The politics of numbers. Framing intra-EU migrants in the Netherlands

ABSTRACT Numbers and data collection play a key role in political framing. To explore how this works, this article provides a case-study analysis of European mobile workers in the Netherlands. Following the increase of intra-European mobility, European mobile workers have emerged as an important but contested policy target group for receiving countries. This article examines how numbers contributed to the framing process of this topic. This study reveals how frames legitimize the strategic usage of numbers and the indication of issues as problems. This contributes to our understanding of the importance of numbers in the framing process of target groups.


Introduction
At a certain moment we observed a stagnation in the increase of official registrations of intra-EU migrants (. . .). But when a new minister came to office and interpreted the same figures, he said: 'the amounts are still increasing'. Which are both true, but it's about how you frame it, you can go any way. Framing is very important in this issue, in a lot of political dossiers, but certainly in this dossier (MCS3) How can we understand the above statement of a Ministerial civil servant and grasp the relevance of numbers in the 'very important' issue of framing? From a rationalist perspective, numbers provide an objective means of identifying issues or problems and thus pinpointing areas for governmental intervention (Niskanen 2007). Various studies, however, have proven this to be far more complex. The methods of quantified data collection (Porter 1995), the selection and definition of items on which data are collected (Jasanoff 2004), as well as the definition of groups involved (Pierce et al. 2014;Schneider and Ingram 1997;Yanow 2003) are often much more contested.
To understand this contestation, we connect the issue of numbering to the framing literature in public policy studies. While various studies consider the role of stories, categories, symbols, myths, and metaphors, less attention has been paid to the contribution of numbers in framing studies (Lakoff and Johnson 1980;Stone 1988;Yanow 2003;Bacchi 2009). Instead, numbers have mostly been studied as an instrument to 'account' (Miller and Napier 1993), as population management (Rose 1991), as 'rhetoric of objectivity' or uniform communication (Desrosières 1998;Porter 1995;Rojo and van Dijk 1997;Cohen 1999). This is an important gap, since 'numbers, in fact, work exactly like metaphors' (Stone 1988, 165).
In this study, we approach numbers as socio-political constructions. We adopt a constructivist approach in which we focus on political actors and authorities involved in the framing process of 'intra-European movement'. In that process, we understand the importance of 'numbers' in the discursive legitimation of 'frames'. We examine how authorities 'clothe their choices in objective technical terms' or in 'simply rationalizations' (Schneider and Ingram 1997, 158-165). By this approach, we contribute to the role of data and statistics in the framing of target groups and in legitimating action toward these groups (van Leeuwen 2007;Pierce et al. 2014).
'Intra-European movement' is a case where numbering of involved groups has played a key role. In the European Union (EU) enlargements of 2004 and 2007, various Central and Eastern-European (CEE) countries joined the EU, such as Poland, Hungary, and Romania, which provided access to the right of free movement within the EU. 1 While the European Commission framed this as 'mobility' and defined these groups as 'EU mobile citizens' making use of their right of free movement, local and national governments in countries such as the Netherlands framed these groups as 'CEE migrants' and their movement as 'migration'. Or, even more specific, Dutch authorities framed these migrants as 'MOElander' ('MOE' as Dutch abbreviation of 'CEE') (Letter to Parliament 2001, 5). In that regard, the Dutch government claimed to have received an inflow of 'more than 300,000 MOE-landers' after 2004 and local governments advocated specific measures to stimulate the 'integration' of these ' MOE-landers' (Statistics Netherlands 2011;Municipality Rotterdam 2013). Key in this political agenda setting involved the data production by various Dutch research agencies and 'planning offices' to indicate the population size and the socio-economic position of these 'CEE migrants'. As such, numbering plays a key role in public and political debates, yet, we do not fully understand what role numbering plays in framing specific target groups. Therefore, this article examines how numbers contribute to legitimate frames of 'intra-European movement'.
For that purpose, we include an empirical case study on 'intra-European movement' in the Netherlands. This serves as a strategic case study because of the perceived relevance to monitor and register this population in the Netherlands. We pose the following research question: 'How has numbering legitimized the framing of "MOElanders" in The Netherlands?' We will address this question through an interpretative case study analysis of official Dutch documents regarding 'EU migrants' and a series of semi-structured interviews with respondents involved in policymaking as well as in registration and data collection.

The politics of framing
From a social-constructivist perspective, policy problems are not 'discovered', but socially constituted as part of a broader social and interactive process. The notion of frames emphasizes how policies are based on plot lines and storytelling that 'holds disparate elements together long enough to energize and guide action' (Weick 1995, 61).
Convincing frames are an ideological and political result of contingent disputes defined as a graspable situation one can act upon (Rein and Schön 1977). It includes a comprehensive definition of social problems related to relevant political solutions. The concept 'frame' points at the more static concept by which subjects and issues are addressed. Instead, framing points at the more interactive, dynamic, and politically sensitive understanding to conceptualize frames. As such, we approach framing as an intersubjective and socio-political process (van Hulst and Yanow 2016). More specifically, we approach framing as how specific language and concepts are used (naming), how involved groups are defined (classifying), and how a causal story with prescriptive solutions is constructed (narrating).
First, by naming, policy-relevant actors draw on language that reflects their understanding of the issue. By the creation of concepts, labels, and metaphors, issues can be defined (Rein and Schön 1977;van Hulst and Yanow 2016). Second, framing involves classifications or categorizations of 'the social' in groups and categories that are distinguished by a convincing set of characteristics (Bowker and Star 1999;Starr 1992;Bacchi 2009). To classify a group is not a neutral but primarily a social and a relational activity reflecting a hierarchical order (Hacking 2007). This occurs by the introduction of new categories, which include or exclude people; reifying the social world they refer to (de Zwart 2005). For instance, it differs throughout societies whether 'Jews' are classified in terms of 'ethnicity', 'race', or 'religion' (Starr 1992). As such, classifications can contribute to the 'recognition of shared characteristics', to create 'a bond of uniformity' or to a 'shared community' (Schneider and Ingram 1997). Third, framing ensures that these names and categories are positioned in a certain story, a narrative with a problem and a convincing perspective on its control (Stone 1988). With these elements, policy framing can include a feed-forward effect, which could trigger all kinds of consequences such as policies, instruments or beneficial arrangements (Schneider and Ingram 1997). This is the 'normative leap', the call for policy action, which is its performative potential (de Zwart 2005; van Ostaijen 2016).
By the elements of naming, classifying and narrating, policy frames can be constructed and studied to display how specific values are attributed to specific groups. When political actors highlight an issue, communication informs with the intention to manipulate, silence or mystify some elements (Wodak 2011;van Hulst and Yanow 2016). We are interested in how political actors construct legitimate names, classifications or narratives with the use of numbers. In the following, we will explain what we understand by numbers and numbering.

The politics of numbering
Policies can become more convincing and persuasive by the strategic usage of data, statistics and registrations (Edelman 1971;Yanow 2003). Data and statistics are very important as 'input' for policy design and are used to make policies persuasive (Cohen 1999). Especially on target groups, statistical data plays a key role in policy design processes since numbers can create benchmarks if presented as 'hard data' (Desrosières 1998;Simon, Piché, and Gagnon 2015;Porter 1995).
The use of numbers seems very natural, objective or neutral, because of its 'ordering capacity' and technicality by which one can objectify, quantify and depoliticize subjective problems (Porter 1995;Rose 1999;Cohen 1999). But from a social-constructivist perspective, numbers are not 'discovered' and data collection on populations is not a neutral form of counting but rather social-politically constituted, since numbers include deliberate decisions about how to count and why a phenomenon is perceived interesting enough to bother counting (Simon 2005;Schinkel 2013;Starr 1992;Stone 1988, 172). As such, numbering is a socio-political process of inclusion, exclusion and selection, since numbers make it possible for governments to govern (Rose, 1999). These numbers are not just 'out there', since actors have agency and strategic interests in numbering activities (Porter 1995;Jasanoff 2004). In this context, it is important to study by whom, how and why numbers are produced. This is why we focus on numbering. We approach numbering as the socio-political process of knowledge production by actors in the use of numbers (Porter 1995). We understand numbering by the elements of selection of topics on which numbers are collected and methods of data collection (Starr 1992;Simon, Piché, and Gagnon 2015).
First of all, the selection of topics or domains on which numbers are collected (or equally important, on those which numbers are not collected) is often not merely an expert choice, but part of a socio-political process (Salter 1988;Jasanoff 2004). Topic selection involves a focus on which topics data is collected. Second, methods of data collection focus on how data is gathered. Both can be affected by institutional constraints or strategic motivations. For instance, in some countries like the United States, data collection about ethnicity is based on a census where citizens declare their ethnicity themselves (Simon 2005). In other countries, ethnicity is 'deduced' from a broader set of data on family decent, which has large consequences for the definition of 'minorities', 'migrants' and 'inflows'. By these elements of topic selection and methods of data collection we understand numbering as a socio-political process of 'making up numbers'. 2 We expect a relevance of numbering in framing processes. By this conceptual distinction and operationalization, we aim to study numbering and framing as separate conceptual phenomena. This leads to the following overview in Table 1 which concisely summarizes the operationalized elements by which we analyzed framing and numbering:

Relating numbering and framing
To study how numbers legitimate frames, 'legitimation' is conceptualized as the creation of a sense of understandable, necessary or acceptable actions in a specific setting (Van Leeuwen and Wodak 1999). It is the justification of salient elements of the institutional order by giving a normative dignity to its practical imperatives (Boltanski and Thevenot  . In contrast, delegitimation establishes a sense of negative, morally reprehensible or otherwise unacceptable actions (Rojo and van Dijk 1997;van Leeuwen and Wodak 1999). We study the legitimation of practices of a specific institutional order by discursive practices, and by discursive legitimation. From a discursive perspective, legitimacy is created within specific discourses, since discourses produce knowledge and 'provide the 'frames' with which people make sense of particular issues and give sense to them' (Vaara and Tienari 2008, 987).
To study discursive legitimation, we take up the work of van Leeuwen (2007, 91) who distinguished four key categories of discursive legitimation: authorization, moral evaluation, rationalization, and mytho poesis (van Leeuwen 2007;Rojo and van Dijk 1997;van Leeuwen and Wodak 1999;Vaara and Tienari 2008, 988). As such, this focus on discursive legitimation enables us to study how numbers discursively legitimate frames. Table 2 concisely summarizes the elements by which we operationalize discursive legitimation.
These elements of legitimation can occur separately or in combination with each other and can be used to legitimize or delegitimize a social act (van Leeuwen 2007). By the distinction of these four elements, we analyze how numbers discursively legitimate frames. We expect relevancy of numbers within framing processes, since we presume that numbers are not just 'out there'. But, it is quite unknown how to understand this relevancy, especially in this case. Based on previous studies, we expect that numbers are mostly used to objectify or depoliticize political discourses. Hence, we expect that numbers, as 'rhetoric of objectivity' (Rojo and van Dijk 1997;van Leeuwen 2007) legitimize frames by rationalization or objectification. In that interactive process, we focus on political actors, experts and professionals who generate knowledge, and have the (institutional) justification to legitimate its validity (Hacking 2007, 297).

Methods
This study is based on a qualitative analysis of framing and numbering activities in the field of 'intra-European movement' in The Netherlands. To unravel the dynamics between framing and numbering, 'intra-European movement' is selected as an embedded case study. The Dutch case is selected as it involves a (relatively) new 'migrant' category, which means that processes of categorization may be relatively new and visible. The Dutch case is also selected because of its historical specificity in ethnic registrations (Bijl and Verweij 2012), since data collection on migrants exists longer in the Netherlands than in other countries, such as in France (de Zwart 2005). The Dutch case is a strategic case study to analyze the relevancy of numbering in framing processes. Specifically, to study this relevancy, we focus in particular on the position of governmental organizations and data producers. First, we did a qualitative desk research of the policy process, second we interviewed respondents and stakeholders and third we focused on how knowledge and information circulated between those actors. As such, we focused on a wide variety of data and knowledge producers, such as universities, NGO's and private research agencies related to this topic such as the Dutch Social and Cultural Planning Office (SCP) and the Statistics Office (CBS). Important to mention is that our analysis was not limited to these agencies and that we did not pre-defined the selection of respondents in our research design. The point of departure was not the institutional differences of these agencies, but when data have been used or was of relevance for (policy related) stakeholders in this topic. Furthermore, our analysis focused on the period 2000-2015, a couple of years before the Dutch borders 'opened' for 'mobile workers' from Central and Eastern Europe until recent policies.
First, the research design was based on a qualitative document analysis. This included primary (policy documents, Letters to Parliament) and secondary (literature) sources covering policies regarding EU mobile workers and 'migrants'. Additionally, this involved an analysis of primary (research reports, advisory studies) sources, of all relevant data producers within the topic at hand. More specifically, we studied all studies spurred by Ministries and that were reported within Letters to Parliament.
Subsequently, we completed two rounds of interviews. First, we held 12 qualitative semi-structured interviews with stakeholders regarding issues of registration in general. This included national and municipal policymakers, lobbyists and private parties active with 'intra-European movement' in their daily work. This round of interviews sensitized our perspective on registration issues and its consequences in the Netherlands in general. This evolved into a second round of 10 additional indepth semi-structured interviews with different experts such as national policymakers, researchers and statisticians at data offices. These interviews were held in the summer of 2015 and focused on the preliminary issues found in the first round of interviews.

Data analysis
The desk research was primarily executed by online searches of Letters to Parliament by the number of a dossier. 3 The basis of the search was on the Dutch translated terms 'mobility', 'migration' and 'movement'. From here, four dossier numbers rose, of which '29407' was the most relevant ('free movement of employees from new EU member states'). This dossier included 332 documents (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015), which have been interpretatively studied by their abstracts and selected on their qualitative relevancy. It resulted in 53 documents which contributed to an extensive data file. This, together with the research documents produced by research agencies on this topic collected by extensive agency searches made up our data file. Second, the qualitative semi-structured interviews were conducted along pre-structured interview guidelines. The experts for the interviews were selected by snowball sampling because this study focuses on numbering and framing, the references and network capital of respondents were important to get access to authorities in the field. This holds some biases, since we had to rely on the willingness and cooperation of respondents to get to additional informed respondents. This may have led to a very specific sample of respondents, but because of their expertise this specificity was needed. We did not only focused on a specific set of respondents, but also included respondents with a variety of affiliations, such as ministerial, municipal, and NGO related respondents. This mixture enabled us to check the data from a variety of perspectives along the way. Next to this, our desk research also made it possible to triangulate and check the reliability of our interview data by multiple sources.
All experts agreed to a recorded interview, which were transcribed and transposed to the program ATLAS.ti, a software program used for analyzing qualitative data. The interview data were analyzed along the codes and sub-codes derived from the operationalization of the grid (explained below). This has been done by hand-coding and highlighting all relevant words, phrases and paragraphs in the documents that fitted the grid. By this approach, this type of research can be indicated as a matter of back-and-forth reasoning (Berg and Lune 2004) or abductive research (Timmermans and Tavory 2012). With an abductive approach, the researcher went back and forth between theoretical concepts and the empirical findings. By doing this kind of analysis, the researcher distinguished frame elements in the empirical data. This analysis is based on an analysis of the empirical material by predefined codes, the deductive grid of framing, numbering and legitimation elements (comprehensively visualized by Table 1 and 2). As such, framing has been analyzed by the elements of naming, classification and narrating. Numbering has been analyzed by the elements of topic selection and data collection. And legitimation has been analyzed by the elements of authorization, moral evaluation, rationalization and mythopoesis. For aims of transparency and reliability of the analysis, these elements are included between parentheses (for instance by '[naming]') in the analysis paragraphs to transparently report to the reader which elements we attributed to which data.
We will now first contextualize the Dutch case study. Second, we will reconstruct the findings in three periods and after the presentation of these findings, we will finalize with a concluding section.

Framing and numbering regarding intra-European movement
The Dutch context Migration or mobility from Central and Eastern Europe should be positioned in the institutional context of EU policies and the process of European integration (Gabriel and Pellerin 2008). Since the Treaty of Rome in 1957, leading up to the Schengen Agreement (1985), freedom of movement was of major importance in the European collaboration between member states. An important focus of the EU is to guarantee labor mobility, to enlarge the flexibility in European labor markets to strengthen the Single Market by removing 'barriers', 'obstacles' and 'disturbances' that limit flexibility on the labor markets (European Commission 2002. Removing barriers also means that citizens of EU member-states are not perceived as 'migrants' under EU law, but as 'mobile workers' (Favell 2008;Guild and Mantu 2011). This contrasts with EU policies aimed at Third Country Nationals (TCN's), migrants from outside the European Union, which still need specified permission to get access, work and residency in the EU. Although national and local governments are free to adopt integration measures on a voluntary basis, mandatory measures or any provision related to entitlements are in this perspective perceived as illegitimate 'disturbances' of the EU principle of free movement.
The political context of that moment, with fierce political turmoil in the aftermath on the assassination of right-wing politician Pim Fortuyn, stimulated an anti-immigrants debate. The Dutch context was strongly fuelled by politicized backlashes on 'immigration', 'islam' and what has been termed 'the multicultural drama' (Scheffer 2000). This also legitimated restrictive transitional arrangements, which caused that accession state nationals could travel to and settle but had limited access to the Dutch labor market. By this intervention, the Dutch government aimed to better regulate the inflow of accession state nationals, mainly based on their labor market participation. This resulted in the Netherlands where for instance Polish workers needed a work permit until May 2007 and Romanians and Bulgarians workers until January 2014 to be employed in the Netherlands. Since the lifting of these transitional arrangements, 'migration' from CEE countries and 'migrant' monitoring and registration became a key political concern in the Dutch national political arena in the late 2000s. Not only on the national level but also on the local level of municipalities who asked attention for the 'tsunami' of Eastern Europeans (Zuidervaart 2010). Previous research showed three distinct discursive periods or frames regarding 'intra-European movement' at the Dutch national level: from a more legal-economic discourse (<2007), to a legal socio-cultural discourse (2008-2011) toward a merging of both in the final phase (2011>) (van Ostaijen 2016). We will now concisely summarize these three periods.
The first period is characterized by the transition period installed until 2007, which imposed strict limitations on mobility of EU citizens from new member states. In this period discourses held a mere economic perspective on the intra-EU mobility paradigm and a strong liberal focus on the 'opening' of the borders by minimizing 'administrative burdens' and maximizing the impacts of 'the four freedoms'. Attention is mainly devoted to the most profitable timing in the implementation of new EU legislation for the Dutch economy and to optimize the EU's internal market function (Letter to Parliament 2004a). Indicative, Dutch policies are termed 'accompanying policies', because legislation should not hamper the free movement aims. This first period is thus characterized as a legal-economic discourse. The second discursive period (2007)(2008)(2009)(2010)(2011) is characterized by the lifting of the temporary restrictions and shows a shift toward socio-cultural and legal issues. There is more focus on social 'problems' and 'reverse sides' such as housing irregularities and fraudulent practices. Moreover, in this period, all kinds of parallels are made to 'learn lessons of the past', implicitly defining this too as a migrant (and integration) issue with a focus on housing, language and neighborhood problems (Letter to Parliament 2009; 2011a). 4 Consequentially, all sorts of 'pilots', 'Action Plans' (2011) and 'an integral packet of measures' are proposed to counteract these 'reverse sides' and benefit from those 'lessons' of that past. Thus, this second period is characterized as a legal socio-cultural discourse. Third, both discourses earlier identified get a close issue connection in the final period (>2011). In this period, discourses include both legal-economic and socio-cultural elements by connecting aims to stimulate the labor market with interventions to regulate housing issues and the societal participation of migrants. All kinds of laws were proposed to control housing issues more effectively and to regulate irregularities in the labor market, in order to combat 'shadow sides' (Letter to Parliament 2013c, 175) and to make EU migration 'maintainable ' and'compliable' (Letter to Parliament, 2013a, 172, 2013b, 174). This turned a focus on exploitation in the Dutch labor market, to illegal temporary labor agencies and affordable housing. Thus, the final period is characterized as the merging of both legal-economic and socio-cultural discourses. These three periods constitute the background of the in-depth analysis.

First period (<2007): naming and categorizing 'MOE-landers'
In the pre-2007 period, the issue of European mobility is mostly covered in terms of 'European enlargement' and seen as a positive development (SER 1993, 7). It includes a strong distinction between 'Eastern' and 'Western' Europe, sometimes with a label such as 'Middle and Eastern Europe' (SER 1993(SER , 1999(SER , 2001 [naming]. The label 'Middle and Eastern Europe' originated in a study on the expected trading contributions of potential member states by a 'gravity model' (Gros and Steinherr 1993). 5 Within this 'gravity model', all countries behind 'Eastern Germany' were classified as economic 'periphery' with 'limited gravity' (SER 1993, 152). In this study the label '(Middle and) Eastern Europe' was mainly based on economic index data, territorial proximity and geopolitical developments at that time [classification].
This was not unimportant because this early characterization of 'Eastern Europe' resonated in other official documents. Of large importance was the European Integration Consortium report (Boeri and Brücker 2000) which defined these countries as 'CEECs' (Central and Eastern European Countries). In discussions about European enlargement, the Dutch Cabinet explicitly referred to this report in terms of 'labour migration', 'flows' and the 'migration' of 'migrants' from 'Middle and Eastern European countries' (Letter to Parliament 2000, 3). One year later, the Dutch Minister declared that he had attention for impacts of 'labour migration' and 'mobility' from 'MOE-landen' (again, 'MOE' is the Dutch abbreviation of 'CEE') (Letter to Parliament 2001, 5). From this onwards, the label 'MOE-lander' originated, referring to a migrant who originates from one of the Central or Eastern European countries. As such, this new policy category was strongly characterized by geopolitical developments, territorial enlargements and net economic benefits of member-state accessions [naming and classification]. To contextualize the origination of this label, Figure 1 shows the development of the label 'MOElander' within Dutch media (national and local newspapers) 6  Because the pre-2007 period was characterized by a transitory regime, the Dutch Cabinet could regulate the 'inflow' of 'labour migrants' by means of 'labour market permits'. As such, positioned within the broader EU mobility paradigm, intra-EU mobility was framed purely as a form of labor migration (Letter to Parliament 2004b, 17) and migration as labor market issue [narrating]: The fundament of economic cooperation within the EU is consisted by the Internal Market. The goal of this is to achieve a European Economic Space wherein 'the four freedoms' are secured. The Netherlands has, as trading nation, large interests with a good functioning of that internal market which is free of unneeded trading barriers. (Letter to Parliament 2005, 2) Within that labor market narrative, the Cabinet expected an inflow of 'gross 10.000 temporary labour migrants' with a 'maximum limitation of 22.000 migrants' before May 2005, which may not be 'trespassed', otherwise policies need to be reconsidered (Letter to Parliament 2004b). By May 2005, however, the amount of labor market permits had 'doubled' to almost 25,000. This was more than the critical juncture set before, but this increase was presented as 'minimal' (Letter to Parliament 2005). One year later, the Cabinet wanted to know if labor market limitations could be removed and requested the policy advisory office Ecorys to estimate the population size [topic selection]. They indicated (on the basis of secondary data) that the 'gross labour migration in 2006 will be between 53,000 and 63,000 persons' (Ecorys 2006, 198). The Cabinet presented this as a 'limited' number, and decided to remove all limitations on labor market access combined with an extra focus on 'irregular employment', 'false self-employment' and 'uneven competition' next to the already existing 'accompanying policies' (Letter to Parliament 2006b). The economy needed 'extra hands' and the Cabinet was willing to liberalize the labor market [narrating]. If there were deviancies, they were relativized by the argument that, in general the overall perspective held a mere positive trend and that history has shown a different trend, stated by the Minister of Social Affairs: By references to the past and with new statistics showing even more shortages, the liberalization argument was legitimated. An analysis of the pre-2007 period shows that, with labor market permits as primary source [data collection], it became legitimate to frame the issue as a labor market issue and EU citizens as labor migrants. This data collection made it legitimate to construct European mobility as labor market activity: It causes extra economic growth, more dynamics in the corporate sector and more labour positions. These positive effects are hard to measure or to visualize than when in an individual case Dutch labour supply is repressed by labour supply from the new memberstates. (Letter to Parliament 2006a, 3) By regulating a delineated amount of people to the Dutch labor market, EU mobility was framed by the Dutch state as labor mobility and labor migration. And since this migration originated from the 'MOE-landen' this made it legitimate to label these 'migrants' as 'MOE-landers'.
To conclude, by referring to the tradition of a 'trading nation' and to previous acts, laws and European Treaties, Dutch authorities legitimized 'intra-European movement' mainly as a labor market issue [authorization/mythopoesis]. These legitimized data collection by measuring labor market permits. Limitations on labor market permits were legitimized with arguments to ensure economic growth and prosperity connected with measures to protect certain vulnerable domains in the Dutch economy [moral evaluation]. Interestingly, increasing numbers, and the consequences of increasing numbers, were delegitimized on the basis of past experiences [mythopoesis]. First, new research included new definitions dividing the population into permanent and temporary groups. Since citizens who stay shorter than 4 months are not obliged to formally register themselves at the Municipal Registration Office (GBA), it was stated that these so-called 'flexible' groups of mobile workers remain out of sight for official authorities (Letter to Parliament 2007, 8). Consequently, the Minister turned his attention on data collection methods since 'this group does not register themselves at the GBA, it must be thought through if and in which way a better view can be possible' (Letter to Parliament 2008a, 3). This resulted in attempts to connect information systems of the Tax Service and the Employment Centre enabling 170 municipalities to get a better insight of the residential addresses of temporary labor migrants (Letter to Parliament 2008a, 10).
Second, new research identified 'problems' on 'integration issues of migrants', such as 'mixed marriages, social contacts, modern attitudes and language' (Risbo 2008). Surveys indicated that 'three quarters of the MOE-landers do not or speak minimal Dutch' (Risbo 2009, 14). This caused a change in characterizations: 'three studies 8 show that MOE-landers have limited contact with people out of their own group' (Letter to Parliament 2009, 7). The Cabinet found it important to 'be alert and attention for integration is needed. It is societally unwanted that large groups of newcomers have an isolated existence' (Letter to Parliament 2009, 4). Therefore, new data collection changed the [classifications] toward migrants and their sociocultural position: With the arrival of labour migrants from MOE-landen, the parallel with the sixties and seventies forces itself, when also large extensions of groups came to the Netherlands. [. . .] Now, after forty years, there are still efforts to overcome these backlashes. We can't allow that in a certain time again an extensive group stayed unnoticed and which came at large socio-economic distance. (Letter to Parliament 2009, 4) This integration-related narrative contrasted with the previous economic narrative. New [data collection] was legitimated because of references to the past and previous insufficient integration policies. Therefore, the Cabinet concluded that 'it is important to monitor this group well (by repeated, comparable research) and to keep a renewed integral policy approach of the composition in education-, labour market position, integration and criminality ' (Letter to Parliament 2008a, 4).
Since the Cabinet could no longer rely on labor market permits, they asked several research institutes to estimate the amount of 'MOE-landers'. One policy research institute estimated the amount on 'minimal 100,000 persons' based on Employment Centre data (Regioplan 2008), while the Dutch Statistics Office (based on municipal registrations) reported about '70,000 people with a nationality from one of the MOEcountries' (Letter to Parliament 2007, 2). From this new data collection, the Cabinet concluded that there 'is certainly no rise in the amount of labour migrants' (Letter to Parliament 2007, 2). This approach was continued in the upcoming years 'because the Netherlands is not a preferred country for this group' (Letter to Parliament 2008a, 5). . In a reaction to these new data, the Minister of Social Affairs pointed at the methodological limitations of these studies, problematizing the variety in the studies. He contested the topic selection and data collection processes since he doubted how groups were grouped as 'MOE' and to what extent groups were well distinguished between 'temporary' or 'permanent' categories and their way of data gathering (Letter to Parliament 2008c, 2). Significantly, data collection about this 'MOE group' moved to the top of the political agenda.
To conclude, this period is characterized by the removal of labor market permits and a search for alternative monitoring. Consequentially, this legitimized a variety of data collection to keep the population in sight [authorization]. Additionally, those studies included a focus on socio-cultural issues, legitimized by references to insufficiencies of past integration measures [mythopoesis]. Moreover, growing numbers were delegitimized on the basis of insufficiencies in the data collection process [rationalization]. It caused a narrative change from migrant laborers toward labor migrants, which changed the classification of this 'group' and 'topics' of new data collection.
Third period (2011)(2012)(2013)(2014)(2015): legitimating numbers The third period was characterized by the start of a new Cabinet (led by the Liberal Party) and a new Minister of Social Affairs (Kamp, Liberal Party) coming to power in 2011. This Minister changed the tone of the issue, as a civil servant indicated that 'we got a General on the top. That was a new and confrontational experience' (MCS2). This new tone can be illustrated by the approach on population size, since this Minister stated: The increase of the amount of labour migrants from MOE-landen has been much larger than expected. In the first estimation in 2004 it was about 15,000, in 2007 the expectation was that 100,000-120,000 labour migrants would work in the Netherlands, according to the last estimation it is in the meanwhile almost 200,000ʹ which could be expected to increase in the upcoming years. (Letter to Parliament 2011a, 1) The Minister of Social Affairs thus put attention to the 'sharp increases' from the past, the low expectations and the 'expected increases' for the future adding up to 'almost 200,000 labour migrants'. The amount '200,000 labour migrants' was legitimated by an estimated '165,000 migrants' of earlier research together with 18,000 registered or 27,000 estimated self-employed (Risbo 2009). However, this has been indicated as 'an underestimation of the factual amount of self-employed' (Letter to Parliament 2011a, 4). Three months later, all of a sudden the Statistics Office also confirmed the Ministerial statement of '200,000 labour migrants'. By a 'new approach', as one respondent indicated it, connecting two population sizes (of municipal data of '117,000 MOElanders' combined with employee data of '81,000 employees'), they confirmed the new statement of the Minister that there were 'more than 200,000 MOE-landers' in the Netherlands (Statistics Netherlands 2011) [data collection]. Subsequently, this new number of '200,000' legitimized a wide variety of 'problems', 'negative effects' and 'nuisances' (concerning registration, labor market fraud, social security, housing, language and integration) (Letter to Parliament 2011a). New data collection made new problems 'legible' and contributed to a focus on 'integration' [narrating].
Notwithstanding the fixation on the new number of '200,000', the existing official data collection system was regarded inadequate: 'The total number of migrants from MOE countries is unknown. This is mainly caused because of a large part of them is not registered. We are left with estimations' (Letter to Parliament 2011a, 3). By referring to new research it is stated that sometimes 'up to 80% of the total population' did not register in the Municipal or Alien Administration (Risbo 2009) and the Minister demanded clear data, illustrated by a Ministerial civil servant: 'Our own director and Minister asked repeatedly: With how many are they? How many are there? And a sort of incomprehension of why can't you just tell this to me? Because we do know everything in the Netherlands and why can't you just tell me how many Polish people are here?' (MCS1) This led the Ministries of Social Affairs and Internal Affairs to spur Utrecht University to estimate the total population size. By a capture-recapture approach, coupling the databases of the municipal administration systems (GBA), Employment Centers (Polis) and the Recognizing Service System (HKS) of the Police, this study gave an estimated overview of registered and non-registered MOE-landers (van der Heijden, Cruyff, andvan Gils 2011, 2013) [data collection]. This resulted in estimations 'between 260,000 and 305,000 MOE-landers' in 2008, '286,000 and 325,000 MOE-landers' in 2009and an amount of '340,000 MOE-landers' in 2010(van der Heijden, Cruyff, and van Gils 2011. The data collection was based neither on interviews, surveys nor employer statistics but on estimations of coupled data systems and 'on statistical methods' (Letter to Parliament 2011b, 2). This resulted in amounts that were 'much higher than those circulating before' (Letter to Parliament 2011b, 2). While the Cabinet acknowledged the methodological limitations of this data collection, the number '300,000 migrants' 'caused an enormous shock' (MCS1) and became an overall reference point (Letter to Parliament 2011c), confirmed by a Ministerial civil servant: 'Because those numbers were so high, that remained the number what was mentioned all the time. Van der Heijden set the bar very high that other figures always were a bit disappointing. [. . .] That became the number where everybody worked with at that time' (MCS1) These new numbers gave input for new narratives, such as that: 'We must ensure that with the expected increases of the amount of labour migrants that the problems will not increase as well' (Letter to Parliament 2013b, 6). As such, the selection of topics also became characterized by 'disadvantages' and 'problems' that could lead to 'uneven competition, exploitation and overcrowded housing' (Letter to Parliament 2012, 1) [narrating]. In the meantime, other research (Engbersen et al. 2011) indicated that substantial parts of the Polish migrants held high unemployment rates (13%), which was 'three times as high as the autochthonous population' (Letter to Parliament 2011c, 132, 25). As a consequence, the Cabinet concluded that new data collection laws were needed to register all persons from the beginning of their stay in the Netherlands (Letter to Parliament 2011a, 6, 2011b, 3). The Minister of Internal Affairs concluded that: 'the high percentage of non-registered underlines the importance of the recently developed policies [. . .]. Especially the measures taken around the improvement of registration of labour migrants plays a role' (Letter to Parliament 2011b, 3).
We can conclude that the delegitimation of data collection legitimated the demand for new data collection, by estimations. Ironically, when these estimations were ready, they again legitimated the demand for new data collection, by refined registrations. Consequentially, the Minister declared that 'the Cabinet strives, together with the Statistics Office and the Social and Cultural Planning Office, to get more unity in the usage of definitions and estimation-methods' (Letter to Parliament 2011c, 4). To meet that aim, the Ministry of Social Affairs spurred the Statistics Office to develop an overview study 'Migrant monitor 2007-2012' for a comprehensive insight on the total population of migrants from the EU. 10 On the basis of both data sources, a new label was introduced, 'EU-10', which consisted all former 'MOE-8 countries' including Bulgaria and Romania [naming]. On the basis of this data collection, the Cabinet concluded that 'this group increased from 139,090 persons to 236,620 persons in the first quarter of 2012' (Letter to Parliament 2013b, 2) and then to '246,660 persons in the last quarter of 2012' (Letter to Parliament 2014, 4). In the end, this 'unified' data collection revealed a strong discrepancy between registered data of Bulgarian and Romanian migrants of the Statistics Office ('almost 32,000 persons') and estimations of Van der Heijden ('110,000 migrants'). On the basis of this discrepancy, the Cabinet concluded: 'The big difference between the amount of registered migrants from Bulgaria and Romania and the estimations of prof. dr. Van der Heijden emphasizes the need to improve the registration, harmonizing the data systems and to attack illegal labour and irregular employment'. (Letter to Parliament 2013b, 3) It shows that delegitimized numbers can also legitimize new laws and legislation. Furthermore, while the Cabinet initiated all kinds of new policies, they were confronted with lower numbers on two topics. First, they were confronted with relatively small amounts on welfare claims (by the Statistics Office), but the Minister stated that: 'Despite this is about a relative small amount of people, it does not mean that the Cabinet will not put efforts to prevent a "honeypot effect" on the Dutch system' (Letter to Parliament 2013a, 6). By references to an unknown future the Minister legitimized repressive measures to prevent 'welfare tourism'. Second, the European Commission publicized a 'fact finding' study on 'the social security claims and non-contributive cash benefits' of intra-EU migrants concluding that 'welfare tourism' is not or has limited presence in Europe. As a response, the Dutch Minister stated that 'this conclusion does not hold any guarantee that this will not occur in the future' and that '[. . .] problems are not always findable in statistics. While the bottlenecks caused by freedom of employees are certainly visible in society' (Letter to Parliament 2013c, 1-2). Furthermore, the Minister pointed at the methodological limitations and that: 'comforting developments in the past are no guarantee for the future' (Letter to Parliament 2013d, 2). Next that it shows a struggle to push forward a particular narrative with certain numbers, it also shows that by references to an (unknown) future [mythopoesis] or limitations in data collection [rationalization] alternative policy solutions are legitimized.
To conclude, in this period there was an urge to delegitimate past data and to legitimate improved future data [mythopoesis/rationalization]. By this new estimated data, 'improved' policies were legitimized to 'combat' 'illegal labour' and 'irregular employment' and to contribute to 'the manageability of EU-labour migration' (Letter to Parliament 2014, 7). Even when numbers were low, this was delegitimized by reference to an unknown future or by methodological limitations [mythopoesis/rationalization]. It shows discursive legitimation as strategic process.

Conclusion
This article posed the question 'How has numbering legitimized the framing of "MOElanders" in The Netherlands?' In the first period, characterized by a more legal-economic framing, the population was indicated and monitored by means of labour market permits. This was legitimated by references to the Netherlands as a 'trading nation' and past agreements on the European level [authorization/mythopoesis]. Consequentially, this category was characterized as foreign laborers on the Dutch labor market. Increasing numbers or problems were delegitimized by references to the past [mythopoesis]. In the second period, labor market permits could no longer monitor the population size. Consequently, this legitimized [authorization] the demand for a range of studies to focus on socio-cultural issues, such as migrant integration, in order to overcome insufficient policies of the past [mythopoesis]. Increasing numbers were delegitimized by references to a variety of methods of data collection [rationalization]. In the third period, previous data collection was problematized since new research showed the importance of temporary groups, which put pressure on past registration procedures [rationalization]. This resulted in estimations which showed steep 'increases' and made it possible to problematize and delegitimize former registration procedures which legitimized new repressive policies [rationalization/mythopoesis]. As such, numbers contributed to the discursive legitimacy of specific political frames, and by focusing on the case of 'MOE-landers', this article displayed how this works.
Theoretically, we expected that numbers legitimize frames by rationalization, as 'rhetoric of objectivity' (Rojo and van Dijk 1997). While we indicated that politicians usually 'clothe their choices in objective technical terms' or in 'simply rationalizations' (Schneider and Ingram 1997, 158-165) our study shows that numbers not only legitimize by rationalization, but also by authorization and mythopoesis. This shows the broad and important function of numbers in the discursive legitimation process, delivering new theoretical insights (van Leeuwen 2007; Rojo and van Dijk 1997;Vaara and Tienari 2008;Vaara and Monin 2010). In line with this, our analysis shows that numbers are not static endeavors, but contingent and on-going sociopolitical constructions, and need to be analyzed as numbering. By numbering, authorized actors can legitimize their framed position. And depending on the contingent frame adaptation of authorized actors, numbers contribute to the framing of 'MOElander' issues.
Second, our study shows that numbers do not have a sole capacity to legitimize frames, but it also shows the relevancy and importance of numbers in legitimating certain frames. Our analysis displays that numbers do not have autonomous authority to speak 'truth to power'. Numbers do not have the authority to define what can or cannot be said, what is legitimate or not to be spoken about (by rationalization). It shows that actor frames can (de-)justify certain data as important or irrelevant on the basis of past performances, unknown futures [mythopoesis], insufficient data collection [rationalization] or institutionalized laws and legislation that push forward a certain decision [authorization]. Like discourses, frames produce knowledge. This nuances the autonomous position of data and numbers in the production of public policy. Nevertheless, our case also reveals the relevance and importance of numbers in the legitimation of an issue as a problem. Or as one statistician noted: 'in the usage of data there are always different points of view. And which point of view you use depends on what you want to describe' (RSO). As such, this study displays the politics of numbers since the legitimacy of numbers depends on the authority of actors, stated by a Ministerial civil servant: 'you have to interpret the numbers and that is politics' (MCS3).
This points out at our third concluding point, since this study not only displays how but also reveals why certain actors are able to impose their perspective on a certain moment. Not coincidentally, our case study focuses on political and policy discourses, frames and narratives on 'intra-European movement'. As such, we focused on 'political knowledge' displayed in power relations of political authorities who mainly have the institutional authority to impose a certain worldview on an issue (Wodak 2011). Why some frames prevail in this case study is much influenced by the fact that Ministers have the institutional capacity to use and neglect certain knowledge to push forward a particular agenda. This political-institutional dimension is important to understand why certain numbers prevail to justify certain frames in a specific period. Next to this, the institutional or professional loyalty of policymakers toward their political principal shows the importance of political authority to understand why certain knowledge prevails in a certain period. It is part of the administrative professionalism and loyalty why certain numbers are considered relevant or significant in a specific period of time. And since an important element of 'political knowledge' is also the ability to manipulate, silence or mystify knowledge, this discursive-institutional or politicalinstitutional dimension is important to understand why and how certain authorities can impose a particular perspective on an issue (Schmidt 2011;Wodak 2011;van Hulst and Yanow 2016). This article revealed the contingent development of these political particularities in the Dutch case on 'MOE-landers'.
Fourthly, our study displays the importance of mythopoesis (reference to issues in the past or future) in the analysis of numbering and framing. It points to the importance of time. The recurrent usage of mythopoesis shows the importance of time scales, timing and time horizons to (de)legitimize numbers in a certain policy approach ('unknown future') or in the delegitimation of past policy approaches ('We can't allow that in a certain time again an extensive group stayed unnoticed'). This points at the importance of how the past, present and future are discursively constructed to (de-)legitimize a certain number. It reveals that numbers can be delegitimized by numbers. This contrasts with metaphors, since metaphors cannot be delegitimized by metaphors. It shows that 'numbers do not exactly work like metaphors' (Stone 1988).
Fifthly, our study speaks to the broader literature on the politics of numbers. Speaking to the more (post-) structuralist, Foulcauldian or constitutive body of literature on numbering (Rose, 1999;Schinkel 2013), this study shows the importance of actor agency to understand frame structures. Within a post-structuralist approach, actors, subjects and agents are the means by which formative rules, structures and mechanisms are displayed. However, our study shows the importance of actor agency to overcome the 'structuralist trap' regarding individuals solely as bearers of a certain structure. It shows that personal, political and professional affiliations of actors contribute to why and how numbers are legitimized or delegitimized.
Finally, speaking to the broader literature on (im)migration studies and studies on 'CEE migration' (Favell 2008;Gabriel and Pellerin 2008;Guild and Mantu 2011) this study unravels the normative dimension in 'migration' research. This article nuances the autonomous and neutral position of numbering agencies, research institutes and academic communities. Therefore, this study adds critical awareness and reflexivity toward the socio-political context in which migration studies as numbering activity of it self is also active. This could be an interesting starting point, looking at the politicalnormative contributions of migration scholars and research, for conditioning future 'mandated' science and research [mythopoesis] (Salter 1988