Measuring Migration: Best Practices in Censuses and Household Surveys

1
Measuring Migration Best Practices in Censuses
and Household Surveys

• Gero Carletto
• Carlo Azzarri

2
Background

• Migration policy research hampered by
  availability of data
• Definitions and measurements of migration varies
  across and within countries (from ? sources), and
  overtime
• Best practices?
• Lack of consensus
• Who is an international migrant?
• What type of migration?

3
What type of migration?

• Flow vs. Stock
• Immigrant vs emigrant (net migration)
• Permanent/L-T vs. temporary/S-T
• Circular, returnees
• Internal
• Undocumented/irregular
• International migrants vs. international
  migration (multiple moves)
• ? Different definitions/dimensions, different
  sources

4
Outline

• Concepts and definitions
• Data sources
• Sampling and survey options
• Questionnaire contents/design
• Conclusions and next steps

5
Concepts

• (Change in) Place of usual residence
• Time/duration of stay
• Purpose of stay
• Citizenship
• Place of birth
• Alone or in combinations
• in which order?
• for which purpose?
• ? Used differently by different sources and
  depending on objective

6
Definition of Intl Mig (FLOW)

• An international migrant is a person who changes
  his/her own country of usual residence (UN RSIM,
  Rev. 1)
• Place of usual residence?
• normally spends daily period of rest
• Usual ? legal
• Boarding students? Weekly commuters?
• No provision for
• Duration of stay
• Purpose of move

7
Long- vs. Short-term migrants (UN RSIM Rev 1.)

• A long-term migrant is a person who moves to a
  country other than that of his/her usual
  residence for a period of at least a year (12
  months)
• A short-term migrant is a person who moves to a
  country other than that of his/her usual
  residence for a period of at least 3 months but
  less than a year (12 months) (except for
  recreation, holiday, business, medical treatment,
  )

8
Definition of Intl mig.(STOCK)

• Persons who have ever changed their country of
  usual residence i.e. spent at least 1 year in a
  country other than the one in which they live at
  the time data gathered (UN RSIM Rev. 1)
• Relevant population groups
• foreigners (non-citizens of country of usual
  residence)
• foreign-born (born in country other than one in
  which they are being enumerated)
• provisions for 14 special groups

9
Alternative definition (Bilsborrow et al, 1997)

• An international migrant is a person who have
  lived for at least 6 months in a country other
  than that in which they are being interviewed and
  whose move into the country of interview occurred
  during the 5 years preceding the interview
• but definition assumes presence!

10
Complementary definition (Bilsborrow et al.,
1997)

• An international migrant is a person who used to
  live in the country in which the interview is
  being conducted and was a member of the household
  of the person being interviewed but who left at
  some point in time during the 5 years preceding
  the interview to live abroad for at least 6
  months

11
Operationalizing definitions

• Usual Place of residence
• Some countries do not use concept of migrant
• Why 3-6-12 months?
• Intended vs. actual duration of stay?
• at time of arrival based on intention
• upon completion of 12-month period
• Lots of provisions!

12
Data sources

• Population registers
• Other administrative records
• Population censuses
• Surveys

13
Data sources Pop Registers

• PROS
• Continuous
• Measure of FLOWS (consistent with STOCKS)
• De facto duration of stay/absence (if
  centralized)
• CONS
• Not run by NSO coordination
• Only limited information
• Often not public access
• Often only nationals only documented
• (dis)Incentive to register/deregister
• Intended vs. actual duration of stay
• Quality variability (if decentralized)
• Feasible for LDCs? (EU from PopCensus to
  Registers)
• Only 15 countries reporting flows figures (mostly
  from pop registers UK, NZ from border statistics)

14
Data sources Other admin records

• Border crossings (flows)
• Registry of foreign workers
• Embassy/consular data
• Resident permit holders/applicants
• Diaspora organizations/NGOs in D country
• Emigration clearance certificates
• Destination country sources
• often incomplete both in coverage and content

15
Data sources PopCensus

• Most reliable source of internationally
  comparable IMMIGRANT stocks.
• PROS
• Universal coverage
• Some characterization possible, based on basic
  demographics and socio-economic, e.g.
• Gender (feminization of migration)
• Age (SSS fertility)
• Occupation (workforce)
• Educational level (brain drain/gain)

16
Data sources PopCensus (contd)

• CONS
• Every 10 years
• Limited information (policy analysis?)
• Poor/delayed tabulation
• Low training of enumerators/data quality
• Different approaches/definition
• De facto (present population) vs. de jure (usual
  population)
• 14 sub-groups for special treatment
• Under-coverage
• Entire families that moved
• Seasonal/temporary, circular (night prior to
  census date)
• Marginal groups (undocumented migrants?)
• Housing arrangements, language, distrust,
  mobility,

17
Data sources PopCensus (contd)
(Chen, 2006)
18
Data sources PopCensus (contd)

• Only 1 country in 89 compliant with UN RSIM Rev.1
  (of 153 of 196)!
• 11 countries close to definition
• citizenship (93 countries)
• stateless
• place of birth (112 countries)
• new countries
• PoR at some point in time in the past (88
  countries)
• (Chen, 2006)

19
Data sources PopCensus emigrants?

• Self-reporting about absentees based on
  socio-econ status (hh membership)
• Two (indirect) methods based on demographics
  (IUSSP)
• Place of Residence (PoR) of siblings
• PoR of children (complementing existing sources
  split surviving children by PoR)
• Even more limited information on emigrants
• Less flexibility
• No sensitive info (affect coverage)
• Feasible at all?
• Quality of information
• Proxy respondent
• Enumerators training
• Balance equation/residual (Albania)
• Using data from PopCensus in D countries?

20
Data sources Surveys

• Despite hype, haphazard efforts
• Little attention in
• UN RSIM Rev 1.
• EGM recommendations
• Countries should explore the possibility of
  using sample surveys to collect data on
  international migration, especially for those
  aspects for which no other sources are available
• PROS
• In-depth analysis
• determinants, distributional analysis, impact
  (??)
• More flexibility
• Emigration (last resort?)
• CONS
• Rare event, clustered
• Sampling error
• Sampling frame (emigrants)
• Richer info but still limitations through proxy
  respondents

21
Survey options prob vs. non-prob

• Probability finding rare events
• Huge sample?
• 8 procedures (Kish, 1965)
• Disproportionate
• Two-phase
• Non-probability
• Case studies
• Purposive geo selection (NIDI/Eurostat)
• Snowball
• Aggregation point intercept (Brazil-Nikkei)

22
Survey options HH vs. non-HH

• Passenger surveys
• UK IPS (flows)
• Low incidence
• Short questionnaire
• HH surveys
• Specialized
• Preferred option but
• Too costly/unfeasible
• Limited thematic coverage
• Piggy-backing
• Richer info (welfare?) but shorter migration
  module
• Marginal costs
• Sampling issues
• LSMS/IHS (small sample Albania), HBS/IES (high
  non-response resistance) LFS (larger sample
  frequency no welfare Philippines)

23
Survey on Overseas Filipinos (SOF)

• Long tradition now integral part of LFS.
• 1982 1-page rider to ISH
• 1987 part of LFS (OFW only)
• 1993 expanded module, renamed (OF)
• 2003 Master sample (51,000 HHs)
• Estimates and characterize stock of overseas
  Filipinos, including OFW (presently AND
  temporarily working overseas in last 5 year
  period) 1.2M in 2004

24

• Questionnaire
• Contents / Design

25
Identifying a migrant in HH

• Approach 1 Typical LSMS/IHS
• Self-reported hh membership (usually eat and
  live together)
• No. months absent in past 12 months
• Provisions, for HH head, new-comers
• Several drawback
• Permanent migrants?
• Exclusion based on arbitrary cut-off
• Ambiguous
• Misses entire HHs

26
Identifying a migrant in HH (contd)

• Approach 2
• in the last 5 years, did anyone who lives or
  lived with you go to live in another country?
• Captures recent migrants but
• Still ambiguous
• Arbitrary reference period
• Still misses entire HHs

27
Identifying a migrant in HH (contd)

• Approach 3 Extended roster (with flap)
• List all present and former hh members
  irrespective of PoR
• Captures all members but
• Still ambiguous
• Incentive to underreport?
• Misses entire HHs

28
Identifying a migrant in HH (contd)

• Approach 4 Expanded fertility module (Albania
  LSMS 2002)
• From 15-49 to above 15
• List all women in age group
• List all children ever born from woman
• PoR for still alive
• However
• Still possible undercount
• mother absent
• Misses entire HHs

29
Once identify (e)migrant

• How much information can you elicit through proxy
  respondent? Whats the min-MAX amount of info?
• Age, gender, occupation (before and during)
• Length stay abroad/year of departure
• Country of destination
• Living with spouse/children
• Legal status?
• Multiple episodes?
• Recall methods
• Less accurate through proxy
• Recall bias (Smith and Thomas, 2003 Som, 1973)

30
Albania LSMS 2005

• Expanded module on (permanent) migrants (non-hh
  members) plus contact information for tracking in
  D country
• Full histories and characterization of migration
  of current household members (15 years)
• Bound grid
• Clear time-marks

31
Migration histories of HH members
32
What if entire HH moves?

• If dwelling occupied
• Ask new occupant
• About location
• About relatives still in communities
• If dwelling vacant/destroyed
• Ask neighbor
• Ask neighbor about relative still in community
• If dwelling moved?
• Internal migrants
• Restricted universe of analysis
• Track them? Feasible?

33
Tracking migrants

• Replace proxy respondent
• Issues
• Info needs
• Attrition
• Costs
• Albania (forward tracking)
• Tonga (backward tracking)

34
Tracking Albanian migrants in Greece

• Contact info
• Process
• Greece only list of migrants with contacts
• 1st contact in Greece
• Return visit in Fall (phone cards)
• 2nd contact in Greece
• Interview
• High attrition
• No contact info
• Unable to locate
• Islands (high costs)
• Refusal (not high)
• Returnees/high mobility
• Selection bias(es)? (undocumented aliens)

35
Conclusions

• Standards not uniformly adopted across and within
  countries (depending on source), and changing
• Definition to be adopted depends on objective at
  hand and sub-populations of interest
• Census stock, but
• Registers flows, but
• Survey Surprisingly, still little attention
• Weakest link emigration!
• Better estimations
• Better characterization
• Better sampling frames

36
Next steps WBs role?

• 2010 Round of PopCensus
• Influence Qx design
• Min set of questions on immigrant stock
• Question on Emigrant?
• Approximation for stratification in sampling
  frame
• Improve admin records
• TFSCB
• Register (if available) and/or border crossing
• Inter-institutional collaboration
• Data access

37
Next steps WBs role? (contd)

• HH surveys
• Support proper sample designs
• Sampling frames of emigrants
• enhanced PSU listing for 2-stage
• Oversampling in planned surveys
• Reduce undercount/misreporting
• Fielding and validation of ? approaches (LSMSIV)
• Adjust for entire migrant households
• Specialized vs. multi-purpose
• Better income-based measures

38
Next steps WBs role?(contd)

• HH surveys (contd)
• Improve content
• Much more flexibility allowed
• Emigrants, Returnees,
• If migrant present
• Histories of migration
• Retrospective data, determinants
• Through proxy
• Minimal set of questions?
• Contact info for tracking (sub-sample?)
• Tracking surveys
• Panel surveys
• Make provision in baseline