A tool for evaluating integration processes Gian Carlo Blangiardo Fondazione Ismu / Università di Milano Bicocca
Three preliminary remarks Integration holds for some specific characteristics: processuality, multidimensionality, and bi-directionality; The process of integration of immigrants is fundamentally influenced by the context of the receiving countries and takes place first of all at the local level; The Process must be monitored: according to EU recommendations [ ] clear goals, indicators and evaluation mechanisms are necessary to adjust policy, evaluate progress on integration and to make the exchange of information more effective. [ ] it is important to know whether integration policy efforts are effective and make progress [ ]. (Common Basic Principles for Immigrant Integration, Justice and Common Affairs Council, 2004, CBP 11).
Monitoring the integration process The identification of well-defined indicators is the first step needed for the process of accountability, but it is only the beginning of the monitoring process. In fact, the implementation of such indicators depends mainly on the available data and the approaches used by different actors interested in the phenomena.
Data collection There has been a clear global effort to identify sources of data able to satisfy the need of producing relevant indicators. Mainly three sources of data are available for this purpose: Administrative data (including censuses, microcensuses, national statistical offices, vital statistics); General surveys (including surveys such as the European EU-SILC, specific national surveys); Ad-hoc surveys specifically aimed at the foreign population.
Data processing Macro and micro approach to integration measure According to the source of data, two different approaches can be used to measure integration: one based on macro-data and one on micro-data. The macro approach, mainly based on official sources data, permits to look at migration and integration in respect to a series of variables of interest such as age, sex, household composition, level of education, occupation and income among others. Following this approach the subpopulation is defined according to a profile based on the average/frequency of the variables associated with the concept of integration. However this approach allows the comparison between different populations only based on one of the dimension at a time. In fact, a specific group can be the most integrated with one dimension (i.e. employment) but lag behind in relation to another dimension (i.e. social inclusion). The micro-approach makes possible to construct more complex measurements of integration based on the individual profiles associated with a set of characteristics. It is important to underline that, while the macro-approach can rely on aggregated data (i.e. tabulation), the micro-approach requires access to individual level data.
The micro-approach: how to construct the individual score of integration Through the micro approach the construction of an overall score of integration at individual level allows to analyze spatial and sub-population differences in the process of integration and to control the efficacy of specifically oriented or local policies. Basic tools needed I) An individual data set (Census data or a representative sample of the target population); II) A methodology able to assign an overall integration score, according to an a priori definition of integration, to each unit.
Steps required in order to assign an integration score to every statistical unit (Blangiardo, 2013) STEP 1 Selection of a set of indicators -according to the definition of integration used- and identification of the associated variables, within the dataset, whose modalities can be ranked following a scale of integration (from low to high) STEP 2 Estimation for each variable (reflecting one of the multiple dimensions of integration) of the integration scores (ranging from -1 to +1). These can be implemented calculating the relative frequency of the variable and assigning to each modality a score. The latter is calculated as the difference between the sum of the frequencies of the previous modalities and the sum of the frequencies of the following ones
Step 2: an example I Consider a variable (extracted from the individual data set) related to the following question: Do you believe that for identical jobs women should have, compared to men: 1 - Very inferior wages; 2 - Moderately Inferior wages; 3 The same wages Scores could be 1, 2 o 3 (cardinal scores by the set of integers or similar), but efficiency of such scoring could be improved according to the principle that for assessing integration level of any individual, in a relative sense, each answer should be considered in connection to the opinion of the whole population". Therefore for our respondents the modalities 1 and 3 should be the more unfavorable or favorable to integration, respectively, the more they are expressed by a restricted minority. In fact the highest integrating modality (or the lowest) shrinks its contribution (rewarding or penalizing), if it is widely shared.
Step 2: an example II In our example, if the distribution of frequencies is (on 1000 units): 1 - Very inferior wages 400 ( 0.4 ) 2 - Moderately Inferior wages 500 ( 0.5 ) 3 The same wages 100 ( 0.1 ) Total 1000 ( 1.0 ) The corresponding scores, according to step 2 rules, are: 1 - Very inferior wages - 0.5-0,1 = - 0,6 2 - Moderately Inferior wages + 0.4 0,1 = +0,3 3 The same wages + 0,4 +0,5 = +0.9 Eventually, it can be remarked that, for any variable, the mean score for the whole population will be zero.
Steps required in order to assign an integration score STEP 3 to every statistical unit (continued) Assign to each unit a vector of scores according to the modalities of the observed variables; STEP 4 Calculate the overall integration score averaging the vector of scores for each unit; STEP 5 Summarize the overall integration score, within subgroups of the population defined, according to specific variables (i.e. gender, nationality, education and so on).
Let s see two examples of monitoring integration by a microapproach
1. Monitoring integration by micro-approach: an European experience A practical example can be useful to show the procedure for constructing, through a micro approach, measures specifically addressed to check, whether (and, if so, to what extent) some policy oriented events e.g. access to citizenship, long term residence, family reunion- can really foster the integration process into the following four domains: 1) to get a job; 2) to get more education; 3) to get more involved in the receiving society; 4) to feel more settled in the receiving society.
In this regard, we will consider the outcomes of a recent European survey, the Immigrant Citizens Survey (ICS)*, carried out in 2011-2012 through a sample of 7473 immigrants born outside EU. The sample was collected in: Portugal, Belgium, Germany, Spain, Hungary, Italy and France *Huddleston, Tjaden, 2012
Questions in the ICS survey Did / Will Becoming citizen Becoming long term resident Reuniting your family help you To get a Job To get Education To feel more settled To get more involved? Not at all A little A lot
Average of individual scores for each country interviewees
Effects and Expectations scores related to the integration path ICS survey: NATIONALITY 60 40 20 0-20 -40-60 Nationality Effects Nationality Expectations
Effects and Expectations scores related to the integration path ICS survey: LONG-TERM RESIDENCE 60 40 20 0-20 -40-60 -80 Long-Term Residence Effects Long-Term Residence Expectations
Effects and Expectations scores related to the integration path ICS survey: FAMILY REUNION 20 15 10 5 0-5 -10-15 -20-25 -30 Family Reunion Effects Family Reunion Expectations
Computation in each country -for the three events- of the average score related to the integration path ICS survey (i.e. for effects) 30 20 10 0-10 -20-30 -40-50
It can not be ignored that the rank scores are dependent on the context conditions (norms, culture, administrative practice, etc.) that are in place in each destination country
Individuals Context conditions Job Education Settlement Involvement Access to Citizenship Events Policy oriented L.T. residence Family reunion Integration
Looking at the context In order to propose an assessment of expectations and effects of those events, in connection to more or less favorable national contexts, we shall link the ICS survey outcomes to the MIPEX scores (British Council and MPG, 2011). This is since the last focuses on measuring policies and their implementation in the countries (and in the domains) we considered.
MIPEX III score for some integration dimensions (selected countries) Scores x100; ranges: 0 very bad; +100 very good 95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 Access to Nationality Family Reunion Long Term Residence Average (3 scores)
Is a favorable context (for citizenship, LTR, family reunion) positively correlated with a similarly favorable integration path in the domain of job, education, settlement and involvement into the receiving society?
Correlation coefficients between ICS survey scores & MIPEX III scores Expectations related to the acquisition of nationality ICS vs. Access to nationality MIPEX 0.25 Effects related to the acquisition of nationality ICS vs. Access to nationality MIPEX 0.83 Expectations related to the acquisition of a long-term residence permit ICS vs. LTR MIPEX -0.32 Effects related to the acquisition of a long-term residence permit ICS vs. LTR MIPEX -0.45 Expectations related to the family reunion ICS vs. Family Reunion MIPEX -0.62 Effects related to the family reunion ICS vs. Family Reunion MIPEX -0.72
A figure to read jointly integration and context
Comparison between the effects of the acquisition of citizenship (ICS scores) and the contextual conditions (MIPEX scores) Standardized values through the Z-scores* (*)Previously ICS and MIPEX scores have been transformed to give both average 0 and variance 1 Countries legenda: BE=Belgium; DE=Germany; ES=Spain; FR=France; HU=Hungary; IT=Italy; PT=Portugal
A revised ranking According only to ICS scores concerning effects According to z scores of effects minus context Belgium 46 Belgium 0.82 Germany 12 Germany 0.38 France 9 France 0.29 Portugal 6 Spain 0.01 Italy 0 Hungary -0.05 Spain -40 Italy -0.23 Hungary -58 Portugal -1.22
How the joint analysis could change the outcome: an additional example The top and the last five categories as regard the integration of immigrants/foreigners into labor market (EU countries) Without considering the context Portugal Males Foreigners born in the country Italy Males Citizens born abroad Sweden Males Foreigners born in the country UK Males Citizens born abroad Luxembourg Males Citizens born abroad Sweden Females Foreigners born abroad Greece Females Citizens born abroad France Females Foreigners born abroad Greece Females Foreigners born abroad Spain Females Foreigners born abroad Top Considering the context Luxembourg Males Citizens born abroad Luxembourg Males Foreigners born abroad Luxembourg Males Foreigners born in the country UK Males Citizens born abroad France Males Citizens born abroad Spain Males Foreigners born abroad Portugal Females Foreigners born abroad Portugal Females Foreigners born in the country Spain Females Foreigners born abroad Sweden Females Foreigners born abroad Bottom Source: A.DiBartolomeo and S.Strozza (2014). Dataset European Union Labour Force Survey (2011). Macro approach with multivariate factorial analysis
2. Local observatory and monitoring of the integration process: the ORIM application The integration process can be analysed - taking into account the contextual conditions focusing more on local realities- by processing the data from local observatories. Osservatorio Regionale per l Integrazione e la Multietnicità (ORIM-Regional Observatory for Integration and Multiethnicity), which operates in the region of Lombardy (Italy) since 2001 and is run by Ismu Foundation, represents a good example.
Osservatorio Regionale per l Integrazione e la Multietnicità (ORIM-Regional Observatory for Integration and Multiethnicity) In 2001 ORIM was created in Lombardy, for which ISMU Foundation was given the responsibility of management. Annually, through the application of the Centre Sampling Method (Baio et Al., 2011), a sampling survey was conducting in all the 11 provinces (12 from 2005). The sample size was about 8,000 units in the first five surveys (2001-2005) and has been increased to 9,000 units in 2006-2009 (back to 8,000 in 2010-2011, 7,000 in 2012 and 4,000 in 2013 and 2014). Further, since 2001, in order to monitor locally the immigration integration, ORIM developed and coordinated a network for monitoring consisting of 12 the Provincial Observatories.
Through micro approach to individual data of ORIM 2013 survey Integration indexes of foreign immigrants living in Lombardy provinces Scores x1000; ranges: -1000 very low; +1000 very high Provinces Economic & labor market integration Socio territorial integration Average scores Bergamo 53 34 Brescia 20 13 Como -23-33 Cremona 40 73 Lecco 114 127 Lodi 39 71 Mantova -19 16 Milano (city) -34-30 Milano (extra city) -37-78 Monza e Brianza -22-2 Pavia 29 64 Sondrio 16 73 Varese 36 53
Some context parameters - like unemployment rates, housing benefits for migrants, social spending for immigration, etc.- may be combined in order to assess the effectiveness of local policies. For example
What about correlation between public expenditure for immigrants and integration? I According to integration indexes and the pro capite spending in the municipalities of Lombard provinces, no correlation exist! Correlation indexes Spending & Labor market integration index = -0.33 Spending & Socio territorial integration index = 0.27
What about correlation between public expenditure for immigrants and integration? II Such outcome is not so amazing. It can be observed as regard the correlation between public spending for migrants and integration level in the municipalities of 32 Italian provinces where in 2009 a set of integration indexes were computed (Cesareo V., Blangiardo G.C., Integration Indexes, Quaderni Ismu, 2/2011). Correlation indexes Spending2009 & Cultural integration index = +0.04 Spending2009 & Social integration index = -0.01 Spending2009 & Political integration index = 0.00 Spending2009 & Economic integration index = -0.21 Spending2009 & Total integration index = +0.04
Thanks for your attention giancarlo.blangiardo@unimib.it