COPAFS
 

MINUTES June 7, 2013 COPAFS QUARTERLY MEETING


Report from the Board of Directors

Noting troubling legislative initiatives, COPAFS chair Maurine Haver explained that COPAFS has adopted a strategy to educate congressional members and staff on the importance of federal data. Given the importance of press reports, the Board is considering a conference on federal data for journalists. A conference of this type some years ago focused on economic statistics, and it has been suggested that the proposed conference could focus on higher education, employment and immigration statistics. Maurine encouraged COPAFS reps to communicate thoughts and suggestions on the proposed conference. 
 

Update from the Executive Director

Kitty Smith reported that she attended the PAA and AAPOR conferences, and hopes to attend more conferences of member associations. She noted that COPAFS helps host the FCSM research conference, which is scheduled for November 4-6, and has an impressive program in the works. She also described the COPAFS membership campaign that kicks off the week of June 10. As Smith put it, many associations and businesses that use federal data are not yet COPAFS members, and she asked for referrals to add to the list that she plans to contact.

Smith then introduced the first session, noting the great uncertainty over budgets in the wake of sequester, and introducing the panel of agency officials who had joined the meeting to present on the topic. 


Panel of Statistical Agency Officials on 2013 and 2014 Budget Implications

Tom Messenbourg.  Acting Director, US Census Bureau

Messenbourg expressed how pleased the Census Bureau is to have John Thompson as the nominee for its next director.  The confirmation hearing is set for July 9, and it is hoped Thompson will be confirmed later in the summer.  Messenbourg explained that the Commerce Department sent “spend plans” including sequester cuts to congress, but they were rejected – primarily in objection to NOAA furloughs of weather service employees.  NOAA has found a way around the furloughs, but the Census Bureau still has no spend plan to work with.  However, it is clear there will be cuts, so the Census Bureau is slashing contracts, and imposing strict constraints on hiring, travel, and training.  They do not plan furloughs, but some products will be eliminated and others delayed.  The 2014 budget calls for much restoration – with the biggest increase being for 2010 census, as it concludes its research and testing phase.  It is doubtful that Census will receive the full request, and the biggest concern is that a continuing resolution could leave 2014 funding at 2013 levels.  Messenbourg also mentioned the Duncan bill, which would repeal the Census Bureau’s authority to collect most surveys – including some that it no longer collects.  There is also the Poe/Paul bill that would make response to all but four ACS questions optional, and require prominent messaging to that effect.  If it passed, the bill would have a negative impact on response rates – probably similar to that experienced in Canada, where many communities no longer have data from the now voluntary sample census. 

Jack Galvin.  Deputy Commissioner, Bureau of Labor Statistics

Galvin remarked that the BLS situation sounds similar to that at Census.  With budgets sharply reduced, they have eliminated some programs – such as statistics on mass layoffs, green jobs, and international labor comparisons.  All were producing useful data, but other BLS programs had higher priority.  Even these cuts were insufficient, so BLS has a freeze on hiring, and is cutting spending on contracts.  The 2014 request restores most of the funding cut in 2013 – but would not restore the eliminated programs.  Small increases are requested for supplemental CPS labor force questions, and another to support the Census Bureau’s development of an alternative poverty measure.   

Charlie Rothwell.  Acting Director, National Center for Health Statistics    

Rothwell said the funding situation at NCHS differs from that at Census and BLS, as much of their funding comes not from Congress, but from other agencies and partners – and those funds cannot be sequestered twice.  The concern is that as these agencies and partners absorb cuts, they may be less interested and able to fund the data they get from NCHS.  The NCHS budget has been flat-lined in recent years.  But while that is better than what many agencies have experienced, it is preventing improvements to programs, and like other agencies, NCHS is cutting back on travel and other expenses.  The 2014 request calls for an increase over the pre-sequester 2013 amount, and Rothwell said the request at least signals a recognition of the need for additional funding.            

Joe Reilly.  Associate Administrator, National Agricultural Statistics Service

Reilly described NASS as a statistical agency within USDA that also serves as the statistical arm of every state department of agriculture.  He said there has been major suspense over the census of agriculture, as there were indications from Congress that there would be no additional funds for data collection.  Eventually, they received some of what was needed.  NASS also is in the middle of a reorganization, which makes this a difficult time to absorb the sequester cuts.  In response, NASS has prioritized its reports and surveys – cutting some, and keeping as many as they can.  They have received many expressions of concern over the elimination of the milk production reports, which are used to determine the proper price of milk.  For now, they will run the data based on a model, the thinking being that something is better than nothing.  NASS also has cut hiring, stopped work on outlying areas of the US, and will not produce special tabulations – such as data for ZIP Codes.  They just received their 2014 House mark, and it is a sharp reduction from the requested amount.  They will finish the census of agriculture, but are proceeding with transformational initiatives (started before sequester) including the elimination of 24 field offices – reducing their footprint to 12 regional offices.        

Tom Williams.  Assistant Administrator, Energy Information Administration 

Williams described EIA as the statistical arm of the Energy Department, and noted that data collection is 60 percent of its budget, with analysis accounting for 30 percent, and 10 percent other activities.  Petroleum and natural gas are the biggest focuses of their work.  The EIA 2013 budget has been tough, since it was a cut from 2012 even before the sequester.  The 2014 request is an increase – higher even than the 2013 request, and Williams described initiatives they would like to pursue with that level of funding.  As he put it, they like the going in position, but are not betting that it is what they will have coming out.      

William Sabol.  Acting Director, Bureau of Justice Statistics

Sabol gave a brief overview of BJS and its 46 statistical collections.  Its budget numbers are much smaller than the other agencies, and like NCHS, BJS has not fared that badly in recent years.  They are still waiting for their final 2013 spend plan from Congress, but with cuts inevitable, already have initiated cuts in research, evaluation and statistical activities – including programs such as the National Crime Victimization Survey.  Much BJS funding comes from set asides – money from grant and reimbursement programs set aside for statistical purposes.  These funds are at the discretion of the Attorney General, and the concern is that as these grant programs take budget hits, there will be pressure to reduce the amount set aside for statistical activities.  BJS also has been affected by the Department of Justice hiring freeze, and with staff reductions, they have to delay or cut back on programs.        

Marilyn Seastrom.  Chief Statistician, National Center for Education Statistics   

With time running short, Seastrom gave a quick summary of the NCES cuts from 2012 to 2013, and confirmed that they are deferring some work, and have canceled the 2014 national household survey – the source for parent reports, and other information they cannot get from schools.  They also have delayed the adult training and education survey, are cutting assessment programs, technical assistance to states, and a study on teacher compensation.  The 2014 request calls for an increase over 2013, and it is hoped it will enable them to restore the national and international assessments.  NCES also hopes to move the survey of post-secondary education financing from a 4 year cycle to a two year cycle.   In contrast to many agencies, NCES has not been under a hiring freeze, but they have not been hiring to replace attrition, so staffing is reduced.   


Panel of Congressional Staff on the Use of Federal Statistics and Outlook for Statistical Agencies


Kim Kowalewski.  Deputy Assistant Director for Macroeconomic Analysis, Congressional Budget Office

Kowalewski explained that CBO provides independent, nonpartisan data to Congress for budget analysis purposes.  They do not make policy recommendation, but rather evaluate the pros and cons of budget proposals, and provide cost estimates of bills that Congress is considering.  Kowalewski is in the macroeconomic analysis division, where they produce economic projections of the US economy, and examine the macroeconomic implications of proposed policies.   

CBO’s economic projections sometimes are criticized as inaccurate, but Kowalewski explained that their projections are designed to show what would happen if specific policy alternatives were pursued.  He described the models they use to project GDP and potential GDP.  After the 2001 recession, real GDP was only slightly lower than potential GDP, but since 2008, real GDP has been sharply lower than potential.  The shortfall is a measure of unused resources in the US economy (amounting to $6 trillion), and the difference is a frame of reference, telling Congress how weak the economy has been.    

Kowalewski also presented statistics on labor income as a percent of gross domestic income.  The numbers have gone up and down since 1950, but have dropped since 2000 to their lowest level.  Put another way, the percent of total income coming from capital income (as opposed to labor) has increased sharply.  Possible explanations include technical change, globalization, and increased trade with China.  Kowalewski also showed data on inflation and unemployment, where recent trends also are a departure from the expected pattern, with inflation hardly moving despite large swings in employment. 


Jeffrey Post.  Professional Staff (Majority), House Committee on Oversight and Government Reform.    

Post started with a recitation of the increasing per household cost of the census, and the difficulties it poses as budgets become more constrained.  And with sequestration apparently not ending anytime soon, he said the Committee seeks to help the Census Bureau do more with less.  The focus, he suggests, needs to be on initiatives such as promoting online response to the census and ACS.  Post said he could not be specific on budget numbers because they are not known, but offered to respond to questions. 

The first question concerned the risk that funding withheld now could impair the ability to reduce census costs through future efficiencies.  Post responded that everyone is experiencing cuts (including the Committee), and the money is just not there.  He expressed confidence that the Census Bureau has good people who will find ways to get things done with reduced budgets.

Asked about the pressure to make ACS response optional, Post said his boss (the Committee chair) supports the ACS, but some in the caucus question the government’s right to ask questions of the type on the ACS.  He commented that the initial ACS rollout caused much of the problem by not explaining why the questions are asked, and said it does not take many complaints from angry constituents to move representatives to oppose mandatory ACS response.  Post then praised the Census Bureau for communicating the justification for questions more effectively now, and ensuring that ACS field reps are not overly zealous in obtaining a response.  He reiterated that the chairman supports the ACS, and knows it is important to business and other users. 

Not surprisingly, there was a question about the Duncan bill, which would eliminate the Census Bureau’s authority to collect most of the survey data it currently collects.  Post said it is important to show members that there are legitimate reasons for these surveys.  He also stressed the importance of better highlighting the confidentiality protections in federal data collections.

Given that the data user community has been trying for years to educate congressional staff, Post was asked for advice on ways to do this.  Post acknowledged the challenge, noting that the census is a peripheral concern to the Committee, and thus assigned to junior staffers.  In other words, the turnover in congressional staff is likely to continue.  He cited as a positive step, the resources one is directed to now on the online ACS forms.  The links provide answers to respondent questions, and Post contended that once provided with the reasons for asking ACS questions, much respondent anger is defused.  He stressed however, that data users will have to continue taking an active role in education.

Kowalewski was asked if CBO is looking at the impacts of immigration reform on productivity. Calling it a huge issue, Kowalewski said they are working on this now, and are hoping to come up with a consensus estimate.    


The Committee on National Statistics Principles and Practices for a Federal Statistical Agency – Fifth Edition.  
Connie Citro, Committee on National Statistics, The National Academi
es

Citro provided a brief history and overview of the National Academy of Sciences, and the Committee on National Statistics, which conducts workshops and convenes panels of experts to address statistical matters.

CNSTAT also produces the Principles and Practices for a Federal Statistical Agency. The publication is a statement of principles the academy considers fundamental to a federal statistical agency, and a statement of practices that are ways of making these principles operational. It is not a guide to data collection, estimation or dissemination, nor does it provide data quality standards. It is a set of guidelines, not a set of prescriptions.

CNSTAT produced the first edition of Principles and Practices in 1992, when the Bureau of Transportation Statistics was being established. Congressional staff and cabinet departments inquired about what constitutes an effective statistical agency, and CNSTAT recognized that not all government officials are knowledgeable about proper practices for federal statistics. Among the reasons CNSTAT identifies for establishing a statistical agency are the need for information beyond the scope of operating units; the need to establish independence of data from policy or operating control; and the need to consolidate collection, compilation, analysis and dissemination of statistics in one unit.

The first edition of “P&P”, issued in 1992, was edited by Margaret Martin and Miron Straf. It defined a federal statistical agency as “a unit of the federal government whose principal function is the compilation and analysis of data and the dissemination of information for statistical purposes.” It identified the following three principles.

• Relevance to policy issues.
• Credibility among data users.
• Trust among data providers and data subjects.

And eleven practices.

• A clearly defined and well-accepted mission.
• A strong measure of independence.
• Fair treatment of data providers.
• Cooperation with data users.
• Openness about the data provided.
• Commitment to quality and professional standards.
• Wide dissemination of data.
• An active research program.
• Professional advancement of staff.
• Caution in conducting non-statistical activities.
• Coordination with other statistical agencies.

Citro described the “strong measure of independence” as perhaps the most important, but also the trickiest objective to achieve, as agencies must operate within the framework of congressional, OMB, and departmental oversight, while maintaining credibility as impartial purveyors of information. Aspects of independence include:

• Organizational separation from program activities.
• Qualified head appointed for a fixed term and with direct access to the secretary.
• Broad authority over scope, content, and frequency of data.
• Primary authority for selection and promotion of staff.
• Authority to release statistical information without prior clearance.
• Authority to speak on agency’s program before Congress, with congressional staff, and before public bodies.
• Adherence to predetermined schedules for release of data.
• Maintenance of clear distinction between statistical releases and policy interpretations by others.

As Citro described it, not all agencies qualify on these measures. For example, some are so “layered down” in a program agency that they cannot speak for themselves.

About 2000, CNSTAT determined it made sense to update the Principles and Practices, and it is now updated every four years. The second edition added an explicit discussion of the need for federal statistics – noting that private sources are not likely to provide all the data that are needed, or to make data widely available. It also added a practice calling for the continual development of more useful data (in recognition of the growing importance of the Internet). The third edition (2005) added an appendix on relevant legislation (such as CIPSEA), and added “authority over information technology systems” as a characteristic to strengthen independence. The fourth edition (2009) added a section on the organization of the federal statistical system, and another on legislation and regulations governing federal statistics (such as the Paperwork Reduction Act, CIPSEA, and the E-Government Act). It also added a practice calling for a strong internal and external evaluation program. The fifth edition (2013), dedicated to the memory of Margaret Martin, included a beefed up discussion of independence and non-statistical activities, separated privacy from confidentiality protection, and stressed the need for active collaboration (not just cooperation) among agencies – especially in an age of constrained budgets.

Citro described some of the uses of the Principles and Practices publication, and remarked that she has already started work on the sixth edition


On the Census Undercount of Young Children.
Bill O’Hare.  O’Hare Data and Demographic Service

O’Hare noted that young children have been historically undercounted in the census, but there is almost no systematic information on the topic. He described the two basic measures of census coverage – demographic analysis (DA), which compares census counts against estimates based on births and deaths, and dual-system estimation (DSE), which involves a coverage measurement survey.

DA appears to be the more relevant measure for the undercount of children. Almost all the DA estimate of population age 0-4 is based on births, so if the birth certificate data are good, the DA estimate is accurate. And it is the DA estimate that shows the large net undercount of young children. As O’Hare described it, the overall DA estimates for the 2010 census look good, as the 2010 census total population was very close to the independent DA estimate. But looking at coverage by age, one sees an undercount of children that was offset by an overcount of adults. The pattern of undercount by age is striking, with high undercount for the youngest ages becoming (and remaining) an overcount by about age 12. And this tendency is greater for minority populations, such as Blacks and Hispanics. Also notable is that the net undercount of both children and adults improved from 1950 to 1980, but since then the undercount of children has been getting worse.

Key questions, according to O’Hare are – Why is there a net census undercount for young children, why is there a net overcount of 14-17 year olds in 2010, why is there such a strong correlation between undercount and age among children, and why has the undercount of young children increased since 1980?

Potential explanations for the undercount of young children include the possibility that the DA estimates are too high, the extra effort required to include more than six people in a census response (10 percent of children are in households with seven or more persons), or the possibility that many who should have had age imputed as 0-4 were imputed as age 10-14. It has also been suggested that the parents of young children might not complete the census due to time demands (there is evidence of lower response rates for single parents with children), and young children may live disproportionally in households that are difficult to enumerate.

Summing up, O’Hare noted some implications of the undercount of young children. For example, if the 2010 census undercounts young children, Census Bureau estimates based on these counts will underestimate the size of this population, and these underestimated totals will be used as weights in surveys.

O’Hare’s presentation prompted much discussion, with COPAFS reps offering thoughts on how one might explain the high undercount among young children and why it is getting worse. One suggestion was that unauthorized immigrant households might exclude young children in their census response, believing it could protect them from detection.


Input from COPAFS Constituencies


Because not all statistical agencies were represented in the session on sequester cuts, Kitty Smith was asked if that was because the others were not impacted. Smith said that was not the reason, but noted that IRS Statistics of Income and Social Security have not been impacted because their work is so central to their departments that they are pretty much protected. In the case of BEA, they have not had to cut its core programs, but quality has become an issue as resources are stretched.

The meeting wound down with an extended discussion of options for educating congressional staff about federal statistics.

Presentation from June 7, 2013:   Read More