The Use of Household Surveys to Collect Better Data on International Migration and Remittances, with a Focus on the CIS States Richard E. Bilsborrow University of North Carolina at Chapel Hill (consultant to World Bank) with Maka Lomaia, World Bank Presented at Video Conference for CIS States, World Bank Headquarters, Washington, DC, November 23, 2010
Outline of presentation The importance of international migration in the World and the ECA region and challenges to collecting good data on migration and remittances Recent household surveys in the region and limited data collection on migration and remittances Improving data by adapting existing surveys at low cost Collecting better data using specialized surveys of international migration Conclusion
Importance of international migration in the world, demographically UN (2009) estimates 214 million persons live in a country other than that of their birth in 2010, 3.1% of the world population. Above is data on the stock of foreign born migrants accumulated in a country over their lifetime: data from the UN Population Division excel files/wall Chart (2009) show how small the flows are in any recent time period compared to the population sizes of countries. Thus only 3 countries in the world (with over 1 million population) having a net annual immigration rate as high as 1% during 2005-2010, and only two having a net emigration rate over 1% (Zimbabwe and Georgia). This means that recent migrants are rare elements in the populations of countries, and hard to find! This is crucial to bear in mind in surveys seeking data on international migrants!
Table 1. Basic Data on Select Countries in Eastern Europe and CIS States Country Year of Latest HBS Number of Households in HBS Population, 2009 (000) Foreign-born as % of Population, 2010 Mean Annual Net Migration per 1000, 2005-10 Remittances as % of GDP, 2008 Albania 2008 3599 3,155 2.8-4.8 12.2 Armenia 2008 7872 3,083 10.5-4.9 8.9 Azerbaijan 2008 5587 8,832 3.0-1.2 3.4 Bulgaria 2007 4300 7.545 1.4-1.3 5.3 Bosnia & Herzegovina 2007 7468 3,767 0.7-0.5 14.8 Belarus 2008 5328 9,634 11.4 n.a. 0.7 Estonia 2004 3165 1,340 13.6 n.a. 1.7 Georgia 2007 5257 4,260 4.0-11.5 5.7 Croatia 2008 3108 4,416 15.9 0.5 2.3 Hungary 2007 8547 9,993 3.7 1.5 1.7 Kazakhstan 2008 12000 15,637 19.5-1.3 0.1 Kosovo 2006 2392 1,998 n.a. n.a. n.a. Kyrgyz Republic 2008 4995 5,482 4.0-2.8 27.9 Latvia 2008 4002 2,249 15.0-0.6 1.8 Lithuania 2008 6102 3,287 4.0-6.9 3.1 Moldova 2008 6133 3,604 11.4-9.4 31.4 Macedonia 2008 4145 2,042 6.3-1.0 4.3 Montenegro 2006 1300 624 6.8-1.6 n.a. Poland 2008 37358 38,074 2.2-0.6 2.0 Romania 2008 31743 21,275 0.6-1.9 4.7 Russian Federation 2007 212330 140,874 8.7 0.4 0.4 Serbia 2009 4592 9,850 5.3 n.a. 9.8 Tajikistan 2007 4643 6,952 4.0-5.9 49.6 Turkey 2008 8548 74,816 1.9-0.1 0.2 Ukraine 2008 10622 45,708 11.6-0.3 3.2 Uzbekistan 2003 9620 27,488 4.2-3.0 n.a. Sources: World Bank, columns 3, 4, 8; UN, International Migration Wall Chart (2009) rest.
Highlights from Table 1, focusing on select CIS states As of 2009, 8 of 26 countries in the table have over 10% of their population who are migrants in the sense of being foreign born, with the percentage for the 10 select CIS countries ranging from 4% (Georgia, Kyrgyz Rep., and Tajikistan) to over 10% in Armenia, Moldova, Belarus, Ukraine and Kazakhstan, 19.5% in latter). Only Russia has net immigration since 2000, at 0.04% per year. The other nine CIS countries of interest have been experiencing net emigration, from -.13%/year in Kazakhstan to -1.15%/year in Georgia. Sample sizes of HBS/LSMS surveys of the 9 vary from 4,643 in Tajikistan to 12,000 in Kazakhstan, implying small numbers of households with recent international migrants (emigrants).
Examples of CIS States Collecting Better Data than Others Best HBS survey on collection of data on immigrants: Kyrgyz Rep. : country of birth, citizenship and previous residence; whether came in past 10 years, year, reason; if return migrant. Best HBS on emigrants: Georgia and Moldova, both collected current age, sex, education, and why person left Georgia: collected for everyone leaving but only for those leaving in previous 3 months vs. Moldova those in past 12 months; but Moldova only collects data on those leaving for work. Best HBS survey on remittances: Georgia and Uzbekistan (incomplete)
Defining and Measuring Migrants Based on place (country) of birth being different from country of current residence = foreign born Based on country of previous residence being different from current, plus time of arrival (yields fixed-period migration) Based on citizenship Sources of data: Census, border/admission statistics, current population register, registers of foreigners/foreign workers, naturalization statistics, surveys.
Need to consider: What is the main purpose of the data collection on migrants? Identify/count international migrants-immigrants, emigrants, and/or return migrants? Over some fixed recent time period (cut-off of 1, 2, 5, 10 years)? Characterize migrants: age, sex, education, work experience, skills, language ability.? Collect data on remittances sent and/or received? Study determinants and/or consequences of international migration?
Two ways of collecting better data on international migration/remittances with surveys 1. Adapt existing survey - which has an established major purpose other than migration, sample size and design, questionnaire, stakeholders, etc. Major advantage: costs of survey already covered, inexpensive to add and process questions/modules Disadvantages: cannot add many questions, sample of migrants may be inadequate 2. Design and implement a new specialized survey focusing on international migration
Types of existing surveys to think of adapting Household Budget Surveys (HBS) and Living Standard Measurement Surveys (LSMS): main advantage: obtain detailed data on income, sometimes on remittances received by households disadvantages: small sample size, no data on immigrants or emigrants Labor force surveys (LFS) advantages: larger sample, obtain household composition, employment, data on immigrants disadvantages: no income data; usually no remittances data Other (e.g., Multi-purpose) surveys Longitudinal surveys
Adapting existing surveys to collect more/better data on International Migrants Requires, if possible: (1) large sample size and/or (2) high prevalence of migrants of interest (e.g., households with emigrants), because (recent) migrants are usually. Rare elements, especially if the focus is on recent migrants vs. lifetime migrants, as it should be for policy Example, if a country has even a high rate of emigration of 0.5% per year, a survey of 10,000 households asking about those who left in past 5 years will yield only about 250 emigrants in 200 households, if all migrated as individuals Most useful are surveys that already collect useful data for studying migration, viz., household composition, education, employment, wages, etc., which applies to HBS/LSMS and LFS.
Surveys analyzed in the framework of this project
Countries included in the analysis
11 Tables in the report show the availability of data on migration in 28 countries of Eastern Europe and the CIS States From Household Budget Surveys (HBS), LSMS, Labor Force Surveys, and Specialized Surveys On immigrants: age and sex; country of birth, citizenship, previous residence; when came, reason; education at arrival; if return migrant, etc. On emigrants, if identified in household: age and sex, when left, destination; reason for leaving; education and employment before leaving and in destination. On remittances (sent and) received: from former member or not?; country of residence of sender; amount in money, goods in past 12 months; transfer mechanism, use of remittances.
Summary of results from Tables 2-12 (see Table 13 in Report) 1. Of the 24 most recent HBS surveys: one country identified emigrants from the survey household, with five more partially identifying; 5 identified whether someone left within the past year but only one identified if in a year before that; 3 identified if emigrant currently working in destination country. Regarding remittances, HBS data are much better: 14 identify if household received money from former members in the previous year, though 10 only partially; none identified the country of residence of the person sending, only one the transfer mechanism or use of remittances. 2. Of 18 Labor Force Surveys: 15 identified immigrants and year of arrival, but only 3 inquired if household had an emigrant (6 more partially); only one identified if the person is currently working and destination. No LFS survey obtained any data on remittances.
Examples of best practice among the 10 CIS focus countries On immigrants: Kyrgyz Rep. HBS obtained country of birth, citizenship and previous residence of immigrant; whether came in past 10 years and year; reason; and whether person is a return migrant. On emigrants: Georgia and Moldova HBS are best, both getting age, sex, education and reason for leaving, but Georgia for only the previous 3 months vs. Moldova for previous 12 mos.; but Moldova only obtains data on persons who left for work, while Georgia covers all emigrants. On remittances received: Georgia HBS is best, obtaining data on both money and goods received and from former household members and non-members; but recall period is only last 3 months, and country of origin for money received is not reported. Uzbekistan is second best, with both also collecting data on money sent to out-migrants but only for past month. LFS in the ECA region cover only immigrants, and only one CIS country carried out a LFS, Moldova, but it obtained no data on migration. LSMS and specialized migration surveys collected considerably more data on migration and remittances in Albania, Armenia, Kosovo, Tajikistan.
Examples of questionnaire modules to add to HBS/LFS in countries of emigration or immigration (provided in report) To identify emigrants/immigrants On basic personal characteristics of emigrants, such as age, sex, education and work activity prior to departure On reasons for migration On education and work activity in country of destination On migration intentions On remittances received and use On return migrants
Example of Armenia: testing migration questions provided by ILO This was not a regular LFS of 3,600 households, but rather a one-time special survey of 1,985 households, nationally representative, carried out in 2006, with modules on immigrants, emigrants, and return migrants (40 questions) For emigrants and return migrants, it covered all persons aged 16+ who had been away over 3 months any time since 1990 Household head is asked the main characteristics of the emigrant (512 reported) currently living abroad; if he/she is working in the destination country; if ever sent remittances in money and/or goods, when first and last time, amount/value in past 12 months, how sent, and what it is used for. Return migrants in the household (593) are asked similar questions Most migrants are married men, living in or returned from Russia. Unknown if special sampling design was used to find migrants.
Conclusion regarding the state of knowledge on international migration The state of knowledge is weak, partly due to lack of good data and studies Better data on the number of migrants, basic characteristics, and remittances can be obtained by adapting existing surveys, but only if sample size is sufficient. Otherwise, need to increase sample size or change sample design. If a country wants to go beyond the above, to study the determinants and/or consequences of migration, a specialized migration survey is needed. This can use specialized methods of data collection, including (1) a sample design to find enough migrants, and (2) special questionnaires that collect the needed detailed data
Designing a specialized survey on international migration requires A sample design that finds migrants, which involves 1. Stratification with oversampling to select sample areas E.g., in country of emigration, use latest census (or other source) to form strata based on the expected proportion of international migrants living in each area; then oversample areas with high proportions 2. Two-phase sampling in the last stage (smallest) areas sampled, involving listing in the field households according to whether they had a member migrate abroad in past (e.g.) 5 years, then oversample those with recent emigrants. 3. Interview households using detailed specialized questionnaires.
Examples of specialized surveys of international migration Albania, 2008 LSMS with Migration Modules Kosovo, 2009 Migration Survey Argentina example of subnational survey, 2002-2003 Morocco longitudinal survey, 2008-9 Egypt Labour Market Panel Survey, 2006- UNHCR-funded Survey of Colombians in Ecuador, 2006 NIDI Push-Pulls survey linking countries of origin and destination
NIDI - Eurostat Push-Pulls International Migration Survey Project, 1997-1998 Five countries of origin of migrants: Turkey, Egypt, Morocco, Senegal and Ghana Two countries of destination: Italy and Spain Defined migrant households as having a member who left to live abroad in previous 10 years (or as having someone arriving in previous 10 years) Use common methodology: sample designs appropriate for rare elements, with stratification, etc.; similar questionnaires Sample sizes range from 1100-2000 households in countries of emigration; 700-800 migrants in destination countries
Conclusions To collect useful data on international migration and remittances, it is necessary to use clear definitions of migrants (emigrants, immigrants, return migrants). It is desirable to use common definitions across CIS states. A commitment of CIS states to collect better data will make it possible to compare data on emigrants from A to B collected in A with data on immigrants from A to B collected in B, which will lead to improved data over time. Modules can be added to existing questionnaires to fit country needs, but sample sizes may not be large enough to provide sufficient numbers of migrants for analysis. This should be checked before adapting existing surveys. Specialized migration surveys can more efficiently concentrate data collection on households with migrants but involve costs.