COPAFS QUARTERLY MEETING
SEPTEMBER 9, 2011
Executive Director’s Report. Ed Spar
Spar described the budgets a “grim,” with early House mark ups calling for severe cuts. Late word is that the plan now is to have no bills, but instead Congress will wait and move to an omnibus bill perhaps in December. So agencies will be working with their 2011 budget levels for the foreseeable future with possible interim cuts, will little knowledge about the final appropriation.
Spar highly recommended a new Census Bureau report, “Management Challenges of the 2010 Census,” by Dan Weinberg. The report can be found on the COPAFS website. Final criteria for defining urban areas based on the 2010 census counts have been published in an August 24, 2011 Federal Register notice. The Bureau of Transportation Statistics has resurrected its advisory committee, and the committee will meet September 28. Spar said these meetings are worth attending, but added that those who do should be aware that the Department of Transportation building is tough to get into. Another item for calendars is the 2012 Applied Demography Conference in San Antonio January 8-10.
The next COPAFS meeting is December 2, and the dates for 2012 meetings are March 16, June 1, September 14 and December 7.
Spar then turned to COPAFS Chair Felice Levine for a discussion of the Common Rule. Levine referenced a July 26, 2011 Federal Register notice that requests comments on how current regulations for protecting human subjects in research might be modernized and revised to be more effective. The current regulations date back more than three decades, and 20 years have passed since15 federal departments and agencies adopted the “Common Rule” to promote uniformity in human subject protections. Levine said changes to the Common Rule could have major implications for social research – with an emphasis on protections at the potential expense of data access. She recommended that COPAFS reps read the (lengthy) Federal Register notice, but also check out additional material on the subject at firstname.lastname@example.org on the American Education Research Association website.
A Review of the Upcoming 2012 Economic Censuses
Ahmed started with background on the economic censuses (EC), which are conducted in years ending in 2 and 7. The EC covers more than 28 million establishments, with forms sent to less than 20 percent of businesses, and administrative records used for very small and non-employer businesses. The economic census program includes a survey of business owners, a business expenses survey, the commodity flow survey, and economic censuses for Puerto Rico, Guam and the island areas. The program covers over 1,000 industries, 10,000 products and 17,000 different geographies.
Shirin Ahmed. US Census Bureau
Work on the 2012 EC started in early 2009, and the schedule calls for data collection starting late 2012, editing and imputation in 2013, data analysis in late 2013, and data dissemination shortly after that.
The EC works with a six year budget cycle, and Ahmed remarked that they are entering 2012 with the lowest increase they have ever seen. Some attendees wondered why economic census costs would be increasing, and Ahmed’s citation of staff, benefits and IT surprised some given sharp decreases in the cost of hardware. Ahmed explained that it is not just hardware, but also the expense of updating old programs to more contemporary languages. Still, there was comment that the Census Bureau needs to make a better case for increased costs, with explanations citing security and confidentiality protection probably resonating better with appropriators.
Objectives for 2012 are to use newly developed parts of the North American Product Classification System, incorporate revisions to NAICS, and add product manufacturing product detail to account for Current Industrial Report industries and the green economy. Another initiative is to reduce respondent burden. Online response options are being expanded and enhanced over those available in 2007, and the business help site (which helps businesses respond to the census) is being modernized. The census also is expanding the account manager program – which assists large companies with EC response – from 1,300 to about 2,000 companies. The Census Bureau also is expanding advance outreach, with an expanded mailing to 80,000 businesses with a focus on selected industries and selected metropolitan areas. Ahmed noted that the “mandatory” response message is an effective motivator, so it is prominent on outreach materials.
Economic census data are currently on the legacy American FactFinder system, but should be on the new American FactFinder by the end of October. Ahmed described some of the features that will be available on AFF2. Ed Spar thanked Ahmed for the presentation, and reminded attendees that all of this assumes there will be a budget for the economic censuses in the first place – as there has been talk of eliminating the 2012 EC if cuts to the Census Bureau budget are severe enough.
Using a Composite Index of Financial Condition Indicators to Predict Turning Points in the U.S. Business Cycle
Ozyildirim described himself as one of a small group of economists at the Conference Board (TCB). Founded in 1916, TCB is a non-advocacy and non-profit research organization that provides businesses and other organizations with practical knowledge to improve their performance and better serve society.
Ataman Ozyildirim. The Conference Board
The Conference Board has a long history of economic work on a wide range of indicators, including help-wanted data, consumer confidence, US and global composite, coincident and leading indexes, and others. In 1996, TCB took over the Business Cycle Indicators Program from BEA, and while they have introduced several new methods, there have been no major revisions since that time. Ozyildirim commented that the time may have come for such revisions.
As Ozyildirim described it, TCB has taken an indicator approach to the measurement of business cycles – looking at the level of economic activity, and seeking to identify indicators that are coincident with downturns and recessions. He presented a number of graphs showing monthly economic activity back to 1959, and how the Conference Board’s indexes coincided with the beginning and end of recessions. One is the Leading Economic Index for the US (LEI) which is based on 10 components that track economic output, and suggest future activity – such as new orders, housing permits, average weekly hours for manufacturing, and consumer expectations. TCB also produces a Coincident Economic Index for the US, which is based on a wider range of measures. TCB research seeks to determine how well the indexes and components perform in reflecting and predicting recessions, but the task is a challenge because data on some of the components only date back as far as two or three recessions.
The criteria for selecting indicators in TCB indexes are economic significance, statistical soundness, and conformity with business cycles. In particular, an indicator should give few false signals of recession. Given the growing importance and complexity of the financial sector Ozyildirim commented that financial indicators need to be reviewed. For example, he described one financial component that no longer reflects recessions, and needs to be replaced. TCB is committed to researching and developing indicators that better correspond with the business cycle, and measures that provide advance indication of economic recession are especially valued.
With housing often cited as critical to recovery from the current economic downturn, Ozyildirim was asked why the LEI has only one indicator related to housing. He agreed that housing is a key economic component, but noted that recessions typically do not start with housing – that is something different with the recent recession. And asked why all index components are weighted the same, Ozyildirim explained that BEA had used weights, but stopped in 1989 (before TCB took over). The weights had little impact on the results (served mostly to complicate the calculations), so there is no sense of urgency in re-establishing the use of weights.
Restructuring the Census Bureau Regional Offices
On June 29, 2011, the Census Bureau announced that it is permanently closing six of its 12 regional offices (ROs). Mesenbourg explained that the Census Bureau typically reviews its regional office structure after each census, but this is the first major change in about 50 years.
Thomas Mesenbourg. US Census Bureau
The reduction is in response to realization that future budgets will be constrained and possibly declining, while agencies that sponsor Census Bureau surveys are demanding lower costs and increased efficiency. And with survey organizations increasingly moving to leaner infrastructures, it has become clear that the survival of the Census Bureau’s survey business requires changes at headquarters and in the field.
Mesenbourg described the year long process in which the Director, Deputy Director, Associate Director for Field Operations, Chief of Field Division and all 12 Regional Directors worked together on the restructuring plan. He stressed that the process was goal and data-driven, with over 20 designs considered (ranging from 4 to 12 regions). The alternative designs were rated on eight goals, with a focus on minimizing costs and improving data quality.
The process yielded two elements of change – a change in the number of offices (a reduction from 12 to six), and a change in the management of data collection. The offices that are staying open are New York, Philadelphia, Atlanta, Chicago, Denver, and Los Angeles. The offices that are closing are Boston, Charlotte, Detroit, Kansas City, Dallas, and Seattle.
There were two prerequisites in delineating RO boundaries – the regions had to be contiguous and not divide states. Also considered were geographic compactness, the number of interviewers, workload, projected 2020 population, and difficulty of data collection (2010 non-participation and hard-to-count scores). The selection of RO cities also was data-driven, with two prerequisites – existing RO cities were preferred over new cities, and the RO city had to be in the region it serves. The criteria for city selection included large population, proximity to hard-to-count populations, centrality within the region, an educated population, low locality pay, and low lease costs. Candidate cities with the highest scores on these criteria were selected. Mesenbourg remarked that this data-driven, non-political approach was a key to selling the results.
Mesenbourg also described and touted the changes in the management of data collection. Currently, field reps work for survey statisticians and program coordinators in the office, and survey statisticians are responsible for all aspects of a specific survey. The restructured offices will be staffed with two Assistant Regional Directors, four Program Coordinators, office-based Survey Statisticians, and support staff to accommodate an increased workload. Home-based Survey Statisticians and Field Supervisors (a new position) will manage the field Representatives.
The transition to the new RO structure has begun. Posting for new positions began in July, training for new duties begins in October, field staff transition to the new supervisory structure in seven waves from January through November 2012, and the entire transition will be complete no later than January 2013.
Asked about savings, Mesenbourg said the new structure is projected to save $18 million per year starting in 2014, quickly recovering the estimated transition cost of about $30 million. The Census Bureau is covering all transition costs (not passing any costs to the survey sponsors). The six eliminated offices will close January 1, 2013 through a reduction in force affecting about 320 people. Some will be eligible for retirement opportunities, and others can apply for positions in the remaining ROs and elsewhere. Additional benefits – such as career counseling, job placement, financial planning, and other services are being provided to staff impacted by the closings.
Asked about the reaction of local governments and congressional representatives in the affected areas, Mesenbourg acknowledged that people are not happy, but said there has been less objection than expected. Given the difficult environment, the need for such change is understood, and again, the data-driven approach seems to have dampened negative reactions.
Redistricting and Reapportionment Following the 2010 Census
Kim Brace, president of Election Data Services, explained that the company has been working with states on redistricting since 1979, and provides other election-related data, such as voting equipment, which became of interest following the 2000 election.
Kimball Brace. Election Data Services, Inc.
Brace commented that just as “cameras don’t make pictures – photographers do,” redistricting is a human rather than an automated process. And with the 2010 elections changing the party in control of a number of state legislatures, redistricting will be influenced, and often executed by technicians with limited experience (it seems many who worked on redistricting after 2000 want no part of it in 2010).
Before proceeding, Brace clarified some definitions, describing reapportionment as the allocation of districts to an area (such as congressional districts to states), and redistricting as the crafting of district configurations within an area. These could include congressional, state legislative and county council districts, city wards, and similar areas. Redistricting is usually accomplished once per decade.
A major factor guiding redistricting is the one-person-one-vote principle, which requires districts of equal size. The standard is especially strict with respect to congressional districts, and some find it surprising that the size calculations are based on total persons rather than population by age or other qualifiers. Redistricting also is guided by the Voting Rights Act, which mandates that districts reflect the strength of minority populations. Brace also described a “Kitchen Sink” factor, consisting of anything else one could possibly consider in redistricting. Courts have upheld a wide variety of factors.
Brace then explained that the data used in redistricting are a complex mix of block level census counts, spatial data from TIGER/Line, and data reported for political boundaries, such as precincts and electoral districts. The complex definitions and tabulations of census data on race and Hispanic ethnicity add to the challenges. Even the enhancements to TIGER spatial information are a complication, as 2010 and 2000 block boundaries do not overlay well. Even where blocks are unchanged, boundaries are different due to the enhanced precision in 2010, and this makes it difficult to determine where data are comparable. On the positive side, the 2010 TIGER boundaries are now accurate enough for overlay with parcel data and aerial images.
Brace finished with a report on the status of redistricting – which states are done, which have completed legislative districts only, which have completed congressional districts only, and which are “in progress.” It is a real mix at this point, and lawsuits are in progress all over the place. Brace remarked that “it never stops,” and added that redistricting provides little opportunity for vacations, but steady employment, for those involved. The process has been described as one of outstanding legislative warfare, after which the combatants go to court, and ask a judge who won.
A question was raised about a case in Texas that seems to hinge on whether the ACS is part of the census (ACS is of interest because it has data on citizenship). Brace commented that the real question is how good ACS data would be at the block level. He noted that the major vendors of redistricting software have not included ACS data, but that some states have acquired ACS data, and seek to apply them in redistricting. Presumably, this would require the use of ACS data at the block level, and presumably that would be accomplished by mechanically distributing the already severely stretched ACS block group data down to census blocks. The practice had many COPAFS heads shaking, but Brace noted that many states are intent on using citizenship data in redistricting, and many courts are open to the possibility.
Concerns From COPAFS Constituencies
Mary Jo Hoeksema (PAA Government and Public Affairs) wanted to make sure the attendees were aware of a briefing that will highlight the uses and importance of the Economic Census. The briefing is scheduled for September 26 at Rayburn House Office Building.