module 6 - international organization for migration

MODULE 6

Strengthening administrative data sources: towards harmonization and standardization of migration data

Jason Schachter, Chief, Net International Migration, US Census Bureau Jason Gagnon, Lead Economist on migration and skills, OECD Development Centre

Nougbodohoue Samson Bel-Aube, Statistician, STATAFRIC African Union (AU)

PART 1

Integration of Administrative and Household Survey data at the US Census Bureau to Measure International Migration

Jason Schachter

Administrative data• Data collected by Federal agencies for some administrative purpose

• Often not collected with the purpose of measuring international migration

• Advantages• Timely• Complete coverage (as long as included in the system)• Low cost• Often includes a number of characteristics (but often limited)• Can be linked if personal identifiers are available

5

Strengthening administrative data sourcesPart 1: Integration of Administrative and Household Survey data at the US Census

Bureau to Measure International Migration_Jason Schachter

Data Integration• The process of combining two or more data sources to produce

statistical outputs• Macro: The combination of data based on aggregates (statistics) of individual-

level records. • Micro: the integration of data based on record linkage or statistical matching

of individual-level records using key identifying variables. • Difference between Micro- and Macro-integration

• Micro-data integration creates new combined data sets, which can produce statistics• Macro-data integration creates new statistics (estimates) from outputs from multiple

data sets• Often used when micro-data integration is not possible

6



Why data integration is needed to improve measurement of migration statistics: Macro• Multiple data sources yield different results

• Due to inherent challenges measuring international migration (different concepts, universes, coverage, etc.)

• Data source availability• Micro-data integration is not possible• Lack of specific information (e.g. stock data available, but not flows; lack of geographic detail, lack of timely

information, etc.)• Specific migrant groups missing in regular data sources (e.g. coverage of irregular migrants, refugees, circular

and temporary migrants, etc.)• Missing characteristics of migrants in a particular data source (e.g. demographic characteristics, legal status,

etc.)

• Respond to recent migration events which regular data measurement lags or cannot measure adequately

• Natural disasters/pandemics• Humanitarian crises (e.g. war/mass refugee movements)• Sudden drastic policy changes

7



Why data integration is needed to improve measurement of migration statistics: Micro

• Combined data sets provides richer information about migrant populations, including data without migrant identifiers

• Different administrative sources• Surveys to administrative sources• Census to survey or administrative sources• New data sources (such as big data) combined with traditional data sources

• Compilation of data sets for policy analysis or estimation, including longitudinal analysis• Migrant integration indicators • Economic impact of migration• Change of legal or economics status over time

• Synchronizing/harmonizing variables, terminology and timeframes across different sources, particularly helpful for administrative data collected by different registries.

• Reduced costs for data collection/processing• Reduced response burden• Improved quality of data 8



Macro-Data Integration Methods• Additive methods (data compilation)

• Use different data sources to produce different sub-components of international migration estimates

• Combination of different estimates to produce a single estimate (data triangulation)• Adjust migration estimates from one data source using estimates from one or more other data sources

• Combination of different data sources (borrowing strength from one source to improve another source)• Applications

• Improve conceptual coherence (e.g. flight data is very timely but conceptually weak)• Add information to an existing source• Improve the balance between accuracy and timeliness• Improve small areas and geographies estimates

• Examples: Combine different data sources, with differing levels of accuracy• To produce migration estimates at different levels of geography (e.g. survey data for national totals/administrative data for sub-national)• To produce migration estimates by various characteristics of the population (e.g. Bayesian methods to provide estimates for areas/characteristics with scarce

observations)

• Examples: Information from one data set used to inform another data set• Combining migration flow data gathered by other countries to estimate international migration flows• Characteristics missing from one data set are imputed via the identification of similar individuals on another data set (e.g. trained logistic regression models to

identify the probability that an individual has a certain characteristic, such as refugee or legal status) 9



Micro-data integration methods• Record linkage of Administrative data

• PIN/SSN (especially countries with population registers)• Family identifiers (names)• Combination of multiple personal identifiers, aka deterministic linking• Probabilistic linking of administrative sources

• Countries without PIN/SSN• Statistical matching with survey and census data• Parametric and nonparametric approaches• Integration of multiple survey databases• Integration of survey and census data• Integration of administrative data with census data

• Data quality assurance

10



US Census Bureau: Net International Migration (NIM) Estimates

• Produce annual estimates of international migration flows to and from the United States

• Geographic Levels: Nation, state, and county• Characteristics: Age, Sex, race/Hispanic origin

11



NIM Components

12

Foreign-born immigration

Foreign-born emigration

Net Puerto Rico migration

Net Native-born

migration

Net International

Migration



Presenter

Presentation Notes

Quickly go through the different components of the NIM estimate they see in the release.

U.S. Census Bureau SurveysAmerican Community Survey

(ACS)

• Implemented in 2005, estimates released annually

• Large sample allows for studying small levels of geography

• Relevant questions: country of birth, U.S. citizenship status, year of naturalization, year of entry, residence one year ago, race, ethnicity, language spoken at home

Current Population Survey (CPS)

• Primary data source for U.S. labor force statistics

• Began in 1940’s so allows for time series analyses

• Contains most foreign bornitems in ACS plus parental place of birth & reason for moving

Survey of Income Participation (SIPP)

• Longitudinal survey; each panel is about four years long

• Relevant questions: whether born in U.S., citizenship status

• Topical module contains information on immigration status upon entry to the U.S., whether and when status changed to permanent resident, country of previous home



Presenter

Presentation Notes

ACS is primarly Survey data has some limitations for measuring flows due to sample. ACS subnational, CPS and SIPP national estimates only. ACS 1-year files allow for analyses of areas with populations of 65,000 or more, and 5-year files allow for analyses down to the block level. 2016: 3.5 million address samples, 2.2 million interviews, 206,000 GQ selected, 60,000 interviewed. CPS: CPS has a sample size of about 60,000 households. Although labor market information is central to the CPS, the survey provides a wealth of other demographic, social, and economic data that are widely used in both the public and private sectors. This includes several questions of interest to migration researchers, such as country of birth, U.S. citizenship status, and the year the respondent came to live in the United States. Most notably, the CPS is the only U.S. Census Bureau survey to include questions on parental place of birth for all respondents. SIPP: Sample size ranging from approximately 14,000 to 52,000 interviewed households. The duration of each panel ranges from 2 ½ years to 4 years. SIPP 2008 was 5 years long. Most recent panel began in 2014 with approximately 27,000 households.

Flow Estimates from Administrative Data• Migration flows can be estimated via visas issued, landing cards, legal resident status changes,

population registration, etc. • Administrative data are not well integrated into the U.S. statistical system, particularly for

international migration

• U.S. administrative data collected by many different agencies• Department of Homeland Security Office (DHS), State Department, Department of Justice, etc.

14



Presenter

Presentation Notes

About admin data

U.S. Administrative Data Sources on International Migration

15

Internal Revenue Service

Federal Tax Exemptions•Movement•Subnational

totals

Individual Tax ID Number (ITIN)•Movement•Implied non-

citizenship•Implied Year of

Entry

Social Security Administration

Numident•Age•Sex•Country of Birth•Implied Year of

Entry

Citizenship and Immigration

Services

Lawful Permanent Resident (LPR)•Class of Admission

•Country of Birth/Cit

•Demographic Characteristics

•Year of Status Change

Affirmative Asylee

Statistics

Customs and Border

Protection

Arrival and Departure

Information System (ADIS)

Immigration and Customs Enforcement

Student and Exchange

Visitor Information

System (SEVIS)

Bureau of Transportation

Statistics

Airline Passenger Data•Entries•Exits

Land Border Crossing Data•Entries Only

Department of State

Visa Statistics•Year Visa Issued•Country•Visa Class

Worldwide Refugee

Admissions Processing

System (WRAPS)

Department of Justice

Defensive Asylee

Statistics

In use

Needs processing

Not acquired



Presenter

Presentation Notes

Some Census uses, most do not have access to.

Challenges for U.S. Data System• Decentralized Federal statistical system

• Census Bureau, Bureau of Labor Statistics, Department of Homeland Security, State Department, Department of Justice, etc.

• Difficult-to-count migrant groups• Sample size limitations• Non-response• Legal status not collected on most census data

• Refugees• Irregular migrants

• Measurement of emigration• Not collected

• Foreign born• Residual method

• Native born• International data sources

16



Presenter

Presentation Notes

Some of these apply to Mexico and Canada, as well. Mention Disclosure Avoidance?

Challenges for improving international migration estimates at the US Census Bureau

• The American Community Survey (ACS) is our primary data source• Annual survey of about 2 million households

• Sample size limitations for characteristics, particularly at the subnational (county) level

• Pool multiple years of data (impacts recency of estimates)

• Increased concern about non-response• Legal status not collected

• Refugees, Irregular migrants, etc.

• Data quality• Ex: Year of Entry: heaping, first or most recent move, etc.

• Recent events impacting migration (e.g. Hurricane Maria and Puerto Rico, COVID-19, drastic policy changes, etc.)

• Timeliness of data17



Presenter

Presentation Notes

ACS. 3 to 3.5 initial addresses selected, 1.9 to 2.3 million final interviews. + 200,000 GQ, 161,000 interviewed.

Macro-data integration example: Adjustments for the impact of COVID-19 on Net International Migration Estimates

• Needed to measure the impact of an event that was not covered by current data• Prior experience integrating flight data and household survey data to adjust US to Puerto Rico

migration in the aftermath of Hurricane Maria

• Investigated a wide array of alternative data sources to measure the impact of COVID-19 on international migration to/from the United States

18



US Migration Policy Changes Resulting from the COVID-19 Pandemic (March 2020-Onwards)

• Border closures• Closure of US land borders with Mexico and Canada, except for commerce and essential travel (initially implemented March 21,

extended in July)• Flight restrictions from China (February) and European Union (March), for example

• 98% reduction in international flight traffic in April

• Visa restrictions• Visa issuing offices closed outside the United States, greatly reducing number of approved visas (March 20, 2020)

• Executive Orders• Ban travel of non-residents from certain countries• “Pausing” legal immigration (issuance of green cards, family based visas) (April 22, 2020, extended to the end of year)

• Expanded to include most forms of legal migration, especially employment-based (H1B)

• Other• US citizens/legal residents living abroad encouraged to return for fear of border closures (March 2020)

19



Presenter

Presentation Notes

Through December

Measurement• The Population Estimates Program uses American Community Survey (ACS) data on residence one year

ago to produce net international migration (NIM) estimates (July 1st, 2020)• The Vintage 2020 estimates will use 2019 ACS data, which would not measure the impact of COVID-19

• Two tasks to measure impact of COVID-19 on NIM• Develop method to make adjustment to NIM

• Run normal production to develop NIM estimates, and then:A. Do nothingB. Set total monthly NIM to 1/2 for March, "0" for April-June using basic assumption of zero net migration gain/lossC. Adjust total NIM based on other data sourcesD. Adjust individual NIM sub-components based on other data sourcesE. Combination of B-D

• Utilize more up-to-date data sources (March-June 2020)• Limited availability at time of production (September 2020)• Special tabulations from different agencies

20



Presenter

Presentation Notes

Production occurs in September….

Adjustment Methods/Data Sources• Total NIM

• Flight data adjustment based on past historical relationship of net passenger movement and NIM (unsuccessful)

• Foreign-Born Immigration• Visas issued abroad; I-94 arrival data

• Immigration decreased significantly

• Foreign-Born Emigration• Canadian flight data and Mexican border data

• Emigration decreased significantly

• Net Native Migration• Administrative data on permanent and temporary migrants to Canada from the US• US flight data

• Suggest there was not a net loss (large return in March, some net outflow from April-June)

• Puerto Rico• Flight data to/from Puerto Rico and mainland US (March-June “observed” vs “expected”)

• Reduced net outmigration from Puerto Rico 21



NIM Estimates and Net Air Passenger Traffic from/to the United States: 2010-2018

Sources: U.S. Census Bureau Vintage 2010-2018 Population Estimates; U.S. Bureau of Transportation Statistics 2010-2018..

-200,000

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

2010 2011 2012 2013 2014 2015 2016 2017 2018

Net Canada Air Travel Net Mexico Air Travel Net Total Air Travel Census NIM Estimate



Presenter

Presentation Notes

Landborder, not airdata Research into flight data for US vs NIM estimates. Based on current estimates methodology, ACS input data always lags one year during current estimate year. E.g. V2018, uses 2017 ACS data, and project that data forward one year. Flight data and NIM track very well, surprisingly. 2018 our projected year of NIM, based on 2017 ACS. If used flight data, would have shown a decline. 2018 ACS data showed a similar decline as well. Since can get access to flight data earlier than ACS (later September of the following year), could be used to help project our current year estimate. 2016 policy change….. While seems to track NIM on national level, when looking at Mexico, does not work (many more coming to Mexico than coming back)….

23

Data Source: Bureau of Transportation Statistics Form 41, T100 (International) Segment All Carriers; US Census Bureau, Population Division, V2019 Population Estimates



Presenter

Presentation Notes

This shows seasonality of BTS estimates

Flight Data: Inbound and Outbound Passengers

In 2020, inbound and outbound passenger totals fall well below 2019 levels:

• Higher-than-expected net inflow for Feb and Mar (previous slide), but both inbound and outbound passengers declined for those months

• Outbound passengers declining faster than inbound passengers

Data Source: Bureau of Transportation Statistics Form 41, T100 (International) Segment All Carriers

0

2,000

4,000

6,000

8,000

10,000

12,000

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Inbo

und

and

Out

boun

d Ai

r Pas

seng

ers (

x 1,

000)

Month of Travel

2019 Inbound 2019 Outbound 2020 Inbound 2020 Outbound

Inbound and outbound passengers for Apr 2020 fell to 2% of Apr 2019 levels

24



Presenter

Presentation Notes

Update for May

Visas Issued Overseas and Foreign-Born Immigration

• Bureau of Consular Affairs began reporting monthly numbers in March 2017

• Seasonal patterns consistent before March 2020

0

50

100

150

200

250

300

350


Visa

s Iss

ued

(x 1

,000

)

Month of Issuance

2017 2018 2019 2020

Visa issuance in June 2020 fell to 12% of June 2019 levels

Data Source: Department of State, Bureau of Consular Affairs Visa Statistics

Visa issuance in Mar 2020 was 62% of Mar 2019 levels

25



Presenter

Presentation Notes

Walk through slide.

Adjust Foreign-Born Emigration: Country Data

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000


Land

Bor

der T

rave

lers

(x 1

,000

)

Month Surveyed in Mexico

Travelers to Mexico by Land

2019 Entry to Mexico 2020 Entry to Mexico

Travelers in Apr 2020 fell to 41% of Apr 2019 levels

Data Source: Mexican Institute of Statistics and Geography (INEGI), Border Traveler Survey (EVF); Statistics Canada, Border Enumeration from Primary Information Kiosks

0

5

10

15

20

25

30


Inbo

und

Air P

asse

nger

s (x

1,00

0)

Month of Entry into Canada

Canadian Citizens Returning to Canada from US by Air2018 Entry to Canada 2019 Entry to Canada2020 Entry to Canada

Jun 2020 fell to 9% of the average of Jun 2018 and Jun 2019 levels

26



Presenter

Presentation Notes

Describe graphs.

Adjust Net US-Born Migration: Country Data

0

200

400

600

800

1,000

1,200


Perm

anen

t Im

mig

rant

s

Permanent Immigration to Canada by Country of Birth

2019 US Born 2019 Other Foreign Born

2020 US Born 2020 Other Foreign Born

Data Source: Statistics Canada; Immigration, Refugees & Citizenship Canada (IRCC)

-2,000

-1,500

-1,000

-500

0

500

1,000

1,500


Net

Tem

pora

ry M

igra

nts

Net Temporary Migration to Canada (US Born Only)

2018 2019 2020

More US-born temporary migrants were leaving Canada than entering

27



Presenter

Presentation Notes

Describe graphs.

-60000

-40000

-20000

0

20000

40000

60000

January February March April May June July August September October November December

Net

Pas

seng

ers

Calendar Year

Puerto Rico Net Airline Passenger Flow Data: 2016, 2019-2020

2016 2019Source: Bureau of Transportation Statistics, Airline Passenger Data



Presenter

Presentation Notes

Describe graph.

NIM Adjustment Scenarios• Data were limited for making adjustments

• Special tabulations from US Bureau of Transportation Statistics, US Office of Immigration Statistics, Statistics Canada

• Tried multiple methods/scenarios• Flight data could not be used to adjust total NIM

• General assumptions• Very likely that migration declined in March 2020 and remained low during the last quarter (April-June) of

“Estimates Year” 2020 (near net “0”)• Averaged estimates derived from eight different 2020 NIM series based on varying assumptions

applied to March-June 2020.• First four scenarios assume “0” net migration for total NIM, or part of NIM, while last four adjust individual

subcomponents of NIM• “0” net for April-June• ½ NIM for March• 10% monthly levels for sub-components (April-June)• Different estimates for Puerto Rico (based on ratio of observed-to-expected net passenger flights)• Adjustment for net native return in March

29



Presenter

Presentation Notes

Summary of the conditions we were facing to make adjustments, and the final method we used to make the adjustment.

Results/Questions• Range of 87,000 between high and low estimates (8 series)• Changing adjustment factors (.05 or .15) for scenarios had little impact on

NIM estimates (+/- 5,000)• COVID-19 adjustments decreased EY2020 NIM by 24%

• Essentially the same as setting NIM to “0” for 3 months• Net native migration a big mystery

• Return of US Citizens in March, potential departures in April-June?• We will potentially find out the answers once more data become available in

coming months/years (OIS/2020 and 2021 ACS/Puerto Rico Community Survey, Internal Revenue Service tax returns, etc.)

30



Micro-data integration example: Integrated Database on International Migration (IDIM)• Difficulty obtaining detailed information from the Department of Homeland Security on

international migration• US Census Bureau has standing agreements with the Internal Revenue Service (IRS) and Social

Security Administration (SSA) to receive data for use in our estimates, but have not been used to the measure international migration

• Developing an integrated data base which links these administrative sources, as well as with the American Community Survey (ACS). Uses SSA data as “spine”

31



Data Linking across data sets• The Census Numerical Identification File (NUMIDENT) is a dataset of unduplicated SSA records,

containing one record for every person ever issued a Social Security Number (SSN)

• Process to match persons across Census surveys and Federal data• Person Identification Validation System (PVS) • Probabilistic matching to match person data from an incoming file to a reference file (derived from

the NUMIDENT)• Name• Date of birth• Address

• Each matched person record is assigned a Protected Identification Key (PIK)• Unique identifier for each individual• Ensures confidentiality• Person linkage key

• PIKs link individual records between datasets • IRS, NUMIDENT, Census, and ACS linked to create IDIM

32



Presenter

Presentation Notes

PIK RATES. COB TO IRS PIK RATE. ,Probalisitc: Paper, Leach, Jensen,?...really probabilistic? (more of a direct matching…direct method first, then use probabilistic methods)…(probabilistic data linking or stochastic/probablisitc matching to assign value). SSN have highest matches. Parameters for matches set by CARRA (Full name, first, last, middle). Record linkage is both an art and a science. It is a balancing act between link quality, processing speed, and setting optimal parameters. PVS employs its probabilistic record linkage software, Multi-Match first step of the PVS process is to edit data fields to make them homogenous for comparisons between incoming and reference files. Name software….address standardizer…..assigns unique address identifiers (MAFID). Add tons of census info the record. Goes through different passes to match individual on different variables. 86% PIK rate 2010 Census to NUMIDENT. (91%). 96% IRS. ACS 88-90%

IDIM Administrative Data linked to Census

NUMIDENT IRS Census 2010

• Age • Sex • Country of Birth

• Geography • Race• Hispanic origin

*Race and Hispanic origin are imputed for the foreign born who entered after 2010.

33



Presenter

Presentation Notes

Variables gleened from the different data sets Age—date of birth

IDIM Country of Birth Information

• “Country of Birth” (COB) collected from SSA (included on NUMIDENT)• Two steps to edit and clean COB:

1. Maximize record count for each country by combining records• Ex: Vietnam (VM), North Vietnam (VN), and South Vietnam (VS) recoded to Vietnam

(VM)2. Correct erroneous country codes

• Ex: Records coded to China (CH), that list city of birth as Santiago, recoded to Chile (CI)

34



Application: NIM subnational distribution

• Current methodology distributes most NIM components based on recent foreign-born stock population (5-year ACS file)

• Lacks recency and accuracy of characteristics at the county-level

• Possibilities • IDIM to estimate national characteristics and subnational totals and characteristics

(completely replace ACS distributions)• IDIM to estimate totals and characteristics of counties below a certain population

threshold, while continue to use ACS for larger counties and all states (combine IDIM and ACS)

• Keep current methodology, but model county age distribution (or other characteristics) based on IDIM results

35



Presenter

Presentation Notes

4 geographic iterations. Ordered from more DCF to less ACS Replace ACS DCF for counties, ACS for states, DCF for countries and small state, ACS for rest DCF for smaller counties, ACS for larger counties and states.

Additional Research Applications• Foreign-Born Emigration: Take advantage of longitudinal nature of IRS tax return records, to track

foreign born who drop out of file, either temporarily or permanently (estimate rates/probabilities)

• Net Native migration• IRS tax filing addresses outside the United States (either at T1 or T2), with a consecutive year filing

address within the United States (measures both in- and out-flows)

• “Undocumented” migrants• Theoretically, ACS includes all migrants in sample, both authorized and unauthorized.

• Examine characteristics of ACS survey respondents not present in administrative data sources• Individual Tax Identification Records

• Tax IDs given to those not eligible for Social Security Numbers, many of whom are unauthorized migrants

36



IDIM Questions• Evaluation of IDIM foreign-born coverage

• Missing recent/irregular migrants?• How does IDIM compare to sample-based estimates?• Adjust IDIM to account for “new arrivals”?

• Do we still have to use a IDIM proxy universe (stocks) or can we estimate flows directly?

• Evaluation of ITIN tax data (tax IDs given to those ineligible for Social Security numbers)• Can we include them on the IDIM (currently on NUMIDENT)? • How much would this improve coverage?

37



Presenter

Presentation Notes

ITINs included on NUMIDENT certain classes of people from certain parts of the country are less likely to be represented in the set of records assigned to a PIK. The problem may be due to under-coverage of these groups in the reference file, an association of the group’s records with poor quality survey records— records lacking good name, address and date of birth information—or both.

Considerations when using administrative data to estimate international migration

• Data Accessibility• Data sharing mechanisms need to be in place (no guarantee data will be delivered)• Communication between agencies critical

• Data linking procedures• Deterministic or probabilistic• Longitudinal analysis

• Data protection

• Data quality issues• Data cleaning necessary• Emigration still problematic

• Data comparability: Operational/Definitional• Coverage (under and over)• Time period (Calendar vs Fiscal year)• How migrants are defined across data sources

• Counting events vs people• Duration of stay, usual residence, actual vs intended stay, temporary vs. permanent

• Status is fluid

38



Presenter

Presentation Notes

Potential for longitudinal analysis Emigration still problematic to measure

The Future• Data Integration will become an important feature of international migration statistics

• Draw upon the strengths of each data source, mitigating their weaknesses

• Data privacy concerns will make micro-data integration much more difficult in the future• US Census Bureau will continue to pursue these avenues of research, including greater

incorporation of administrative data, as household survey-based estimates become less tenable (data privacy adjustments, increased non-response, especially by foreign-born respondents)

• Non-traditional data sources will likely need to be incorporated as well

39



PART 2

Pitfalls in using administrative dataJason GAGNON

Statistical realityGeneral concept of

migration

Statistical concept and recommendation

Actual reflection in data

Module 6: Strengthening administrative data sourcesPart 2: Pitfalls in administration_Jason Gagnon

Current statistical reference:

United Nations 1998 Recommendations on Statistics on International Migration

A long-term migrant is a person who moves to a country other than that of his or her usual residence for a period of at least a year (12 months)

1953: Focused on impact of long-term settlement

1976: Focused on long-term settlement, but made statistics complicated

Since 2018: the United Nations Statistical Commission requested the UN Expert Group on Migration Statistics (created: 2017) to initiate a revision

Definition


Presenter

Presentation Notes

IOMchief: we need to collect #data reflecting what might be needed to plan for #future #migration #scenarios, rather than just collecting data for the sake of it Global governance and standards are important, particularly on statistics where comparisons can provide a better grasp on policy options But, migration policy and therefore statistics must be firmly based on bilateral needs and contexts, if they are to be effective Examples of SS vs. SN; informal vs. formal · Migrant stocks are higher in SS On the other hand: intra-Africa trade, is abysmally low at 16% of total · Most forcibly displaced persons are disproportionately in developing countries · SS remittance costs are 4-5x higher than · Informality is more common than formal flows (both migration and remittances) · The way we capture statistics may not be adequate for policy making in developing contexts (example of border economies) · Diasporas are increasingly looking for ways to contribute

Current statistical reference:

United Nations 1998 Recommendations on Statistics on International Migration

A long-term migrant is a person who moves to a country other than that of his or her usual residence for a period of at least a year (12 months)

1953: Focused on impact of long-term settlement

1976: Focused on long-term settlement, but made statistics complicated

Since 2018: the United Nations Statistical Commission requested the UN Expert Group on Migration Statistics (created: 2017) to initiate a revision


Definition

Presenter

Presentation Notes


Three parts to the definition:

1. Whether individual moved from one country to another country by crossing an international border

2. Where individual has changed her/his usual residence

3. How long the individual has been living in the current usual residence (duration)


Breaking down the standard definition

Presenter

Presentation Notes


Breaking down the standard definition

Three parts to the definition:




Usual residence: the country in which he or she normally spends the daily periods of rest.

Temporary travel abroad for purposes of recreation, holiday, visits to friends and relatives, business, medical treatment or religious pilgrimage does not change a person’s residence.


Presenter

Presentation Notes


Breaking down the standard definitionThree parts to the definition:




3 months 12 months

Long-term international migrant

Short-term international migrantVisitor


Presenter

Presentation Notes


Statistical interpretations of the 1998 statistical recommendation

Born in the current country Not born in the current country

Not national

of the current country

National of the

current country

Not an international migrant

International migrant

International migrant

Not an international migrant

sometimes…


Major harmonised global sources

• OECD: based on census data in OECD countries

• WB: since 2007, first attempt at a global bilateral matrix

• United Nations, bilateral matrix

Focus on WB bilateral matrix

• Uses the foreign-born definition due to greater number of countries report data by place of birth

• definition less vulnerable to differences in naturalisation policies across countries

• it more readily accords to the actual movement of migrants, “which is the economist’s principal interest”

• At 2005 census round: the dataset contained information on source countries of foreign-born migrants for 96 destination countries and source countries of migrants counted by foreign nationality in 105 destination countries


Consequences of the standard definition use

• IOM Chief at IFMS:

• We need to collect data reflecting what might be needed to plan for future migration scenarios, rather than just collecting data for the sake of it

• What policy realm are we trying to inform?

• Understanding overall human mobility flows

• Migrant integration experience

• Determinants of migration

• Impact of policy

• African migration realities

• Informality, circularity, fluid

• Border economies

• Forced displacement


Data harmonisation

The case of Burkina Faso and Côte d’Ivoire

How many Burkinabe are there in Côte d’Ivoire?

• According to CIV authorities: 3.5 million

• According to the UN: 1.4 million

How many emigrants from Burkina Faso?

• According to BFA authorities: 60k in 2006

• According to the UN: 1.2 million in CIV alone (in 2005)

3.5 million

1.4 million


Policy questions vs. definitional statistics• Reason for migration?

• Forced displacement• UNHCR (official status), legal protection• Forced displacement [>80% in African countries]

• From whose point of view?• Based on status?• Based on pre-departure intentions (ie. Gallup)• Based on household surveys?

• Integration experience?• Access to services? Impact of policies?• Diaspora engagement?



Reasons for migrating – status vs. individual responses

Government employment

agenciesVocational

trainingPublic

employment programmes

Agricultural subsidiesTraining

programmesInsurance-

based programmes

In-kind distribution

programmes

Cash-based programmes

Government subsidies / Tax

exemptionsAccess to bank

accountsFinancial training

programmes

Formal labour contractsMedical

insurance / Pensions

Access to health facilities

HHS flexibility: Integrating policy into migration surveys


Beyond standard flows and stocks• Remittances

• Informal vs. formal channels• Costs (remittances within Africa are 4-5x more expensive from outside)

• Demographic trends• Implications for future flows, channels


Some publications on migration and development2006

2007 2011

20172010 2014

2016 2018


Thank you

Jason GAGNON

OECD Development Centre

[email protected]

twitter: _jasongagnon

mailto:[email protected]

PART 3

Administrative data sourcesNougbodohoue Samson Bel-Aube & Brian Okengo

Made possible with funding from the Federal Ministry of the Interior, Building and Community, Germany, represented by the Federal O�ce for Migration and Refugees (BAMF).

• Immigration Office• Ministry of Labour• Directorate of Work Permits• Social Security Agencies• Population registers• Border control statistics

Main sources of administrative data on migration

Part 3: Administrative data sources_ Nougbodohoue Samson Bel-Aube & Brian OkengoModule 6: Strengthening administrative data sources


• More in-depth information can be collected through administrative sources

• No single source of information can tell us everything our users want to know, or fully reflect the complexity of our changing population

• Can provide highly detailed and accurate data on the individuals

• Can be linked alongside surveys, to improve the way we measure population stocks and flows

Why strengthen data on migrants through Administrative Sources?



• Different administrative sources can help in understanding better migrations aspects and reveals some of the different travel patterns made by migrants, such as circular patterns of movement

• Allows statistics to be produced more frequently

• Increase the timeliness of statistical outputs by allowing access to more up to date information concerning certain variables

Why strengthen data on migrants through Administrative Sources?



• Set up legal framework to access to the administrative data on migration. The legal framework should be in line with the African Charter on Statistics;

• Create a statistics unit/department in the Ministry of Immigration, Ministry of labour, Immigration Office, Social Security Agency

• Policy frameworks

• Organizational frameworks

• Technical frameworks - mechanisms of data transfer

Political Commitments

Module 6: Strengthening administrative data sourcesPart 3: Administrative data sources_ Nougbodohoue Samson Bel-Aube & Brian Okengo


• Set up/strengthen national coordination committees

• The Committee should comprise by all the main actors involved in data collection on Migration statistics

• The committee should approve the main tools for recording and collection of data on migration

• The committee should approve all the administrative data sources on migration statistics

Technical Commitments



• Harmonization of concepts, definitions and methodologies for collecting migration data

• Elaborate Manual on concepts and definitions on migration

• Development of new insights of international migration trends by sector, education, occupation etc.

• Develop a comprehensive methodology for capturing reliable administrative data from various sources

• Develop data-driven rules and build an integrated system for measuring population and migration

Technical Commitments



Thank you


module 6 - international organization for migration

Documents