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

The workshop concluded with a summary presentation by Robert Parker, former chief statistician at the Government Accountability Office and the Bureau of Economic Analysis (BEA), in which he highlighted some of the key points made during the workshop. He also posed a few questions and offered suggestions to the agencies for going forward. Kathleen Utgoff, commissioner of the Bureau of Labor Statistics (BLS), Howard Hogan, associate director for demographic programs at the Census Bureau, and Rosemary Marcuss, deputy director of BEA, provided concluding remarks.

In a paper written for this workshop (see Chapter 5), Parker summarized the history of data-sharing legislation and various reports on the topic. Among his points were the following (the first three are from his paper and the last two from his presentation):

1. Prior to the 2002 Confidential Information Protection and Statistical Efficiency Act (CIPSEA), there was a significant amount of interagency business data sharing. Most Census Bureau and BEA access to tax return data was granted under Internal Revenue Code Title 26, Section 6103; more modest data sharing among BEA, the Census Bureau, and BLS was authorized by the International Investment and Trade in Services Act of 1990. For data sharing related to use in the economic censuses, the Internal Revenue Service (IRS) facilitated the success of this program by adding questions to selected tax returns solely for the purpose of the censuses.

1. Since the passage of CIPSEA, additional data-sharing programs have been undertaken and, based on the agency presentations at the workshop, others are in the planning stages.

2. The administration has not actually submitted its version of the “companion legislation.” According to agency representatives, the Joint Committee on Taxation has attempted to develop its own version, but it has not met agency needs (there also seem to be different opinions as to whether the Treasury Department supports the legislation).

3. To expand the use of data sharing in the absence of passage of companion legislation, there should be a comprehensive review of IRS Publication 1075, which sets the rules that determine the extent of data sharing under Section 6103 by the IRS and the CIPSEA-designated agencies. This review could result in changes in the rules that might further facilitate data sharing within the provisions of Section 6103 (current rules were not developed in conjunction with the agencies, and there is no public record of the underlying legal interpretations).

4. Discussions have taken place between the Census Bureau and BLS about developing a single business register for employer firms (for which both agencies have data), perhaps in time for the 2007 economic censuses. Although the current business comparison project will yield a substantial amount of information about current differences, BLS and the Census Bureau presented no plans on how this information might be used to develop a unified business register (because of problems with various state laws, it may be necessary for BLS to maintain a separate national register as the sample frame for its establishment surveys).

Following up on the fourth point, Parker suggested that IRS and the statistical agencies could work to clarify the rules governing what the Census Bureau is permitted to do with data, such as those indicating business names and addresses, which, at the time they are received, are considered federal tax information. The Census Bureau must reenter the information on another form and send it to businesses for verification to become sharable data under CIPSEA. Because the IRS has never published legal interpretations explaining how they went from Section 6103 to these regulations, Parker suggested that the agencies and the IRS might collaborate to see whether or not current regulations are appropriate or necessary and provide input on what can be done to make it easier to convert tax information into Title 13 information.

Parker also reflected on Carol Corrado’s presentation, particularly her point that maintaining two different establishment lists is unnecessary and counterproductive. Parker noted that the comparison project has a seemingly infinite research agenda, but that nothing concrete was presented by the BLS and Census Bureau presenters about producing a single

list. He argued that users would ultimately gain little from that particular data-sharing opportunity unless there were plans for some implementation of a single list.

Kathleen Utgoff, Howard Hogan, and Rosemary Marcuss responded to Parker’s remarks as well as to the workshop presentations in general. Utgoff and Hogan both discussed the various data-sharing projects and agreements that are planned or already taking place and the players in data sharing outside the IRS. They specified the various programs that would benefit from data sharing, reiterating the advantages of a single business list, and discussing the short- and long-term effects of CIPSEA on resources and on the work of the agencies. Marcuss suggested that the agencies seek advice on best practices from important data users and work toward increased access in the long term.

Utgoff responded to the call for a single business establishment list, acknowledging the potential value, but also noting that this is just one reason for extending data sharing. She discussed other aspects of BLS work that may be enhanced by data sharing, including improved unemployment and income measurement and development of better survey sample frames. She also noted the potential of data sharing to improve the capacity of state governments to forecast revenues and for underpinning economic development plans.

Utgoff reported that both BEA and BLS are motivated to improve data sharing and highlighted other data-sharing projects among BLS, BEA, and the Census Bureau. Two such agreements exist between BLS and BEA—one to share consumer price index data and another to share company and revenue data (collected under the International Investment Trade and Services Survey Act). Sharing the price index data will allow BEA to determine the feasibility of developing interarea price indexes, which could be used to adjust personal income estimates for price differences across geographic areas. New price indexes to measure royalties and license fees in international transactions could also be created from revenue data shared by BEA. The agreements between BLS and the Census Bureau authorize sharing of the BLS Quarterly Census of Employment and Wages data for use by the Census Bureau Longitudinal Employer-Household Dynamics program, as well as use of the Census Bureau Business Register by the BLS producer price index program to facilitate more efficient sampling.

In order to fully realize the benefits of CIPSEA, these interagency projects require a dedicated commitment of resources and support from the data-user and policy-making communities. One aspect of the overall benefit of these projects is the opportunity to reduce costs and respondent burden. Workshop participants paid particular attention to the data-sharing component of the CIPSEA legislation, but Utgoff acknowledged

the importance of Subpart A, establishing uniform statutory confidentiality protections, which BLS has long needed as part of their effort to maintain survey response rates.

Hogan began his remarks with the point that conducting the research and development on data sharing marked the first real use of CIPSEA, as people learned about such things as the documentation required, the approvals needed, and the mechanics of applying it. Beyond IRS data, Hogan noted opportunities for data sharing in diverse areas, such as product line and price index information.

On the business list topic, Hogan stated that short-term goals can be accomplished while proceeding with long-range plans. For example, research can begin on the differences in the lists at the aggregate level. The Census Bureau has already seen benefits from list comparison work—for example, 10 percent more firms have been added to the research and development sampling frame, and costs and respondent burden have been reduced. Before a single list is possible, Hogan observed, tax data sharing and the role of the states (which provide and share data with BLS but not with the Census Bureau) must be addressed. Overall, there are a number of benefits to the Census Bureau made possible by data sharing, and he expressed excitement on behalf of the agency over the cooperative opportunities.

Adding to the comments of Fixler and Landefeld, Marcuss reiterated the BEA goal of bringing analytical energy to the task of demystifying data inconsistencies and making data from diverse sources fit. BEA will continue to adjust its methods and pursue new approaches to more fully exploit the currently sharable and available data. In addition, BEA seeks advice from the agencies and the IRS on the best, least intrusive, procedural changes in data-sharing practices that can improve estimates of gross domestic product. She added that the agency would quickly utilize information on the sources of differences between the Census Bureau and BLS data from any comparison work—for example, from a partial comparison of the business registers, even short of movement to a single list.

Marcuss added that advice from important data users—particularly policy makers in the administration and Congress, and businesses—on where the greatest weaknesses exist would continue to serve as valuable input. As a long-term goal, she hoped for increased data access for statistical purposes, confirming that BEA has already taken steps forward and is ready to participate in and use the information gained from the data-sharing projects. Her comments echoed those of the workshop in general. Overall, the agency representatives, workshop presenters, and other participants were supportive of extending data sharing to the extent that such arrangements can be safely implemented to increase the efficiency, accuracy, and consistency of the nation’s economic statistics.