BRAIN DRAIN – BRAIN GAIN: Portugal 73 8.1 CONCEPTS, SOURCES AND DATA PROBLEMS Although the concept of»immigrant« seems relatively easy to define, migration scholars are well aware of the opera­tional limits of this term(Boyle, Halfacree and Robinson 1999; Koser 2007). Without entering into a discussion about the time limits that separate»simple movements« from migra­tion(for the United Nations, a minimum of three months abroad, for instance) or temporary from long-term migration (Rosa, Seabra and Santos 2004), the question relevant for understanding the data we are using concerns the difference between foreigners and foreign-born. The principles used in Teney’s study follow the tradition of the majority of EU countries and refer to foreigners. As this is the category used in most analyses devoted to the economic – namely labour market – and demographic features of immigration to Portugal 29 (Rosa, Seabra and Santos 2004; Malheiros and Esteves 2013), being also the key organiser of the informa­tion gathered by key providers such as the Foreigners and Borders Office or the GEE 30 /Ministry of Employment, it was not a problem to adopt it in the brain gain section. However, as far as the brain drain section is concerned, data sources were clearly more limited. Actually, to obtain systematic information about Portuguese workers – and particularly highly skilled Portuguese workers – in other EU member-states proved very difficult. To use the EU Labour Force Survey was not an option, not only because the sample of Portuguese people is not large enough in the majority of countries, but also because we were not able to obtain – through specific requests – data for the samples of Portuguese abroad. The alternative was to use DIOC (Database on Immigrants in OECD countries) information, selecting the data for Portuguese abroad, namely those with education levels corresponding to ISCED 5 and ISCED 6(first and second stages of tertiary education). This OECD data are based on national census data(2000/2001 and 2010/2011 in the present case) and counts foreign-born people and not foreign nationals. Therefore, this information is not directly comparable with the data on immigrant workers in Portugal that use nationality and not place of birth as criteria. Because the alternative was a complete lack of data about Portuguese skilled migration by countries of destination, we decided to use the DIOC data. In short, whereas for labour migration to Portugal we use data on foreigners, for labour migration by Portuguese we use data on people born in Portugal. Comparisons between the two movements are thus not possible. However, we can identify evolutional elements for each of the movements and make comparisons between groups of countries, both for 29 It is worth mentioning that several thematic research projects on cultural and social issues related to immigration(for example, integration processes, material culture, housing, immigrants descendants), as well as specific studies about some immigrant groups(for example, ethnic Indians, people from Portuguese-speaking African countries) adopt a criteria that is not based in nationality but on ethnicity or even place of birth(Baganha and Góis 1998/1999). 30 GEE stands for Gabinete de Estratégia e Estudos do Ministério da Econo­mia(Office for Strategy and Research of the Ministry of Economy) arrivals in the case of immigration and for destinations in the case of Portuguese departures. An additional problem associated with identifying interna­tional migrants(both immigrants and Portuguese emigrants) relevant to the present analysis relies in the specific free cir­culation conditions within the European Union(Teney 2014). In fact, although some administrative registration tends to occur, the free circulation right given to EU citizens abolished the need for permits and visas and frequently people do not register in the administrative systems of destination countries (for example, social security, local registers) in the initial pe­riod of migration or if the period of stay is perceived as likely to be short. Therefore, this imposes limits on how far the number of EU foreigners registered in several administrative sources reflects reality and leads to a more extensive use of both sample data and census-based information. The second thing that could pose some difficulty was the definition of»highly skilled«. Generally, the easiest and most generalised way of establishing that somebody is highly skilled consists in verifying whether they have a tertiary education diploma. This was the criterion used for all data treated in this chapter, although they come from different sources. In the case of data on foreigner workers in Portugal, in both census and GEE/Ministry of Employment data highly skilled workers correspond to people aged 15 or more who have completed at least a three-year university course. Con­cerning Portuguese people abroad, the DIOC database uses equivalent categories of ISCED, namely levels 5 and 6 that correspond to people with first and second stages of higher education. Despite this apparent harmonisation of categories, we must be aware that national education systems and clas­sification education levels have differences and reclassification by international organisations, such as the UN or the OECD, is frequently not linear or direct. 31 In addition to the conceptual issues, other limitations on information comparability rely in the collection systems that lie behind the databases that we have used in our analysis. For instance, although census data for 2011 and GEE/Ministry of Employment data for the same year correspond apparently to the same population of highly skilled EU foreigners, the first source totals around 20,000 people and the latter only 6,000. This difference shows that the first database is more encompassing than the second one because it incorporates non-working highly educated foreigners, such as students and retired people. Furthermore, the universal application of the census takes in more workers than the Employee registers of the Ministry of Employment that exclude workers from the Autonomous Regions of Madeira and Azores and do not include employees on public administration contracts. Never­theless, this last source is a continuous register that provides comparable yearly information about workers. Therefore, we have decided to use both data sources – with some caution – because they complement each other. 31 On this see, for instance, the OECD notes on the methodology behind the construction of DIOC databases in the education variable.