The U'S' Census Bureau Population Estimates Program

1
The U.S. Census Bureau Population Estimates
Program
Victoria A. Velkoff U.S. Census Bureau APDU
Annual Conference September 25, 2008
2
Outline

• What population estimates we produce and why
• Basic methodology and data sources
• Challenges and special censuses
• Research we are doing

3
Annual Estimates

• Population
• National by age, sex, race, and Hispanic origin
• States by age, sex, race, and Hispanic origin
• Counties by age, sex, race, and Hispanic origin
• Incorporated places and minor civil divisions
  (total population only)
• Puerto Rico Commonwealth and municipios by age
  and sex
• Housing units
• States
• Counties

4
Why Do We Produce Estimates? Federal Mandate

• The Intercensal Estimates Program is required
  under U.S.C. Title 13, Section 181
• During the intervals between each census of
  populationthe Secretary, to the extent feasible,
  should annually produce and publish for each
  State, county, and local unit of general purpose
  government which has a population of fifty
  thousand or more, current data on total
  population and population characteristics and, to
  the extent feasible, shall biennially produce and
  publish for other local units of general purpose
  government current data on total population.

5
Uses of Population Estimates

• Allocation of over 300 billion in federal funds
  each year
• Controls for surveys
• Current Population Survey
• American Community Survey
• Survey of Income and Program Participation
• American Housing Survey
• Denominators for statistical rates
• Birth and death rates (NCHS)
• Per capita income (BEA)
• Other Federal Uses and Program Planning
• Department of Education, National Cancer
  Institute
• Location of public and private services

6
Current Methods

• National level
• Cohort-component method
• P2 P1 B - D NIM (International Migration)
• State and county level
• Component of change method
• P2 P1 B - D NM (Domestic and International)
• (controlled to the national estimates)
• Subcounty level
• Distributive housing unit method
• County population is distributed to subcounty
  parts based on updated estimates of housing

7
Data Sources for Base Population and Vital
Statistics

• Base population
• Latest census in current geography
• Vital statistics
• National Center for Health Statistics (NCHS)
• Federal-State Cooperative Program for Population
  Estimates (FSCPE)

8
Data Sources for Migration

• Domestic migration
• Federal tax returns (IRS)
• Medicare
• International migration
• American Community Survey
• Foreign born net migration
• Net movement to/from Puerto Rico
• Other sources
• Native emigration

9
Challenges and Special Censuses

• Challenges
• An alternate estimate provided by the
  governmental unit and accepted by the Census
  Bureau
• Usually based on housing unit data
• Special Census
• A basic enumeration of population, housing units,
  and group quarters conducted by the Census Bureau
  for governmental units

10
Housing Unit Method

Populationt HUt PPHt Ot GQt Where HU
Number of housing units PPH Persons per
household O Occupancy rate (we say vacancy as
well) GQ Group Quarters Population
11
2000-2006 Challenge Results
12
Research Efforts in the Population Estimates
Program

• HUBERT - Housing Unit Based Estimates Research
  Team
• Current Methodology Ongoing efforts to improve
  our current methodology

13
HUBERT Background

• Census Bureau sponsored a data user conference in
  July 2006
• Consider alternative methods and data sources for
  producing population estimates
• Emphasis on the housing unit method
• Research use of the American Community Survey
  (ACS) and Master Address File (MAF) in population
  estimates

14
HUBERT Areas of Research

• Evaluations and basic comparisons of housing unit
  method and administrative records (ADREC) method
  for estimating population
• Review of current methodology for estimating
  housing units
• Research into ways to estimate persons per
  household (PPH) and vacancy

15
HUBERT Topic 1 Evaluation

• Created a set of county household population
  estimates using the housing unit method
• Compared these estimates to Census 2000
• Used mean absolute percent error (MAPE), mean
  algebraic percent error (MALPE) and percent
  difference as evaluation measures

16
In addition to that evaluation

• Evaluated two alternative set of estimates
• ADREC (the Census Bureaus 1990s based household
  population estimates)
• Average of the two methods
• Compared MAPEs, MALPEs, and percent differences
• County size categories
• County growth between 1990 and 2000

17
Mean Absolute Percent Errors (MAPEs) and Mean
Algebraic Percent Errors (MALPEs)
18
Method Closer to Census 2000
19
MAPE by Population Change 1990 to 2000
20

• HUBERT Topic 1 Findings
• ADREC method produced estimates that were closer
  to Census 2000 for all size and growth categories
• In 30 percent of counties, the housing unit
  method was closer to Census 2000
• Current estimates of each component (HU, PPH, and
  Vacancy) are necessary to obtain more accurate
  Housing Unit-Based estimates

21
Topic 2 Review of Housing Unit Estimates

• Looked at input for our current housing unit
  estimates (e.g., housing loss, building permits,
  etc.)
• Examined the Master Address File to see if we
  could use it in our estimates
• Looked at cadastral (tax assessor) files

22
HUBERT Topic 2 Findings

• There are regional differences in housing unit
  loss rates by year-built categories not accounted
  for in the current methodology for estimating
  housing unit loss.
• By excluding specific types of addresses and
  including others in the MAF extracts, the total
  number of addresses in the MAF extracts becomes
  more consistent with the housing unit estimates
  at the national and county levels.

23
HUBERT Topic 3 Persons per Household and Vacancy

• We examined using American Community Survey
  estimates of persons per household and vacancy
• We also examined using regression models to
  estimate persons per household and vacancy

24
HUBERT Topic 3 Findings

• For a large number of counties, the usefulness of
  the one-year estimates of PPH from the ACS for
  population estimates is limited by the ACS sample
  size.
• As was found in previous research, more accurate
  estimates of PPH can be obtained through the use
  of regression models and extrapolations rather
  than by simply using the values from the previous
  census.

25
HUBERT Next Steps

• Production
• Incorporate new housing unit loss rates into 2009
  housing unit estimates
• Develop strategy to include tallies from the MAF
  in
• housing unit estimates review
• Compare vintage 2008 ADREC method estimates with
  Housing Unit-based estimates as part of estimates
  review
• Continue examination of the ACS as a source of
• estimates of PPH and occupancy rates

26
Next Steps in Terms of Population Estimates
Research

• Goal To conduct and document research into the
  use
• of multiple methods for producing population
  estimates
• so that informed methodological decisions can be
  made post 2010
• Continue to collaborate with our state partners
  and external stakeholders
• Look at other methods for population estimates
  (e.g., ratio correlation)
• Develop an evaluation plan for population
  estimates and the 2010 Census

27
Improvements to Current Method

• Vintage 2007 Changes
• Net international migration
• Change to person-based migration processing
• New data source for military data (DMDC)
• Vintage 2008/2009 Changes and Research
• Birth cohorts since 2000 by race and Hispanic
  origin
• Old-age mortality

28
Summary

• We produce estimates
• disaggregated by age, sex, race, and Hispanic
  origin
• at low levels of geography (e.g., counties)
• for many purposes
• We are researching
• alternative methods and data sources for
  population estimates
• improvements to our current method