Minutes of the September 14, 2007 COPAFS Meeting

The COPAFS board meeting ran late, so in the board’s absence, Ed Spar started the meeting with his Executive Director’s report.

Ed Spar. Executive Director’s Report.

With respect to budgets, we can expect the continuing resolutions that we have become accustomed to in recent years. Concern over the NCHS budget continues, but additional funding has been requested that (if passed) would cover vital statistics. There is also concern that the proposed budget numbers might require some cuts in Economic Census activity. Spar reported that the Brookings Institution has revised the estimate of total federal dollars distributed based on census data from $200 billion to $326 billion.

Spar described a Federal Register notice that presents a proposed OMB Statistical Policy Directive providing guidance for the dissemination of federal statistical products to ensure that it is equitable and free from partisan influence. OMB Directive 3 already provides such guidance for the release of Principle Economic Indicators, but the proposed directive would apply to the full range of statistical products disseminated by federal statistical agencies. While the proposed directive’s objectives seem to be favorably received, Spar is concerned that some of its details could have unintended consequences (perhaps holding up some data products indefinitely), and will provide comments on behalf of COPAFS.

Asked the status of Steve Murdock’s nomination as the next Census Director, Spar remarked that all we know is that Murdock has the book of questions from Congress, and that it is hoped that the hearing will be in October.

Turning to housekeeping items, Spar plugged the upcoming APDU conference, and noted that COPAFS meetings will switch from the second week to the first week of each quarter – in order to avoid conflict with FESAC meetings (Federal Economic Statistics Advisory Committee). The final 2007 meeting is still scheduled for December 14.

Spar then introduced Bill O’Hare, from the Annie E. Casey Foundation, who alerted us to a bill that would produce a new federal survey that would provide information on the well being of children. O’Hare spoke highly of what the new survey would provide, such as consistent data across states. He also explained that the Casey Foundation does not normally promote federal legislation, but is in a position to promote this survey because they would be putting about $1 million into it – if the legislation passes.

Implementation Guidance for the Confidential Information Protection and Statistical Efficiency Act.

Brian Harris-Kojetin. Office of Management and Budget

Harris-Kojetin recited the (very) full name of the Confidential Information Protection and Statistical Efficiency Act, demonstrating why it is popularly shortened to CIPSEA. Although recently issued, CIPSEA was developed by an inter-agency team at OMB that was formed in 2003. The act is about strengthening and fostering trust in federal pledges of confidentiality, prohibiting disclosure, and controlling the uses made of statistical information. Harris-Kojetin explained that CIPSEA replaces a patchwork of protections with uniform protection that covers all federal data collected for statistical purposes. CIPSEA promotes protection by establishing strong penalties for disclosure ($250,000), and exempting data from FOIA requests.

Proposed guidance for implementing CIPSEA was announced in an October 2006 Federal Register notice, and final guidance was published in June 2007. Distinctions between statistical and non-statistical uses of data, and between statistical and non-statistical agencies are critical to the guidance for CIPSEA implementation. Statistical purposes relate to identifying information about groups, without identifying individual respondents (including persons, state or local governments, or businesses) within these groups, while non-statistical purposes involve the use of data on individuals that can affect the rights or benefits of respondents. Statistical agencies are those whose primary activities involve the collection of data for statistical purposes. The distinctions are critical, as CIPSEA prohibits the use of statistical data for non-statistical purposes, and applies somewhat differently to statistical agencies. For example, only statistical agencies can designate agents (contractors or other non-employees) to perform statistical activities, such as data collection.

CIPSEA requires that agencies inform respondents about the uses of the information they provide, and pledge to keep the information confidential. And when statistical agencies collect non-statistical information CIPSEA requires that they make this very clear to respondents that this is the case. The act also allows for the sharing of business information among three agencies – BEA, BLS and the Census Bureau—so long as the agencies inform businesses that the data are being shared, and for what purposes.

Agencies must submit reports to OMB (by April 30 of each year) on their use of the CIPSEA pledge, the use of agents, and the use of data sharing.

Business Establishment List Sharing: Ongoing Issues and Research.

Robert Parker. Consultant
James Spletzer. Consultant.
Thomas Mesenbourg. U.S. Census Bureau

Parker, presenting on behalf of the National Association for Business Economics, noted that this topic is a follow up to the data sharing provisions of CIPSEA described in the previous presentation. The objective of this provision is to improve the comparability and accuracy of federal economic statistics by allowing BEA, BLS, and the Census Bureau to share business lists in order to improve coverage, reconcile differences, and promote commonality across the agencies. Parker described a business list comparison project that dates back to 1998, but credited the arrival of CIPSEA with allowing the sharing of business data to move ahead. Parker also described “companion tax legislation” that would further these objectives, but which was not passed along with CIPSEA. He concluded by suggesting some important questions for BLS and the Census Bureau. For example: Has BLS made progress on legislation allowing them to share Census data with states? How is the Comparison List project being used to decide when Census should use BLS records for its register? Are there plans to further eliminate non-comparability between lists in the event that the companion tax legislation is passed?

James Spletzer presented next, and stressed that the business list comparison project is moving forward, and has accomplished much toward the goal of understanding differences between business lists. He then compared Census Bureau and BLS employment and establishment counts – noting that the national totals are similar, but that matching is difficult for multi-establishment firms (with only 13 percent agreeing on number of establishments). Spletzer also described issues related to industry coding, and the flows of data from BLS to Census and from Census to BLS. He concluded by recommending continued targeted research on the most obvious business list discrepancies, and further investigation of the overlap of BLS and Census Bureau efforts to collect employment and wage data. He also recommended proposals to eliminate duplicate efforts, and to resolve industry coding discrepancies.

Thomas Mesenbourg, presenting for the Census Bureau, reviewed the legislation (including CIPSEA) authorizing data sharing, and outlined the projects underway. He described a number of data sharing opportunities, including better leveraging of the Economic Census (which is believed to provide better industry codes because of the detailed information it collects) and the re-engineering of the Annual Report of Organization (because BLS may have better coverage of small, multi-unit firms). Other opportunities include the re-evaluation of industry coding business rules, further use of product information, and determining the best sources for births/deaths. Issues and challenges include enacting the CIPSEA companion tax legislation, ensuring that states use shared data only for statistical purposes, streamlining and documenting data sharing procedures, and increasing IT security requirements. Mesenbourg concluded by identifying next steps, including the completion of the business register comparison project, further research on industry codes, and streamlining data sharing procedures.

Status of the Survey of Income and Program Participation.

David Johnson. U.S. Census Bureau.

Johnson reviewed the Survey of Income and Program Participation (SIPP) program, which is designed to measure the distribution of income and eligibility for and participation in federal programs. He also described a timeline of recent events – in which the Census Bureau had planned to replace SIPP with a Dynamics of Economic Wellbeing System (DEWS), but has responded to user and congressional pressure to “save the SIPP” by focusing attention on a “re-engineered SIPP.”

As Johnson described it, the re-engineered SIPP will incorporate many of the enhancements that were to be built into DEWS. Re-engineering goals include cost reduction, improved accuracy, improved timeliness and accessibility, and improved relevance. Components of the re-engineered SIPP would include improved data collection and processing systems, and the development of event history calendars, which could reduce costs by permitting the collection of data once a year as opposed to once every four months. Samples drawn from the American Community Survey could improve the coverage of selected population segments (such as low income populations), and provide a link to other valuable information. The Census Bureau also is testing administrative data to see if they might replace some of the survey questions. Of course, some re-engineering efforts involve tradeoffs. For example, collecting data only once per year could reduce costs, but might risk greater attrition among SIPP respondents, and place greater burden on their ability to recall.

Johnson also noted that there is a CNSTAT panel on the re-engineering of SIPP, and that the Census Bureau is attending conferences and conducting meetings to get feedback from stakeholders. He described this activity as especially important because SIPP provides no official statistics. As Johnson put it, the data are just out there for users.

Johnson explained that the Census Bureau will continue with current SIPP panel data collection through January 2008, and would start the new SIPP panel in February 2008. This panel would be the same as the 2004 panel (no improvements yet), and its size would depend on the amount of funding available for it. Plans call for re-engineering planning from 2006 through 2011, followed by re-engineered SIPP panel data collection in 2011/2012 and beyond. Data products would include complete public data, synthetic data, and files to be accessed through research data centers (RDCs).

American Community Survey Program Update and Product Proposal Discussion.

Susan Schechter. U.S. Census Bureau.
Nancy Torrieri. U.S. Census Bureau.

Schechter described the recent release of 2006 ACS data, and the upcoming release which will feature selected profiles, workplace-based data, and data on the group quarters population. Because the 1990 and 2000 censuses provided only limited detail on the characteristics of persons in group quarters, the ACS tables will be our first look at some group quarters characteristics since 1990, and in some cases 1980. Schechter described the Census Bureau as excited about the release of group quarters data, but noted that its introduction will make some 2005-2006 comparisons difficult. She also explained that the provision of group quarters detail is determined by the size of the GQ sample – not the 65,000 population threshold for total population. Looking ahead to next year, Schechter noted that we would receive everything provided with this year’s release plus the first set of three-year average data.

Torrieri then described the Census Bureau’s outreach and planning efforts related to the multi-year average data. In addition to engaging stakeholders by attending and presenting at conferences, she described a recent Federal Register notice, which outlines the type of multi-year average products the Bureau plans to provide and the schedule for providing them. Comments on the notice are due by September 28, in large part because the Census Bureau wants to finalize the multi-year products by mid December, and then develop a comprehensive education campaign.

As part of developing the education campaign, Torrieri explained that the Bureau has initiated a project under contract with Saber Systems (and subcontract with COPAFS). The objective of the project is to tap the expertise of experienced users from different data user groups (such as federal agencies, state and local government, academia, and business) to develop ACS handbooks and other training materials tailored to those user communities. The effort will begin with a November 16 workshop that will contribute to the mid-December finalization of multi-year average products. The project’s authors then will prepare draft training materials, which will be reviewed at a May 2008 workshop, and final materials will be disseminated that summer.

Some COPAFS reps are concerned that the schedule provides too little time for users to review the proposed multi-year products, but the Census Bureau expressed confidence that the process they described will generate important feedback, and result in effective educational materials. The Federal Register notice certainly does not provide time for an extensive review of the long list of proposed ACS detailed tables, but the notice does not ask for that level of review. Rather, it seems to solicit comments only on broadly defined products and the general release strategy.

Concerns of COPAFS Constituents

No concerns were raised, and the meeting was adjourned.