New Alternatives for Estimating Net Migration to the United States Using the American Community Survey Alexa Kennedy-Puthoff David Dixon Sonya Rastogi.

New Alternatives for Estimating Net Migration to the United States Using the American Community

Survey

Alexa Kennedy-Puthoff

David Dixon

Sonya Rastogi

Dean Judson Population Division, U.S. Census Bureau

Presented at the United Nations Expert Group Meeting on Measuring International Migration, December 2006.

2

Estimates of International Migration to the United States

3

Population Estimates The U.S. Census Bureau produces annual estimates of the

population and demographic components of change (births, deaths, and migration). These estimates are published at the national, state, and county level.

Estimates are used in: federal funding allocations, as denominators for rates time series, and as survey controls.

One component of the population estimates is net international migration, defined as the net movement across U.S. (50 states and the District of Columbia) borders.*

* The movement of U.S. military personnel across U.S. borders and the in-migration of native U.S. citizens who are not members of the military are included in the population estimates separately from the net international migration component.

http://www.census.gov/popest/estimates.php

4

600700800900

1,0001,1001,2001,3001,4001,500

Estimates of Net International Migration,

1990-2006 (in thousands)

*

5

Net International Migration (NIM) The Net International Migration component of the

population estimates consists of three parts:1. Net migration of the foreign-born population (NFB)2. Net movement between Puerto Rico and the U.S.

(50 states and D.C.) (PR)3. (Net) emigration of native population (NEMIG)

NIM = NFB + PR – NEMIG

NFB accounts for over 90 percent of the total NIM component. This research focuses on the construction of this component.

6

Using ACS Data to Estimate Net International Migration

7

What is the American Community Survey (ACS)? Replaces census sample data (long-form

questionnaire)

Content similar to census long form

Samples every year and spreads sample over 12 months• Monthly data collection• Annual estimates• Low levels of geography

8

ACS Questions

Residence One Year Ago

Citizenship Status

Year of Entry

Native

Foreign born

9

Weighted Estimates of the Population

Citizenship Imputation Rates

Source: American Community Survey, 2000-2004

universe: total population

2000 2001 2002 2003 2004Total Population 273,643 277,018 280,540 282,910 285,691

Native 243,370 245,535 247,491 249,376 251,412Foreign Born 30,274 31,482 33,049 33,534 34,280

Naturalized U.S. Citizen 12,256 12,740 13,536 13,893 14,399Not a U.S. Citizen 18,017 18,743 19,513 19,641 19,880

ACS DATA YEARCITIZENSHIP STATUS

(in thousands)

2000 2001 2002 2003 2004Not Imputed 99.5 99.6 99.6 99.6 99.7Imputed 0.5 0.4 0.4 0.4 0.3

IMPUTATION OF CITIZENSHIP STATUS ACS DATA YEAR

10

Single Year Change(Current Method)

11

Single Year Change The current method used to estimate net

migration of the foreign born Based on the change in the number of

foreign born counted in ACS in time 1 compared with time 2, with an adjustment for deaths to the foreign born during the period

PR

Net migration between the U.S. (50 states and D.C.) and Puerto Rico

Single Year Change

Change in FB population during interval.

SYC = FB t2 – FB t1

Deaths

Deaths to FB population during interval

+ =+ NIM

Net International Migration

NEMIG

Native emigration

-

12

Single Year Change: Estimates and Bounds

universe: total foreign-born population, t1 and t2

(in thousands)


* Data for 2000-2002 were re-weighted to a consistent series of population estimates.The bounds shown here represent a 90-percent confidence interval.

Estimate Lower Bound Upper Bound2000 - 2001 1,209 901 1,5162001 - 2002 1,567 1,215 1,9182002 - 2003 485 128 8422003 - 2004 746 418 1,074

PeriodSingle Year Change (original weights)

Estimate Lower Bound Upper Bound2000 - 2001 1,087 778 1,3962001 - 2002 1,090 741 1,4392002 - 2003 897 542 1,2532003 - 2004 746 418 1,074

PeriodSingle Year Change (revised weights*)

13

Problems with NFB Estimates Based on Single Year Change

High standard errors Sensitive to changes in survey population controls Annual reweighting of ACS data required to provide

consistent series of NFB estimates

14

Year of Entry

15

Year of Entry Was Prior Calendar Year Estimates utilize data on the year of U.S. entry

of the foreign born The “Year of Entry Was Prior Calendar Year”

universe for NFB contains those who indicated they entered the U.S. in the year prior to the survey

PR

Net migration between the U.S. (50 states and D.C.) and Puerto Rico

Year of Entry Was Prior Calendar Year

Foreign born who have a year of entry that is the year prior to the survey.

FBEMIG

Emigration of the foreign-born population during interval

=+ NIM


NEMIG

Native emigration

--

16

Year of Entry Was Prior Calendar Year:

Estimates and Boundsuniverse: foreign-born population, entered in year prior to the survey

(in thousands)



Estimate Lower Bound Upper Bound1999 - 2000 2000 1999 1,574 1,501 1,6472000 - 2001 2001 2000 1,647 1,578 1,7162001 - 2002 2002 2001 1,445 1,379 1,5122002 - 2003 2003 2002 1,260 1,206 1,3142003 - 2004 2004 2003 1,218 1,161 1,276

Estimate Period

Survey Year

Entered U.S. In Year of Entry (original weights)

Estimate Lower Bound Upper Bound1999 - 2000 2000 1999 1,576 1,503 1,6492000 - 2001 2001 2000 1,654 1,585 1,7232001 - 2002 2002 2001 1,427 1,361 1,4932002 - 2003 2003 2002 1,260 1,206 1,3142003 - 2004 2004 2003 1,218 1,161 1,276

Estimate Period

Survey Year

Entered U.S. In Year of Entry (revised weights*)

17Source: American Community Survey, 2000-2004

Year of Entry:Imputation Rates

universe: foreign-born population

(in percent)

universe: foreign-born population, entered in year prior to the survey

(in percent)


IMPUTATION OF YEAR OF ENTRY ACS DATA YEAR


IMPUTATION OF YEAR OF ENTRY ACS DATA YEAR

18

Problems with NFB Estimates Based on Year of Entry

High imputation rates for Year of Entry Conceptual ambiguity / survey question

interpretation difficulties Method requires an estimate of foreign-born

emigration

19

Residence One Year Ago

20

Residence One Year Ago Was Abroad Estimates utilize data on the residence of the

foreign born in the year prior to the survey The “Residence One Year Ago Was Abroad”

universe for NFB contains those who indicated they were abroad (outside the U.S. or U.S. outlying areas) one year ago

PR

Net migration betweenthe U.S. (50 states and D.C.) and Puerto Rico

Residence One Year Ago Was Abroad

Foreign born, who were “abroad” one year ago.

FBEMIG

Emigration of the foreign-born population during interval

=+ NIM


NEMIG

Native emigration

--

21

Residence One Year Ago Was Abroad: Estimates and Boundsuniverse: foreign-born population, residence one year ago was abroad

(in thousands)



Estimate Lower Bound Upper Bound1999 - 2000 2000 1999 1,420 1,354 1,4852000 - 2001 2001 2000 1,421 1,363 1,4782001 - 2002 2002 2001 1,228 1,170 1,2862002 - 2003 2003 2002 1,025 977 1,0722003 - 2004 2004 2003 1,124 1,078 1,171

Estimate Period

Residence One Year Ago (original weights)Survey Year

Residence Was "Abroad" In

Estimate Lower Bound Upper Bound1999 - 2000 2000 1999 1,424 1,358 1,4902000 - 2001 2001 2000 1,421 1,355 1,4872001 - 2002 2002 2001 1,212 1,155 1,2692002 - 2003 2003 2002 1,025 977 1,0722003 - 2004 2004 2003 1,124 1,078 1,171

Estimate Period

Residence One Year Ago (revised weights*)Survey Year

Residence Was "Abroad" In

22Source: American Community Survey, 2000-2004

Residence One Year Ago:Imputation Ratesuniverse: foreign-born population

(in percent)

universe: foreign-born population, residence one year ago was abroad

(in percent)


IMPUTATION OF PLACE OF RESIDENCE ONE YEAR AGO

ACS DATA YEAR


IMPUTATION OF PLACE OF RESIDENCE ONE YEAR AGO

ACS DATA YEAR

23

Problems with NFB Estimates Based on Residence One Year Ago No residence rules for prior residence Method requires an estimate of foreign-born

emigration

24

Comparisons

25

Residence One Year Ago- and Year of Entry- Based Inflows, and Single Year Change (Net Movement)

350

550

750

950

1150

1350

1550

1750

2000 2001 2002 2003 2004

Year of Entry Was Prior Calendar YearResidence One Year Ago Was AbroadSingle Year Change, revised weights

(in thousands)

26

Differences in Quality and Robustness Among Estimates

SEROYA<SEYOE<<SESYC

For the foreign born, imputation rates of place of residence one year ago are lower than those of year of entry

27

Next Steps

28

Components of NIM Warranting Further Investigation

Emigration

Coverage

Puerto Rico

29

Ideas for Estimating Emigration

Record Linking Methods• Current Population Survey (CPS) method• Administrative Records (ADREC) method

Multiplicity Survey Methods

Data Exchange Methods

30

Ideas for Estimating Emigration

Residual Methods• 10-year• 1-year ACS using foreign-born population

counts• 1-year ACS using internal migration

information

31

Sources of Undercoverage of the Foreign Born

Complex Housing Arrangements Irregular Housing Units Language Barriers Fear of Government and Immigration

Authorities High Degrees of Mobility

• ACS Two-Month Residency Rule

32

Estimation of Undercoverage of the Foreign Born

Various Methods and Assumptions Wide Ranging Estimates for Segments

of Foreign Born• Example - Range of 10 to 50 percent

undercount for unauthorized

33

Immigration Statistics Staff

(301) 763-2411

Dean H. Judson, Chief

http://www.census.gov/population/www/socdemo/foreign.html

New Alternatives for Estimating Net Migration to the United States Using the American Community Survey Alexa Kennedy-Puthoff David Dixon Sonya Rastogi.

Documents