National Projections Program, 2005 Population Projections Branch Population Division U.S. Census Bur

1
National Projections Program 2005Population
Projections BranchPopulation DivisionU.S.
Census Bureau
2
Two series under consideration

• Interim projections released in spring of 2004
• Population forecasts under development future
  release

3
Interim projections production overview

• Initially produced by International Program
  Center (IPC) for International Database (IDB)
  resident plus Armed Forces overseas by age and
  sex 2000 through 2050.
• Race and Hispanic origin detail added early
  2003 reduced to resident population
• Resident population by age sex race and
  Hispanic origin released spring of 2004

4
Interim projections technical attributes in brief

• Incorporated Census 2000 as base
• Adapted component assumptions from existing
  projections released early 2000.
• Corrected to population estimates component
  assumptions for 2001 by mixing high medium and
  low series
• Major change in imputation of race of child
  based on OMB 1997 guidelines check all that
  apply

5
Interim component assumptions

• Components not race-specific race and ethnic
  composition effects from old series present but
  not updated
• Very slight decrease in fertility from old middle
  seriesstill a rise above TFR2.2 then a slight
  decline
• No change in mortality from old middle e081.2
  for males e086.7 for females (2050)
• Slight increase in international migration from
  old middle series but no revision of decline in
  early 2000s

6
Race of child the problem

• OMB guidelines yield 62 categories of 5 major
  race and 2 Hispanic origin responses (25-1)2
• Interim projections have only 8 cross-categories
  of race and origin
• Race of child not generally same as race of
  parent race/origin groups not closed with
  respect to reproduction

7
Race of child assumptions

• Transmission of race and Hispanic origin from
  parent to child follows census universe not
  biological rules
• We can approximate transmission from structure of
  households
• Mixed-race childbearing stays at baseline levels
  except for demographic supply effects

8
Race of child census input

• Census 2000 (100) tabulation of households with
  parents and non-adoptive children under 18 by
  race and Hispanic origin
• Imputed race of missing fathers by race of mother
  and race of children
• Race of child may not agree with either Mom or
  Dad OK

9
Race of child imputation

• Apply fertility rates to projected women to yield
  new mothers by race and origin
• Tabulate potential fathers by race and origin
  and age of mother
• Measure shift in distribution of potential
  fathers from base date to target date via
  multiple logits
• Impose multiple-logit shift on actual
  distribution of fathers in census household file
  to yield joint distribution of mothers fathers
  by race and Hispanic origin
• Impute race of child from joint distribution
  based on census household file

10
Forecasts under development

• Greater emphasis on historical time series in
  formulating assumptions
• More systematic approaches to derivation of
  future trends in components
• Measurement of uncertainty prediction intervals
• Min-max race but technical innovations
  directed to three large groups that are close to
  old definition
• Any innovation in method is likely to appear as
  an experimental product