Towards an improvement of current migration estimates for Italy Domenico Gabrielli, Maria Pia Sorvillo Istat - Italy Joint UNECE-Eurostat Work session on Migration Statistics (Geneva, 3-5 March 2008)

OUTLINE Work session on migration statistics - Geneva, 3-5-march Sources for statistics on migration 2.Improvement of the process 3.Is emigration underestimated by current surveys? 4.Household surveys: a complementary source of information on emigration 5.Some conclusions

SOURCES FOR STATISTICS ON MIGRATION Work session on migration statistics - Geneva, 3-5-march 2008 Municipal Population registers: Registrations and cancellations from and for abroad may be assimilated to long-term migration SUMMARY and INDIVIDUAL data on migratory flows are gathered through current surveys carried out by Istat. Population census Sample surveys

CURRENT SURVEYS ON MIGRATION Work session on migration statistics - Geneva, 3-5-march 2008 Summary data (demographic balance) monthly and yearly collection of population flows (including immigrants and emigrants) for each municipality Individual data registrations and cancellations in population registers for migrations between Italian municipalities and abroad, including characteristic of the migration and the migrants Undercount of the outward flows in the data collected with the survey on individual change of residence, compared to the survey on municipal demographic balances

DIFFERENCE BETWEEN AGGREGATED AND INDIVIDUAL DATA Work session on migration statistics - Geneva, 3-5-march 2008

IMPROVEMENT OF THE PROCESS Work session on migration statistics - Geneva, 3-5-march Coherence with statistical requirements Attention to changes in administrative rules and practice 2. Lack of detailed information Better monitoring of survey and actions on the municipalities (respondents) Progress of web data collection New centralized System 3.Delay in the availability of data Progress of web data collection Estimates

DISCREPANCIES IN POPULATION REGISTERS’ (AND IN SURVEYS’ DATA) Work session on migration statistics - Geneva, 3-5-march 2008 Background: Census defines the official amount of resident population Population registers (Anagrafi) give the changes due to natural and migration dynamics therefore updating official population Differences between the registered population (Anagrafi) and the resident population (Census) are due to: Census undercoverage (persons missed by the census) Census overcounting, especially for foreigners who do not have requirements needed by the law to be resident Anagrafi undercoverage, especially for foreigners who do not register Anagrafi overcounting (missed cancellations of those who leave the country)

IS EMIGRATION UNDERESTIMATED BY CURRENT SURVEYS? Work session on migration statistics - Geneva, 3-5-march 2008 Available sources of information: Population census (2001) Census Coverage survey (2001) Population registers (2001 stocks, flows)

A ESTIMATE OF THE EMIGRATION UNDERCOUNT Work session on migration statistics - Geneva, 3-5-march 2008 ∆ Register - Census = Census undercoverage - Census over count + “Shadow emigration” - Anagrafi undercoverage Maximum Estimated SE rate: 27%

HOUSEHOLD SURVEYS: A COMPLEMENTARY SOURCE OF INFORMATION ON EMIGRATION Work session on migration statistics - Geneva, 3-5-march 2008 Requirements: Household survey with large sample size Panel or Pseudo-panel Questions about the permanence in the households of each member Tracking of the household Eligible surveys: EUSILC (European Statistics on Income and Living Conditions) Labour Force Survey

EUSILC: SAMPLING DESIGN Work session on migration statistics - Geneva, 3-5-march Two stage sampling design: primary sampling units are the municipalities; the secondary stage sampling units (SSU) are the households (people living in Institution are not included in the sample) 2.The PSU are stratified according to their size in terms of number of residents. 3.Stratification is carried out inside each region (NUTS2). 4.Four municipalities are selected in each strata with probability proportional to their size in terms of residents. 5.For selected municipalities, households are selected from the municipality population register Macroregion Population (Households) Sample size (households) North10,583,08512,513 Centre4,226,3776,320 South7,197,4536,668 Italy22,006,91525,501

EUSILC: ROTATION OF SAMPLES Work session on migration statistics - Geneva, 3-5-march The whole sample is composed of 4 different panel, each one in the survey for 4 waves (=4 years). 2.Each year 1/4 of the sample is renewed, replacing the group entered in the sample four years before. 3.Each group is associated to one of the municipalities of the strata. Rot. Group \year ABCDEFGHI TA4B3C2D1 T+1B4C3D2E1 T+2C4D3E2F1 T+3D4E3F2G1 T+4E4F3G2H1 T+5F4G3H2I1

LABOUR FORCE SURVEY: SAMPLING DESIGN Work session on migration statistics - Geneva, 3-5-march 2008 Sampling scheme as described for EUSILC 2 sampling stages: Municipalities as PSU and Households as secondary sampling units (people living in Institution are not included in the sample) stratification of municipality (in each province - NUTS3) on demographic size; selection of one PSU in each stratum proportional to its demographic size; SSU are selected from population municipality register. About 1250 municipalities, households each quarter

LABOUR FORCE SURVEY: ROTATION OF SAMPLES Work session on migration statistics - Geneva, 3-5-march 2008 Group Year T quarter IA4B3E2F1 Year T quarter IIB4C3F2G1 Year T quarter IIIC4D3G2H1 Year T quarter IVD4E3H2I1 Year T+1 quarter IE4F3I2J1 Year T+1 quarter IIF4G3J2K1 Overlap of the sample of year T and T+1 (same quarter): 50% of the sample; (year T quarter I and year T +1 quarter I: rotational groups E and F ) Overlap of the sample in two consecutive quarters: 50% of the sample, (year T quarter II and III, rotational groups C and G )

TRACKING Work session on migration statistics - Geneva, 3-5-march 2008 Both in EUSILC and Labour Force Surveys the households are contacted in subsequent occasions. In addition to the information on migration available from the municipality registers (agreeing with the administrative data), some question are addressed to detect migration of households and individuals In EUSILC, when a household is not found, the interviewer can specify the motivation “moved abroad”. Moreover, if a member is not present in the household the same motivation can be given. In the LFS, when a member of the household is no longer present a question is asked about the reason (moved abroad being one of the items). Tracking is much less focused on, as it is essentially a cross-sectional survey.

FIRST RESULTS FROM EUSILC-2006 Work session on migration statistics - Geneva, 3-5-march 2008 Estimates Sampling error (%) Confidence Interval (α=0.05) Lower Limit Upper Limit Households42, ,82856,826 Household members (A)85,654 57,656113,652 Individuals (B)75, ,03195,193 Total (A+B)161, ,687208,845

FIRST RESULTS FROM EUSILC-2006 (2) Work session on migration statistics - Geneva, 3-5-march 2008

SOME CONCLUSIONS Work session on migration statistics - Geneva, 3-5-march 2008 Improvement of statistics on emigration can result from a multi dimensional approach: 1. To monitor the changes in legislation and the actual management of Population registers Strict cooperation with the Ministry of Internal Affairs Ongoing contacts with municipalities and associations of public officers in charge of Registers Activities to make the main actors aware of requirements and utility of statistics.

SOME CONCLUSIONS Work session on migration statistics - Geneva, 3-5-march To ameliorate the production process of current surveys + Accuracy + Completeness + Timeliness Monitoring of respondents Web data collection Centralized systems Statistical estimates

SOME CONCLUSIONS Work session on migration statistics - Geneva, 3-5-march To use other sources to check/complement existing data (EUSILC and LFS) Check Consistency between results of current surveys and household surveys Obtain valuable clues on the size of “shadow emigration” Derive more information on socio-economic status of emigrants