Migration and related efforts to harmonise international data collection on migration have been on the agenda of population statistics since it was institutionalised as a field of data collection and research in the nineteenth century, often closely linked to wider political projects. In 1843, for instance, participants of the General Conference of the German Customs Union grappled with the problem of how to deal with temporary residence in the territory of the Union and discussed at great length who should be considered as belonging to the ordinary resident population, in the aim of harmonising definitions within the (German) member states of the Union (Schmidt 2005: 133). At a truly international level, migration statistics were first discussed at the 1891 congress of the International Statistical Institute in Vienna. The first attempts to standardise definitions in migration statistics followed 30 years later at the International Conference on Emigration and Immigration in Rome in 1924. This conference was followed by several conferences each adopting resolutions on migration statistics. Institutionally, the International Labour Organisation (ILO), created in 1919, became the main actor in regard to promoting efforts to improve international data collection on migration, albeit seeing migration mainly as an issue of manpower and labour rather than one of population.
An ILO-sponsored conference on migration statistics in 1932 adopted the first more systematic set of recommendations for the improvement of migration statistics (United Nations 1949: 1). From the outset, achieving comparable international migration statistics was considered important not only for statistical, or for that matter, scientific, purposes; it was also seen as a precondition for ‘… the regulation of migration by international convention’ [and a tool to] ‘facilitate cooperation of the administrative authorities of different countries’ (International Labour Office 1932: 86, quoted in Kraly and Gnanasekaran 1987: 968). The 1932 recommendations were revised in 1953 and 1976, but both times not widely implemented. The latest revision dates from 1997 (Herm 2006; United Nations 1998). It was preceded by several conferences and studies, also drawing on expertise from outside the UN system such as the Council of Europe, the OECD,Footnote 10 IOM and Eurostat, all of which had become key players in policies on statistical data collection by the time of the 1997 revision, with the role played by Eurostat reflecting the increasing weight of the European Union at the international level more generally (Herm 2006).
3.2.1 Policies on Collection of Migration Statistics in the European Union
Efforts at collecting data on migration at the European level extend back to the mid-1970s and the ill-fated Community Regulation (EEC) No 311/76 on the compilation of statistics on foreign workers. The latter required member states to supply annual statistics both on the number of workers and on their first employment, without, however, providing guidelines on definitions or data sources to be used. This resulted in a potpourri of incomplete and incomparable data, of little use for either policymaking or research. By the late 1980s, Eurostat engaged in renewed efforts to collect basic data on migration and commissioned a number of studies about data collection on migration in the member states of the then European Community, subsequently further extended to European Free Trade Association member states. These studies resulted in a set of tables agreed by member states and Eurostat as a basis for future data collection on migration. A programme for the collection of statistics on international migration was subsequently launched in 1992, and, in the mid-1990s linked to a joint data collection programme run in cooperation with the United Nations Economic Commission for Europe (UNECE), and later also joined by the UN Statistical Division (UNSD), the Council of Europe (CoE) and the International Labour Office (ILO). In 1994 a first publication presenting data on migration flows and stocks based on these new data collection programmes was launched. Data and limited methodological information were also included in Eurostat’s statistical database New Cronos. Data collection, however, remained voluntary, incomplete and inconsistent, and doubtful in terms of the reliability of the data included in the dataset (Herm 2006: 93–95).
The nascent cooperation of EU member states on migration and asylum initiated by the Maastricht Treaty (although to some extent actually preceding it) gave rise to separate data being collected through two Council working groups, respectively on irregular migration (the Centre for Information, Discussion and Exchange on the Crossing of Frontiers and Immigration, CIREFI) and asylum (the Centre for Information, Discussion and Exchange on Asylum, CIREA), both established in the mid-1990s. Following the communitarisation of policies on migration and asylum through the Amsterdam Treaty, and responding to the conclusions of the Justice and Home Affairs Council of 1998 (“Vienna Action Plan”), which called for an improvement of the exchange of statistics and information on asylum, data collection was conducted on a more systematic basis from 1998. Responsibility for collecting data was then handed over from the Council Secretariat to Eurostat (Commission of the European Communities 2001a: 4–5). Before 2004 the data were only partly accessible to the public due to member states’ concerns about the confidentiality and sensitivity of data on irregular migration, and about their problematic quality. Nevertheless the development of migration statistics policy at the EU level was arguably pushed by member states’ interest in statistics directly concerned with migration control, rather than their interest in general demographic data on migration. Tellingly, a Commission Action Plan for the collection and analysis of Community statistics drafted in response to the conclusions of the Laeken Council in 2001, which called for increased efforts to collect comparable statistical data, is most concrete and elaborate in regard to data on asylum and irregular migration, but almost silent on legal migration or migration defined in demographic terms. Implicitly this suggests that the Action Plan simply proposes to carry over Eurostat’s data collection on general demographic indicators on migration initiated in 1998 into a legally regulated regime under future Community legislation (see Commission of the European Communities 2003).
Both the demographic data collection initiated by Eurostat and the data collection on irregular migration and asylum originally initiated and implemented by the Council have been relatively little used. This is because of the limitations of the data collected and the restrictions placed on their public dissemination, but also because a particular policy purpose has been lacking. If they have been used at all, it has been as simple indicators of the size and development of particular target populations, but not for more complex analytical purposes, policy development or policy evaluation. Thus, while the commitment to ‘evidence-based policymaking’ was a strong factor in further developing data collection, the statistical information produced hardly seemed to matter to policymaking initially, as was the case also with the information collected by the European Migration Network launched around the same time (on the EMN see Boswell 2009).
By contrast, reforms to the collection of social statistics, largely absent from European efforts to improve data collection on migration before the mid-2000s, followed a decisively different path. Linked to the introduction of the Open Method of Coordination (OMC)Footnote 11 as a new mode of governance, social statistics became understood as (quantitative) benchmarks of national initiatives and performance indicators (Bruno et al. 2006). To this end a set of indicators – the so-called Laeken indicators for social inclusion – were defined in 2001. Information was to be collected through the European Union Statistics on Income and Living Conditions (EU-SILC) as of 2003,Footnote 12 and complemented by a set of structural indicators drawn from various data-sources, including the EU Labour Force Survey with regard to basic employment indicators. While these datasets in theory – and at least to a limited extent – would have facilitated the monitoring of social and employment indicators in relation to migrants from the early 2000s, 10 years had elapsed before relevant data came to be systematically used at the EU level to monitor the integration of immigrants in the form of institutionalised integration monitoring (see below).
The development of EU policies on migration statistics relied heavily on the mobilisation of ‘external’ expertise. The mobilisation of expertise helped to create an ‘epistemic community’ (Haas 1992)Footnote 13 involving policymakers, academics based at universities and research institutions, statistical institutes and international organisations, sharing a common vocabulary, a common belief in the need to produce comparable statistics, and a common understanding of related challenges. Throughout the 1990s and 2000s, Eurostat commissioned several studies contributing to its data collection efforts. In the mid-1990s, joint meetings on migration statistics were initiated by Eurostat and UNECE, largely drawing on their own expertise and those of national statistical institutes, but also regularly inviting external experts from academic institutions or international organisations (see Herm 2006). Commissioned research and regular meetings were important instruments for creating a like-minded community of experts. In addition, circulation of experts – in the form of posting of national statistical experts at Eurostat or UNECE or career trajectories of individuals spanning both spates of service in public office and academia, etc. – were important drivers as well.
In addition, several research projects were funded under the EU’s 5th and 6th Framework Programmes for ResearchFootnote 14 thus contributing to the development of a legal framework for the collection of migration statistics at the EU level. The adoption by the member states of the ‘Regulation on Community Statistics on Migration and International Protection’ (Regulation 862/2007) in July 2007, first presented as a legislative proposal in September 2005, can be considered a major turning point in EU policies on data collection for migration and asylum.
Reflecting the growing importance of migration issues at the EU level, Regulation 862/2007 breaks with the ‘gentlemen’s agreement’ approach to voluntary data collection by EU member states on issues concerning migration and asylum. Furthermore it provides for mandatory and harmonised statistical data collection on core demographic indicators as well as on key indicators concerning the management of migration. The regulation covers the provision of comprehensive data based on harmonised definitions of population by citizenship and country of birth, migration movements, acquisition and loss of citizenship, asylum, residence permits as well as enforcement of immigration legislation. In an effort to create an incentive for providing accurate data, the allocation of various EU funds on migration and asylum was linked to data collected in compliance with Regulation 862/2007 (Kraler et al. 2006: 70).
While the optional collection of socio-demographic data suggested by the European Commission in earlier drafts of Regulation (EC) No 862/2007 was removed during negotiations, statistical data on migrants’ socio-demographic characteristics has seen an enormous expansion through other EU mechanisms. It now also involves quasi-mandatory elements, for example in the framework of European surveys such as the Labour Force Survey (EU-LFS) (notably through its 2008 ad hoc module on migrants in the labour market, repeated in 2014). In addition, in 2008, Regulation (EC) No 763/2008 on Population and Housing Censuses was adopted, stipulating obligatory topics for population and housing censuses for the reference year 2011 and including core information related to migration, most notably information on country of birth, country of citizenship and other basic data related to migration.
Since the implementation of the Regulation 862/2007, statistics, particularly those related to migration enforcement and asylum, have been increasingly used as performance indicators, notably to derive recognition rates from asylum statistics or to calculate ratios between return decisions and actual returns. Ten years after the Lisbon strategy was adopted, social indicators on migrants, as noted above, came to be systematically collected in the form of ‘indicators of migrant integration’, to be used for ‘monitoring the results of integration policies in order to increase the comparability of national experiences and reinforce the European learning process’ (European Council 2010). The need to evaluate integration policies through indicators has been put forward several times since the Hague programme, adopted by the Council in 2004, while in the Stockholm programme of 2009 the explicit call for the development of indicators of migrant integration was made. This led to the immediate development of indicators by an expert group. Fourteen indicators were developed in the four areas of employment, education, social inclusion and active citizenship, and were endorsed at the Ministerial Conference in Zaragoza in 2010 (European Ministerial Conference on Integration 2010). These indicators were produced in a pilot study by Eurostat (2011) and were subsequently evaluated and tested for their usefulness and robustness.
As a result of these different developments, which in turn were the result of complex discussions involving national data providers, Eurostat, international actors, researchers and policymakers at both the EU and national levels, data availability and accessibility have indeed enormously expanded in the past decades, allowing data users to tap vast and expanding sources of information. Notwithstanding such significant improvements, major deficiencies remain when it comes to availability, comparability and quality of data. The production and analysis of population and migration statistics as well as the institutionalisation of data production and dissemination have always been a concern to representatives of both governance and independent research. Authorities were interested in reliable, comparable data for policymaking but at the same time independent researchers and academia had an interest in using statistics for research, sometimes in order to criticise the authorities. While the adoption of Regulation (EC) No 862/2007 led to an increased availability of harmonised data, major issues are still unsolved in relation to data quality and comparability. It is often unclear what national source data are reported to Eurostat, and how they relate to data published on the national level, an issue of particular concern to administrative data. But also in respect to core demographic data, such as data on migration flows or stocks, data reported to Eurostat still often deviate from agreed data definitions. While availability of data on the European level has indeed much improved, their comparability and quality are still major issues. What is more, the increasing scope and depth of data collection on migration and related areas also raises more fundamental issues regarding the meaning (and meaningfulness) of harmonised statistical categories in an expanding European Union, the limits of such harmonisation, and the dangers associated with the use of concepts such as country of birth and citizenship as master frames. These issues will be discussed in the next section.