, objective process designed to obtain new knowledge, regardless of whether it is ‘pure’ (aimed at deriving general principles) or ‘applied’ (aimed at solving a specific problem or at determining policy).” The term “statistics” in this chapter refers “both to the data obtained through enumeration and measurement and to the use of mathematical methods for dealing with data so obtained.”

Most of the previous reports of the National Research Council on data sharing and related topics touched on issues important to the discussion of changing Section 6103, such as the trade-off between greater access and reduced confidentiality. In general, the findings and recommendations of these reports did not focus on sharing of data and information on busi-

nesses or on data sharing for statistical purposes. Although the most recent of these reports was completed after the enactment of CIPSEA, there was little discussion of the data-sharing part of the act, largely because CIPSEA was limited to the data sharing of business data for statistical purposes, not individual data and not for research purposes. In this section, each of these reports is reviewed in terms of its relevance to business data sharing.

In Sharing Research Data (National Research Council, 1985), the focus is mostly on what researchers should do. It thus has little relevance to sharing of business data for statistical purposes. The report does include a section that identifies several benefits related to sharing business data. These benefits are the verification of results, improvements in measurement and data collection methods, and the development of theoretical knowledge. In addition, one of the papers prepared for the conference that led to the report notes the creation of new data sets as a benefit of data sharing. However, there is no mention of the benefits of data sharing to statistical agencies, such as the ability to develop common business registers.

Sharing Research Data cites three earlier reports related to data sharing: (1) Sharing of Social Science Research Data: An Exploratory Conference Convened by the Committee on National Statistics (National Research Council, 1980); (2) Setting Statistical Priorities (National Research Council, 1976); and (3) Personal Privacy in an Information Society (Privacy Protection Study Commission, 1977). The first report deals primarily with health data, and the second does not deal at all with data sharing. The Privacy Protection Study Commission report is discussed later in this chapter.

Private Lives and Public Policies: Confidentiality and Accessibility of Government Statistics was published in 1993 (National Research Council, 1993). The Panel on Confidentiality and Data Access, jointly formed by the Committee on National Statistics and the Social Science Research Council, developed recommendations to assist statistical agencies in their stewardship of data for policy decisions and research. The panel met between November 1989 and January 1992 and used the results of related workshops: the Longitudinal Retirement Workshop (September 1987), Confidentiality of and Access to Doctorate Records (November 1988), and Confidentiality and Access to National Center for Education Statistics (January 1991).

The panel sought to take into account the trade-offs between data protection and data access and made recommendations to “enhance data access without decreasing data protection and increase data protection without reducing data access” (p. 19). The panel decided to focus primarily on major federal statistical programs and to cover “organizational,” or business, data as well as individual data. This expansion reflected input from

statistical agencies, the experience of panel members, and current events. As a result, the report covers major issues of data protection and access to business data; it includes a chapter on a framework for the treatment of these data (Chapter 7) and covers businesses in other chapters that deal with legislation (especially Chapter 5).

With regard to business data, the panel presents several relevant findings and recommendations. The Executive Summary includes a section on “Barriers to Data Sharing Within Government,” in which the panel reported (p. 6):

The panel recommends that barriers to sharing data for both persons and businesses for statistical purposes should be removed subject to strict controls to protect confidentiality. Recommendation 4.1 includes a statement that “additional data sharing should only be undertaken in those instances in which the procedures for collecting the data comply with the panel’s recommendations for informed consent or notification” (p. 6). Recommendation 5.1(f) states that ‘’a provision that permits data sharing for statistical purposes under controlled conditions be included in the consistent set of statutes and regulations governing the maintenance of federal statistical records” (p. 6). Recommendation 7.4 reflects the panel’s proposal that interagency sharing of data for statistical purposes should include the sharing of lists of businesses by federal and state agencies. Recommendation 7.5 asks for new legislation to expand confidentiality to records collected by all statistical agencies (p. 226); some of this recommendation was subsequently accomplished by CIPSEA.

Although this report cites several examples of reduced sharing as a result of the provisions of Section 6103, there is no explicit recommendation to amend this act. Nevertheless, the panel clearly expresses its views in the following section title in Chapter 7: “Inability to Share Business Lists: An Embarrassment to the Federal Statistical System.”

In addition, the report includes several important discussions relating to sharing business data that appear to conflict with the current IRS interpretation of Section 6103. In a discussion on the Census Bureau use of tax data, it states (p. 192):

The report also notes that there is little or no sharing of business data collected by the Census Bureau (p. 192):

Improving Access to and Confidentiality of Research Data: Report of a Workshop was published in 2000 (National Research Council, 2000). This workshop report does not discuss sharing business data. The report states that a central objective of the workshop was to review the benefits and risks associated with public-use research data files and to explore alternative procedures for restricting access to sensitive data, especially longitudinal survey data that have been linked to administrative records. It focuses on microdata records for research purposes. The workshop did not produce recommendations or specific findings.

Principles and Practices for a Federal Statistical Agency, Third Edition, was published in 2005 (National Research Council, 2005a). Although the report deals with a wide range of activities by statistical agencies, it lists the sharing of microdata records as a “way to improve data quality and develop new kinds of information . . . for statistical agencies that collect similar information” (p. 26). It notes that the sharing of business data would “make it possible to evaluate reporting errors and the completeness of coverage of business firms in different surveys,” and that “such sharing would also make it possible to develop more useful and accurate statistics on the nation’s economy while decreasing the reporting burden on business data providers” (p. 26).

The report discusses the enactment of CIPSEA, noting that the first formal proposal for data sharing under CIPSEA, which involved matching data from BEA’s international investment surveys with data from the Census Bureau’s Survey of Industrial Research and Development conducted for the National Science Foundation, was announced in June 2003. (Additional details on this match are discussed later in this chapter.)

Expanding Access to Research Data: Reconciling Risks and Opportunities was published in 2005 by the Panel on Data Access for Research Purposes, which was convened at the request of the National Institute on Aging (National Research Council, 2005b). The panel report, which is built largely on the earlier efforts discussed above, primarily focuses on data for research purposes. The Introduction states: “Thus, there are questions about how to provide researchers—inside and outside government— access to data that can both inform public policy and protect the privacy of respondents and the confidential nature of the information they provide” (p. 8). Consequently, the report is almost exclusively research oriented although it does cover some research using business records. It focuses on microdata research and notes the policy and cost benefits for data linking and sharing. Although not an explicit recommendation, the report does note the need to revise Section 6103, stating (p. 35):

The final chapter of the report makes 19 recommendations covering such areas as documentation, planning for and expanding access to public-use files and research data centers, informed consent, and safeguarding confidentiality. None of the recommendations in these areas is directly relevant to the sharing of business records.

In a section on the legal environment, a discussion of CIPSEA points to the need for changing Section 6103. The report notes (p. 23):

The report discusses policy research but limits the discussion to the use of microdata. It states: “Detailed microdata permit in-depth analyses of socioeconomic trends and their antecedents and consequences” (p. 40) and continues with “Detailed microdata are also needed for modeling economic decisions and other kinds of social behavior” (p. 41). Examples of such research noted in the report include research to model savings behavior and firm behavior on pollution abatement. It also includes a section on linking survey and administrative record data, but examples are limited and omit any discussion of BEA’s foreign direct investment data. It reports on the 1973 Current Population Survey–Social Security

Administration–IRS (CPS–SSA–IRS) “exact-match file” that was the basis for a major dynamic microsimulation model of social welfare policies and retirement income and for an analysis of the quality of income reporting in the March CPS. The report does not cite the continuing use by the Census Bureau of linkages of survey data and administrative records to assess and improve data quality. It also does not report that BEA and IRS have used the aggregate results of this match program to estimate the extent of nonfiling by unincorporated enterprises. In fact, BEA continues to fund an updated version of this match, but without the SSA records.

Among the papers prepared for this panel, “Privacy, Confidentiality, and Data Sharing” by David McMillen is the most relevant to sharing of business data. It provides an excellent history of CIPSEA, tracing its most recent roots to the Clinton administration’s efforts to enact confidentiality and data-sharing legislation in 1995. (See additional details in the next section.) McMillen also identifies several examples of sharing both business and individual records. For matching of information collected under a mandatory authority, he strongly sees the need to notify respondents when the data they report are to be linked to IRS records or made available to another agency. The report indicates some discomfort with this part of McMillen’s views and does not incorporate them into the panel’s recommendations.

To provide an understanding of the changes incorporated in 1976 into Section 6103, it is important to first review the developments that led to these changes. (Most of the following discussion is based on a 1983 paper by Wilson and Smith and various editions of the Census Bureau series on the history of the economic censuses.) The most important change was that, prior to the Tax Reform Act of 1976, the president had the authority to issue executive orders permitting access to tax records. The revised Section 6103 limited access to tax records to specific federal and state and local government agencies and only for specific types of tax information. Under the previously issued executive orders, access for statistical purposes had been granted to the Commerce Department, covering both the Census Bureau and the predecessor agencies to BEA (1944 E.O. 9499 and 1961 E.O. 10911), the Securities and Exchange Commission and Federal Trade Commission (FTC) (1959 E.O. 10814 and 1961 E.O. 10908), and the Department of Agriculture (1973 E.O. 11697 and 11709). Under the Tax Reform Act of 1976, access by the Commerce Department was continued, but BEA’s access was limited to selected records of corporations, and access by the FTC to conduct the Quarterly Financial Report was continued.

Access by the Department of Agriculture was not continued, although it was restored in the 1990s, when the Census of Agriculture was transferred from the Census Bureau to the Department of Agriculture. In addition to changes in access, the 1976 version of Section 6103 included stricter disclosure rules.

According to Personal Privacy in an Information Society, a report by the Privacy Protection Study Commission, a commission mandated by the Privacy Act of 1974, the changes resulted from an increase in disclosure by the IRS during the first half of the 20th century (Privacy Protection Study Commission, 1977). Most of these new uses were authorized without any significant public discussion and, when made public, were justified by the federal and state agency recipients of the data as essential to an essential government function. The commission stated (p. 27 of Appendix 2):

This report had been limited in scope to individually identifiable information about individuals, so the commission did not study issues regarding disclosure of information about business entities. Nevertheless, the commission generally agreed with the limited access to tax return information included in Section 6103 in the Tax Reform Act of 1976. (The final report was published in 1977, but the commission had provided its recommendations to Congress in June 1976.)

In a 1979 study, GAO reported on the status of efforts of the Census Bureau to create a centralized business listing, or business register, that would be used by the Census Bureau and other statistical agencies to select samples for various surveys (U.S. General Accounting Office, 1979). By selecting samples from the same register, the list would improve the comparability of the data collected in these surveys regardless of the collecting agency and would reduce duplication in the construction of these

lists. The development of such a list had been recommended by various committees and commissions dating back to 1937. In 1968, OMB designated the Census Bureau as the focal agency to develop and maintain the list data. From 1972 to 1978, Congress provided the Census Bureau almost $16 million to develop the list. In 1975, the list became operational and was used by the Census Bureau for the 1977 economic census and current economic survey programs. However, the list could not be shared with other agencies, because the Census Bureau had yet to propose specific legislation to change Title 13 of the U.S. Code to allow it to share the list with other statistical agencies.

The enactment of the Tax Reform Act of 1976 and replacement of previous executive orders with Section 6103 had a major impact on this effort for the centralized business register. Prior to the act, it had been thought that only Title 13 needed to be changed to allow for sharing of the list. However, the act made it clear that because of the extensive use of tax return information by the Census Bureau (as discussed below), it also would be necessary to change Section 6103. In December 1978, the secretary of the treasury informed the secretary of commerce that the Treasury Department would not support legislation to change Section 6103. The secretary stated the reasons for this opposition as follows (p. 56):

After the unsuccessful effort by the Census Bureau to change Title 13 and Section 6103 to allow it to share its business register with other statistical agencies, OMB sought legislation to allow more extensive data sharing, including the sharing of data on individuals. The new effort would include allowing for sharing by more statistical agencies, but it also would provide statutory protection for data collected by all statistical agencies similar to the protection afforded to the Census Bureau by Title 13. Some agencies had no such supporting legislation, while others, under certain circumstances, were to turn over data collected for statistical purposes to other agencies for nonstatistical purposes. Thus, the proposed legislation would protect data collected by all statistical agencies for statistical purposes and allow these agencies to share these data with other statistical agencies.

The renewed effort to change Section 6103 began in 1983 when the administration sought to have the Congress enact so-called statistical enclave legislation, which would provide a single set of confidentiality policies for all federal agencies and their components that collect data for statistical purposes. This legislation, unlike CIPSEA, would have allowed access to tax records by all statistical agencies and included records of both individuals and businesses. As described by Wilson and Smith (1983), this legislation was to provide “(a) a statutory basis for the traditional promise of confidentiality long given respondents to statistical collections, and (b) restricted sharing of individually identifiable records (‘protected statistical files’) for exclusively statistical agencies (‘protected statistical centers’) whose confidential records are provided statutory protection under this legislation” (p. 601). The authors express their personal agreement with the portion of the bill that would ensure that data collected for statistical purposes will be used only for statistical purposes. They also report on the objections of the Treasury Department to increasing access to individually identifiable data for statistical purposes. The authors report that “this aspect of the draft Bill is unacceptable in the opinion of the Department of Treasury because, among other things, it might greatly expand the dissemination of tax return information, attenuate the ability to control such dissemination, and provide insufficient safeguards to prevent further disclosure and erosion of confidentiality.” Interestingly, the authors also report that “there also appears to be no provision in the Bill that would aid the research work needed to manage tax policy or tax administration situations. In particular, no data-sharing arrangements for IRS statistical purposes are proposed. All the ‘sharing’ is in the other direction.” It is not clear whether this is the author’s objection, the department’s, or both.

The authors also make it appear that the basis for the Treasury Department’s opposition was that “even though controls may be provided as to redisclosure of information, if the perception is created that tax return information is widely available to others, that perception alone could be very destructive of public confidence in the security and privacy of information provided under the tax administration system, and ultimately destructive of voluntary compliance” (p. 600).

The next, and most recent, effort by OMB to increase data sharing and change Section 6103 began in 1995, when the Clinton administration sent to Congress a new set of legislation proposals. This legislation was introduced on a bipartisan basis in the House of Representatives in early 1996.

Under the legislation, eight federal agencies were designated as statistical data centers: BEA, the Census Bureau, BLS, the National Agricultural Statistics Service, the National Center for Education Statistics, the National Center for Health Statistics, the Energy Consumption Division of the Energy Information Administration, and the Science Resources Statistics Division of the National Science Foundation. Enactment of this legislation would have improved the efficiency of the federal statistical system, reduced reporting burden on the public, and improved the quality and usefulness of the federal statistics for economic and social policy decisions.

A key feature of the proposed legislation required that data or information acquired by an agency for purely statistical purposes could be used only for statistical purposes and could not be shared in identifiable form for any other purpose without the informed consent of the respondent. This proposal, now the Statistical Efficiency Act of 1999 (H.R. 2885), was passed by the House in October 1999. It increased statutory protections for the confidentiality of statistical data and permitted sharing of data for statistical purposes among designated agencies. The proposal also called for companion legislation to make complementary changes to provisions set forth in the Statistical Use section of the Internal Revenue Code. According to OMB, the complementary proposal was endorsed by the Treasury Department.

To meet congressional concerns about the types of data to be exchanged under this proposed complementary legislation, the administration proposed a revised version of the bill. Under this version, data sharing would be restricted to BEA, BLS, and the Census Bureau, and only data on businesses would be shared. Finally, at the end of 2002, Congress passed the Confidential Information Protection and Statistical Efficiency Act (CIPSEA) of 2002 as Title V of the E-Government Act (Public Law 107–347). This legislation established a uniform set of safeguards to protect the confidentiality of individually identifiable information acquired from the public for statistical purposes, as well as consistently strong criminal penalties for inappropriate disclosure of such information. The legislation reaffirmed that pledges of confidentiality would be honored and gave additional weight and stature to policies that statistical agencies have pursued for decades, assuring respondents who provide statistical information that their responses will be held in confidence and will not be used against them in any government action. CIPSEA also authorizes the sharing of business data among the BEA, BLS, and the Census Bureau. The companion legislative proposal, which would make complementary changes on access to tax return data by these agencies, has been endorsed by the Treasury Department and submitted to Congress; the proposal has not been enacted.

In 2002, Congress passed CIPSEA. The following is the summary of the legislation from Principles and Practices for a Federal Statistical Agency, Third Edition (National Research Council, 2005a, pp. 62-63):

Since the enactment of CIPSEA, OMB has been working again to get the companion legislation to change Section 6103 enacted by Congress. In Statistical Programs of the U.S. Government: Fiscal Year 2005, OMB reported on the latest developments in the effort to change Section 6103. In a section on “Statistical Confidentiality and Data Sharing,” OMB reported the following (p. 44):

This section describes the public positions of various organizations with an interest in data sharing, particularly with tax return records. (This section excludes OMB, which as discussed above, has consistently supported changing the tax laws to permit more data sharing.) The views of GAO, the Council of Economic Advisers, the Department of the Treasury and the Joint Committee on Taxation are presented. It also includes the views of a special study commissioned by the Administrative Conference of the United States.

In response to a congressional request, the administrative conference in 1975 commissioned a team of tax experts headed by Charles Davenport to study the operations of the IRS. The result was the Report on Administrative Procedures of the Internal Revenue Service, October 1975, to the Administrative Conference of the United States, (Davenport, 1976) which included a section on tax return confidentiality. This section reviewed the history and rationale for tax return access for statistical purposes. The report concluded that “it appears that the use of tax data by Census, though not consistent with revenue administration, is a use which can be considered beneficial and is one which does not appear have any undesirable side effects” (p. 880). The study came to a similar a conclusion for BEA but not for the Statistical Research Service (SRS) of the Department of Agricul-

ture. The key distinguishing factor for the commission was that Census and BEA were strictly statistical agencies and did not engage in activities related to other functions of the department. In contrast, the commission determined that the statistics collected by SRS were “for policy making by the agency of which it is a part” (p. 886).

In 1979, GAO issued a report on recent developments in the Census Bureau’s efforts to create a centralized business register for use by other statistical agencies (U.S. General Accounting Office, 1979). GAO, which supports changing Section 6103, summarized its findings as follows (title page):

In 1998, GAO testified before Congress on Statistical Agencies: Proposed Consolidation and Data- Sharing Legislation (U.S. General Accounting Office, 1998). The testimony was about the data-sharing legislation submitted in 1996 and 1997 that would permit limited sharing of data among designated statistical agencies for statistical purposes, subject to procedural safeguards. The testimony included the following statement of GAO’s position (pp. 5-6):

In 2001, GAO issued Record Linkage and Privacy: Issues in Creating New Federal Research and Statistical Information (U.S. General Accounting Office, 2001). This report reviews the background of the key issues related to data sharing. However, the focus on the report was “on linkage projects that involve person-specific data, are conducted under federal auspices (or with federal funding), and produce new research or statistical information” (p. 10). The report did not cover business data sharing.

In the 2002 Economic Report of the President, the Council of Economic Advisers noted the critical need for reliable data and that data sharing would increase their quality. This excerpt notes the submission of proposals in 1999 and that the Administration would continue to work for their passage (p. 25):

Section 3802 of the Internal Revenue Service Restructuring and Reform Act of 1998 requires the secretary of the treasury and the JCT to conduct separate studies of the scope and use of provisions regarding taxpayer confidentiality and to report the findings of such a study, together with any recommendations deemed appropriate to Congress. The JCT published its report on January 28, 2000; the Office of Tax Policy of the Department of the Treasury submitted its report on October 2, 2000.

The JCT and the Treasury Department disagree on changing Section 6103 to permit expanded data sharing. The JCT recommended that “new access to returns and return information should not be provided unless the requesting agency can establish a compelling need for the disclosure that clearly outweighs the privacy interests of the taxpayer” (p. 196). (See below for information on business taxpayer views.) The JCT report did

not explain how to determine both the pros and cons of such a decision, nor did it appear to fully accept the need for the existing access. (Both the JCT and the Treasury Department recommended dropping the Federal Trade Commission from access under Section 6103.)

The Treasury Department recommended “the disclosure authority of Section 6103(j) should be expanded to additional specified statistical agencies, and such agencies should be permitted, upon prior Treasury approval, to share IRS data with each other.” It also specified the agencies to include and that the change to Section 6103 should cover both individual and business records.

In the 2001-2005 Statistical Programs of the United States Government, OMB reports support for the companion legislation that would change Section 6103.

As noted above, the JCT identified the key issues in determining whether to change Section 6103 when it wrote “new access to returns and return information should not be provided unless the requesting agency can establish a compelling need for the disclosure that clearly outweighs the privacy interests of the taxpayer” (p. 196). Although there is a substantial amount of research on individuals’ views on confidentiality and data sharing, much of it conducted for the decennial censuses, there is limited information on the views of businesses. The information that is available seems to indicate that business taxpayers are willing to allow more access to statistical agencies for some types of tax return information.

Private business economists have actively supported data sharing. In 1996, Maurine Haver, president of Haver Analytics and chair of the Statistics Committee of the National Association for Business Economics (NABE), testified before the House Subcommittee on Government Management, Information and Technology. In her testimony on consolidating the three major economic statistical agencies (H.R. 2521), she expressed support for the inclusion in that bill of provisions to provide for data sharing among these agencies, as it would allow for the creation of a single business register. In 2001, Richard Berner, the president of NABE and managing director and chief U.S. economist, Morgan Stanley, Inc., testified before the House Subcommittee on the Census Committee on Government Reform on the Statistical Efficiency Act of 1999. In his testimony, Berner reported on NABE’s support for the reintroduction of this act,

which had been passed unanimously by the House in the previous Congress. He testified “NABE believes that our national data collection efforts should be as efficient as possible. To that end, we believe that Congress should mandate ‘data sharing’ among the agencies, solely for statistical purposes.” He noted that existing confidentiality statutes are barriers to such data sharing because “they virtually guarantee duplication of effort and inconsistencies among related data sets collected by the affected agencies. Moreover, they prevent agencies from undertaking new analyses that could improve the information available to policy makers. This is not a cost-effective way to run any business—either public or private.”

A 2001 article called “Business Perceptions of Confidentiality” reported on the results of a survey conducted by the Urban Institute on these business perceptions (Greenia, Lane, and Jensen, 2001). Of particular interest to the issue of privacy interests is the response to a set of questions about types of data that businesses view as very sensitive. The survey results showed that less than 5 percent of respondents thought their name and address and industrial activity were very sensitive, but 85 percent of respondents were very sensitive to data about their employees. Between 50 and 75 percent of the respondents thought their financial data were very sensitive.

Among the many examples of interagency sharing of business data, the oldest and most extensive is the use of tax information by the Census Bureau to support its economic statistics programs. With the cooperation of the IRS in modifying questions on tax returns and timely delivery of tax records, the Census Bureau has been able to reduce reporting burden in the quinquennial economic censuses for small businesses, to improve the quality of key data from these censuses, and to develop new statistical programs. This section discusses this sharing as well as other selected examples; it excludes the County Business Patterns report (the Census Bureau, IRS, and SSA) and the sharing of BLS quarterly unemployment insurance data with BEA.

As reported in various editions of the history of the economic censuses, the extensive use of tax return records, from both the IRS and the SSA, began with the 1954 economic censuses. In previous censuses, field enumeration was used to collect the data. In 1954, the Census Bureau developed a mailing list from IRS records of employers with the industry classification based on lists from the SSA. In addition, to assist the Census

Bureau, the IRS changed the 1954 tax forms to include information needed to determine the physical location of the business rather than the mailing address and other changes needed by the Census Bureau to determine whether firms had employees. As a result, for the first time, it became possible for the Census Bureau to use tax return information as a source of census data. The Census Bureau decided that because firms with no paid employees represented a very small portion of the volume of retail and service trade but constituted a large number of businesses, it would forgo the collection of certain information from these firms and use the available tax return information to “impute” a complete report. For 1954, they used this technique by selecting a 50 percent sample of retail and service nonemployers. In 1963, this process was improved by the cooperation of the IRS in designing new census coverage questions for the 1963 tax returns, assigning industry classifications to nonemployer returns in the field offices. In 1967, the Census Bureau used all tax records and eliminated the sample. In addition, beginning with the 1987 economic census, a separate publication on nonemployer statistics was introduced. These data were published annually beginning with 1998, and the data for 2003 were released in June 2005.

The use by the Census Bureau of the IRS and SSA records was expanded significantly for the 1967 economic censuses. For 1967, the Census Bureau began using the same information to impute a census report for selected small employers. As a result, about 3 million small establishments were relieved of the task of completing census questionnaires. Overall, data for approximately 60 percent of the establishments included in the scope of the census were obtained through the use of IRS and SSA records. (Although large in number, these small establishments account for only about 7 percent of total sales.) In another expansion of the use of the IRS and SSA records, the Census Bureau also began to use tax return information for the larger employer firms to review the quality of their responses in the censuses.

The Census Bureau also has used its access to tax return information to create a new special supplement to the economic censuses, which is now known as the Survey of Business Ownership (SBO). Using data collected in the economic censuses, business tax return information from the IRS, and administrative data from the SSA, the SBO provides statistics that describe the composition of U.S. businesses by gender, race, and ethnicity, as well as owner’s age, education level, and veteran status. In the 1997 economic census, this survey was conducted as two separate surveys; a survey of minority-owned businesses and a survey of women-owned businesses. These surveys have been conducted as part of the economic censuses since 1972.

Under access to tax returns provided by executive orders and now Section 6103, BEA uses individual corporate income tax return data to estimate adjustments to the published IRS corporate profits data and converts these IRS data to accounting concepts underlying gross national product. For its regional economic accounts program, using a provision of Section 6103 that allows IRS to give access to BEA to review and tabulate individual tax return records for IRS, BEA produces tabulations (after review by the IRS to ensure taxpayer confidentiality) of nonfarm proprietors’ income by state and county that are used to distribute BEA’s national totals.

Another effort to improve the quality of the economic censuses came in 1992 when the Census Bureau and BLS entered into a memorandum of understanding regarding the use of BLS records by the Census Bureau for the economic censuses. Under this agreement, the Census Bureau compiled a list of single establishment employer identification numbers for which it needed the BLS industrial classification codes. The BLS matched this against its own business register and returned a listing of matched records with the appropriate industrial classification codes. The codes provided by BLS were from its three-year classification update program and provided the Census Bureau with more up-to-date classification information, as the Census Bureau typically updated its classification of these businesses once every five years. Records are not available to determine the extent to which the Census Bureau classifications were changed as a result of this effort.

In 1998, BLS and the Census Bureau entered into another memorandum of understanding. This agreement authorized the agencies to conduct research to evaluate (1) the business registers maintained by the two agencies, (2) opportunities to improve each other’s list, and (3) the benefits and risks of sharing list information. After an initial round of comparisons had been completed and a report presented to the Federal Economic Statistics Advisory Committee, the project was stopped by the IRS because the Census Bureau may have given BLS staff access to tax return information embedded in the census register. OMB has reported that this research has resumed under a memorandum of understanding that has been approved by IRS.

The International Investment and Trade in Services Act of 1990 authorized BEA and the Census Bureau to exchange data collected by BEA under this act and data collected by the Census Bureau under Title 13. The purpose of the exchange was to improve the quality of each agency’s data and to allow for the preparation of a report that showed the distribution by industry and geography of establishments owned by foreign direct investors. The act authorized a similar exchange between BEA and BLS.

In 2003, using the authority from CIPSEA, BEA and the Census Bureau entered into an arrangement to match data from BEA’s surveys called Foreign Direct Investment in the United States and U.S. Direct Investment Abroad with data from the Census Bureau Survey of Industrial Research and Development (SIRD). Working with the National Science Foundation, which funds SIRD, the project demonstrated the feasibility of linking the the Census Bureau and BEA survey data to produce new data on the domestic and international dimensions of U.S. research and development (R&D) activity. The project generated new data on R&D activities of U.S. and foreign multinational companies and on the location of U.S. R&D activity by state. The project also allowed for a higher degree of integration between data on the domestic dimensions of R&D and data on the international dimensions. The project also provided benefits to both the Census Bureau and BEA through improvements in data quality. Research on the outcomes of the matching resulted in the Census Bureau’s adding over 500 companies to the sample for the 2003 SIRD and over 60 companies to the 2004 SIRD sample frame. The matching also indicated significant differences in the reporting of both R&D expenditures and industry classification. These differences were resolved and resulted in improved quality in BEA’s estimates.

BLS produced several reports as a result of the sharing authorized by the International Investment and Trade in Services Act of 1990. In 1993, BLS released 1989 employment data on the occupations for manufacturing by industry and country of ownership in foreign-owned establishments. BLS prepared this report using data from the BEA’s 1989 Annual Survey of Foreign Direct Investment in the United States to identify foreign-owned establishments in the BLS Covered Employment and Wages data. These establishments were then linked to establishments from the BLS Occupational Employment Statistics Survey of Manufacturing Industries for 1989.

In 1995, BLS released tabulations for the fourth quarter of 1991 showing a distribution by industry and geography for foreign-owned establishments in the United States based on employment data from their Covered Employment and Wages data. BEA is also planning to provide BLS with company and revenue data from several of its international surveys so that BLS can study expanding their International Price Program to include a new price index for royalties and license fees. BEA hopes to benefit from this sharing by comparing its sample frame with the one compiled by BLS for this price program.

Berner, R. 2001 Testimony before the House Subcommittee on the Census Committee on Government Reform on the Statistical Efficiency Act of 1999, Washington, DC.

Council of Economic Advisers 2002 Economic Report of the President. Washington, DC: Executive Office of the President.

Davenport, C., and the Administrative Conference of the United States 1976 Report on Administrative Procedures of the Internal Revenue Service, October 1975, to the Administrative Conference of the United States. Senate document, 94th Congress, 2nd Session; 94-266.

Greenia, N., J. Lane, and B. Jensen 2001 Business perceptions of confidentiality. Pp. 395-429 in Confidentiality, Disclosure, and Data Access: Theory and Practical Applications for Statistical Agencies. P. Doyle, J.I. Lane, J.M. Theeuwes, and L.V. Zayatz, eds. Amsterdam, The Netherlands: North-Holland, Elsevier Science.

Haver, M. 1996 Testimony before the House Subcommittee on Government Management, Information and Technology, 104th U.S. Congress, Washington, DC.

Joint Committee on Taxation 2000 Study of Present-Law Taxpayer Confidentiality and Disclosure Provisions as Required by Section 3802 of the Internal Revenue Service Restructuring and Reform Act of 1998, Volume I: Study of General Disclosure Provisions (JCS-1-00). Washington, DC: U.S. Government Printing Office.

McMillen, D. 2003 Privacy, Confidentiality, and Data Sharing. Unpublished paper presented at the National Research Council’s Committee on National Statistics Confidential Data Access for Research Purposes Workshop, Washington, DC.

National Research Council 1976 Setting Statistical Priorities: Report of the Panel on Methodology for Statistical Priorities of the Committee on National Statistics. Assembly of Mathematical and Physical Sciences, Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

1980 Report of an October 1979 Conference on Sharing of Social Science Research Data. Committee on National Statistics, Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

1985 Sharing Research Data. Committee on National Statistics. S.E. Fienberg, M.E. Martin, and M.L. Straf, eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

1993 Private Lives and Public Policies: Confidentiality and Accessibility of Government Statistics. Panel on Confidentiality and Data Access, Committee on National Statistics. G.T. Duncan, T.B. Jabine, and V.A. de Wolf, eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

2000 Improving Access to and Confidentiality of Research Data: Report of a Workshop. Committee on National Statistics. C. Mackie and N. Bradburn, eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.

2005a Principles and Practices for a Federal Statistical Agency, Third Edition. Committee on National Statistics. M.E. Martin, M.L. Straf, and C.F. Citro, eds. Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.

2005b Expanding Access to Research Data: Reconciling Risks and Opportunities. Committee on National Statistics, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.

Office of Federal Statistical Policy and Standards 1978 A Framework for Planning U.S. Federal Statistics for the 1980’s. Washington, DC: U.S. Department of Commerce.

Privacy Protection Study Commission 1977 Personal Privacy in an Information Society. Washington, DC: U.S Government Printing Office.

Singer, E. 2001 The potential of confidentiality and attitudes toward data sharing by federal agencies. Pp. 341-370 in Confidentiality, Disclosure, and Data Access: Theory and Practical Applications for Statistical Agencies. P. Doyle, J.I. Lane, J.M. Theeuwes, and L.V. Zayatz, eds. Amsterdam, The Netherlands: North-Holland, Elsevier Science.

Social Security Administration 1991 Social Security Administration’s continuous work history sample. Social Security Bulletin 52(10).

U.S. Bureau of Economic Analysis and U.S. Bureau of the Census 2003 Foreign Direct Investment in the United States: Establishment Data for 1997. Washington, DC: U.S. Government Printing Office.

U.S. Census Bureau 1968-2002 Economic Censuses. Washington, DC: U.S. Department of Commerce.

2005a County Business Patterns: 2003. Washington, DC: U.S. Department of Commerce.

2005b Research and Development Data Link Project: Final Report. Washington, DC: U.S. Department of Commerce.

2005c Nonemployer Statistics Report: 2003. Washington, DC: U.S. Department of Commerce.

2005d 2002 Survey of Business Owners: Advanced Report. Washington, DC: U.S. Department of Commerce.

U.S. Department of Labor, Bureau of Labor Statistics 1993 New Research on Occupations in Foreign-owned Manufacturing Establishments in the United States. (News release.) Washington, DC: U.S. Department of Labor.

1995 Employment and Wages in Foreign-owned Businesses in the United States, Fourth Quarter 1991. (News release.) Washington, DC: U.S. Department of Labor.

U.S. Department of the Treasury, Department of Tax Policy 2000 Report to the Congress on Scope and Use of Taxpayer Confidentiality and Disclosure Provisions as Required by Section 3802 of the Internal Revenue Service Restructuring and Reform Act of 1998, Volume I: Study of General Disclosure Provisions. Washington, DC: U.S. Department of the Treasury.

U.S. General Accounting Office 1979 After Six Years, Legal Obstacles Continue to Restrict Government Use of the Standard Statistical Establish List. Washington, DC: U.S. General Accounting Office.

1998 Statistical Agencies: Proposed Consolidation and Data-Sharing Legislation. Washington, DC: U.S. General Accounting Office.

2001 Record Linkage and Privacy: Issues in Creating New Federal Research and Statistical Information. Washington, DC: U.S. General Accounting Office.

U.S. Office of Management and Budget 2001-2005 Statistical Programs of the United States Government, Fiscal Years 2002-2005. Washington, DC: Executive Office of the President.

Wilson, O., and W. Smith 1983 Access to tax records for statistical purposes. Pp. 595-601 in Proceedings of the Survey Research Methods Section. Alexandria, VA: American Statistical Association.