News from COPAFS
Federal Statistical Agency Budgets
It appears that the Census Bureau is not going to get its
requested funding for the American Community Survey (ACS),
the continuous sampling method that is supposed to replace
the decennial census long form. So the writing is on the wall-there
will be no full implementation of the ACS in 2003. In fact,
there is a possibility the long form is not dead, and if it
is used in the next census, there will be major changes in
content. The Senate mark, if upheld, eliminate the 2002 economic
census, even though is one of the building blocks of the national
income accounts. The BLS has recommended full funding, but
BEA is getting only a 4.1% increase for salaries. In order
to absorb $1.8 million in base cuts, BEA will have to consider
scaling back work to improve the national income accounts,
cutting the RIMS program, dropping the plan to accelerate
the release of quarterly estimates of state personal income,
and cutting the REIS program.
Metropolitan Area Measurement
New metropolitan areas (or Core Based Statistical Areas) are
to be announced next year, and the Office of Management and
Budget has notified the congressional delegations of those
areas that have the option of combining. Census will use a
new urban classification system. "Urban clusters"
will replace "urban places," which are limited to
areas with a population of 50,000 or more. Michael Ratcliffe
from the U.S. Census Bureau explained at the fall COPAFS meeting
that urbanized are (UA) definitions are remaining similar
to those applied after 1990, but that the creation of the
smaller urban clusters is a major new development. Urban clusters
(which serve as the cores of the new metropolitan statistical
areas) are defined with criteria similar to those used for
urbanized areas, but do not require populations of 50,000
or more (10,000 to 49,000 will do). Urban clusters, therefore,
replace some of the urban places of 2,500 or more population
from the 1990 definitions. With increased suburbanization,
the 50,000 population requirement became a problem since fringe
areas with urban density were being classified as rural. The
cities around which they were growing had not grown enough
for the area to attain UA status. The new definitions are
not as constrained by place boundaries, and places can now
be part urban and part rural.
Ratcliffe described how urbanized areas were once defined
by hand using paper maps, then after 19910 they were prepared
with in-house GIS software. There was always an element of
judgment or subjectivity in applying the criteria and debate
over what constitutes "jumps" or "hops"-the
inclusion of low density area to eliminate enclaves. A goal
for 2000 was to establish a more automated and objective process
for delineating urban areas. For 2000, the conditions for
identifying jumps and hops have been revised, and areas are
defined by block group rather than blocks-with the exception
of block groups with land area in excess of 2.5 square miles
(usually in fringe areas), where blocks are still used. The
use of block groups better insures the inclusion of central
business districts that have low residential population density.
Even with automation, the delineation of urbanized areas (and
now urban clusters) remains a massive undertaking. The process
took 40 computers two months to complete.
Ratcliffe also addressed the forthcoming CBSA definitions-where
the designation of "combined statistical areas"
(that is the formal term after considerable debate) is a significant
new feature. Combined statistical areas will consist of metropolitan
and micropolitan statistical areas combined on the basis of
commuting ties or interchange. An interchange in excess if
25% automatically creates a combined statistical area, while
exchanges greater than 15% but less than 25% provide the option
for combination-based on local opinion. The new CBSAs will
be announced by June 2003-or earlier if the designations are
completed before then. For more about urban definitions, see
www.census.gov/geo/www/ua/ua_2ki.html.
Recent Results from the National Household Survey on Drug
Abuse
Joe Gfoerer of the Substance Abuse and Mental Health Services
Administration spoke at the fall meeting about his agency's
2001 National Household Survey on Drug Abuse (NHSDA). The
survey has been annual since 1990, but a 1999 redesign has
enabled the presentation of state data. Respondents are drawn
from the civilian noninstitutional population age 12 and older,
and data are collected through face-to-face interviews. The
2001 sample was 68,929.
Trends show modest but widespread increases in use rates
for 2001 (over 1999 and 2000), and Gfoerer commented that
they are not sure why. The biggest increase is in the age
18-25 range, with growth in older ages described as in part
a cohort effect-as aging baby boomers bring their relatively
high rates to older age categories. The survey also identified
that increasing use rates are accompanied by decreasing perceptions
of risk. Gfroerer then presented results related to alcohol
and tobacco (showing maps of state data), as well as annual
numbers of new users of marijuana over the 1965-2000 period.
He also described some of the mental health data provided
by the survey. Data from the survey are on the Web site www.samhsa.gov
and in a two-volume book, available in print and on the Web
in PDF format.
Developments at the IRS's Statistics of Income Divisions
Thomas Petska, director of the Statistics of Income (SOI)
division, spoke at the fall meeting of COPAFS, The division,
which is technically an administrative rather than a statistical
agency, gets all of its date from administrative records.
SOI's mission is to design and conduct statistical studies,
examine findings, and disseminate results. Their major users
include Treasury, the Congressional Joint Committee on Taxation,
BEA, the Census Bureau, the IRS, other government agencies,
researchers, and the public. Users of the data include revenue
estimation, policy analysis, and the Census Bureau's population
estimates.
Petska described the process through which IRS returns are
received, processed, and edited to provide the data for SOI
products. These products are based on a sample drawn from
the IRS master file. He explained that the IRS master file
contains limited content from all returns, and that SOI uses
a sample of master file records, but with more extensive data
content. The data, which he described as micro data with weights
to national level, are all preaudit. Out of 133 million returns,
the SOI sample includes about 225,000, with high income returns
sampled at a higher rate-up to 100% for returns of $10 million
or more. Among the challenges faced by SOI is their status
as a "distant and second" priority at IRS, with
primary customers in other agencies. They are also constrained
by the limits of the tax form as a data collection instrument,
and compilation is complicated by factors including the decentralized
structure of IRS, the complexity of the tax code, filing extensions,
and tax avoidance/evasion. Petska also noted that raw data
from returns contain many errors, and require extensive editing
efforts. Keeping up with IRS technological changes presents
another set of challenges.
Objectives for the future include the development of data
editing systems that are less "people-intensive,"
expanded use of master file data, protecting the confidentiality
of IRS records, and the development of new means for data
access. SOI products are currently disseminated directly to
customers, and are available on their Web site-which he admitted
has room for improvement. The overall goal is to get more
data more quickly to more customers in more media. The Web
site for SOI is www.irs.gov/tax_stats.
You can write Petska at Director, SOI Divison, P.O. Box 2609,
Washington, D.C. 20013-2608, or e-mail him at thomas.b.petska@irs.gov.
|
| 
| 
| 
| 
| 
| 
