Collectivized Discretion: Seeking Explanations for Decreased Asylum Recognition Rates in Finland After Europe's 2015 “Refugee Crisis”

In 2015, during the so-called “refugee crisis” in Europe, Finland was among the European countries receiving exceptionally large numbers of asylum applications. As the volume of asylum applications surged, however, the percentage of positive asylum decisions in Finland declined substantially. In this article, we explore reasons for this dramatic drop in recognitions rates and examine Finnish immigration control authorities’ use of discretion in asylum credibility assessment. Our approach is unique in its application of mixed methods to examine asylum decisions in pre- and post-crisis situations. We found that asylum caseworkers’ inconsistent assessment of similar facts and lack of faith in the veracity of applicants’ claims were essential to the mass denial of young Iraqi asylum applicants in Finland. This finding is important because it illustrates how asylum officers are able to “shift the border,” or generate a shift in asylum decision-making on a grand scale, without meaningful changes in law. Asylum officers, we show, are able to bring about such a shift via what we call collectivized discretion, or large-scale use of discretion, in asylum status determinations to control migration. Prior research on discretion in asylum decision-making highlights the individual decision-maker. This article expands discretion research by offering new insights on large-scale, collective discretionary shifts in the application of asylum law. We conclude that it is crucial that asylum status determinations be anchored in the individual assessment of each applicant's case, as collectivized discretion can lead to arbitrary results in the application of asylum law, potentially forcing those in need of refugee protection to face deportation.


Introduction
Finland's short-term objective is to stop the uncontrolled flow of asylum seekers into our country, to bring asylum costs under control, and to integrate effectively those who have been granted asylum. 1 Finnish Government, Government action plan on asylum policy (2015).
The number of asylum applications surged in Europe in 2015, 2 and Finland was among the European countries that received an exceptionally high number of applications. 3 The European Union (EU) 4 and its individual member-states like 1 Finnish . Finland 5 perceived this surge in asylum applications as a "crisis" in Europe because, although the conflicts contributing to forced mobility occurred outside Europe, the surge in asylum applications posed a "management crisis" within Europe (Brekke andStaver 2018, 2178). While the number of asylum applications increased across the whole of Europe in Fall 2015, the share of positive asylum decisions dropped substantially in Finland from 2015 to 2016. 6 This drop is surprising because an increase in the number of asylum applications should not automatically result in a decrease in the percentage of applications accepted. This article explores the reasons for Finland's sudden mass denial of asylum applicants in the context of a perceived refugee crisis and Finnish immigration control authorities' use of discretion in asylum credibility assessment. We compare the decision making of the Finnish Immigration Service's (hereafter, Migri) officers as they applied asylum law to individual asylum claims during two distinct time periods: in 2015, immediately before the intense peak in the number of asylum applications in Europe, and in 2017, soon after the peak. Our data consist of 243 initial international protection decisions (i.e., asylum, subsidiary protection, 7 or residence permit on compassionate grounds) regarding Iraqi applicants aged 18-34 years (2015: n = 125; 2017: n = 118). We focus on Iraqi citizens of this age group because they were the largest asylum applicant group in Finland in both periods. 8 From our study's outset, statistics demonstrated that Iraqi applicants' recognition rate plummeted from 2015 to 2017 in Finland, with those granted such protection having previously been in the majority and later in a distinct minority.
Cynthia Gorman (2017) discusses the role that adjudicators have in controlling the number of those deemed to meet the definition of a refugee. As Gorman argues, adjudicators can "shift and change" the border, seen as dynamically constituted through a specific kind of law-making called case law, that is, via new interpretations of asylum law (Gorman 2017, 37-44). Adjudicators can, thus, use legal interpretations, a part of what Gorman (2017) calls "bordering work," to exclude certain applicant populations from an asylum when, for instance, authorities frame the numbers of a particular group of asylum-seekers as constituting "a crisis of illegal immigration" (Gorman 2017, 40). Building on Gorman's ideas, we argue that bordering work also takes place at the stage of initial asylum decision-making and that immigration officers, too, can shift the border collectively by using the interpretive powers afforded them by existing legislation. Our findings suggest that what we call collectivized discretion, or large-scale use of discretion in asylum determinations to control migration, shifts borders as effectively as changes of law, be it statutory or case law. Immigration officers, we show, can draw on legal frameworks in a flexible and instrumental manner and, thus, influence the outcomes of policies and laws because they enjoy a large measure of discretion, as has been demonstrated in numerous studies (Meyers and Vorsanger 2007, 154; Alvesalo and Whyte 2007;Dörrenbächer 2017;Alpes and Spire 2014), and because such decision makers have both the power to choose between and among legal alternatives and a considerable scope for making decisions on both fact and law (Tucker et al. 2016, 67).
Asylum decision-makers, of course, do not make legal interpretations in a vacuum or neutral space devoid of extra-legal considerations. Instead, discretion is shaped by the pressures that decision-makers face from their peers and superiors, and politics and the media can also affect decision-makers' everyday practices (Affolter 2021, 17). Prior quantitative research on the impact of extra-legal factors on asylum status determinations implies that asylum control authorities make use of discretion to control migration on a large scale. Rottman, Fariss and Poe (2009), for example, who examined asylum recognition rates in the post-2001 environment in the United States, observed that between 2002 and 2004, asylum claims were about 7 percent less likely to be accepted than before the September 11 attacks in 2001. Exploring recognition rates in Western Europe from 1980to 1999, Neumayer (2005 noted that recognition rates were lower if many asylum applicants from a specific country had already applied for asylum in the past. While these studies showcase that some scholars have examined asylum officers' possible interpretive shifts quantitatively, more substantive research is needed on how asylum officers generate such shifts. This article addresses this need by examining asylum decision-making, both quantitatively and qualitatively, in the context of a perceived refugee crisis and in a preand post-crisis situation. Although some research shows that asylum officers acknowledge that discretion is "an inescapable feature" of asylum decision-making (Magalhães 2018, 382), there are few quantitative, let alone mixed-method, studies of discretion in asylum control. In research on asylum credibility assessment, discretion is typically examined as a phenomenon relating to the individual decision-maker whose personal judgment, as Kagan put it, "is inconsistent from one adjudicator to the next" (2003,367; see also Herlihy, Gleeson and Turner 2010;Gill et al. 2018). Whether discretion in asylum decision-making can take more collective forms, however, is less understood. This article addresses this gap in scholarship on both discretion and asylum credibility assessment.
We argue that a mixed-method approach is needed to explore immigration officers' use of collectivized discretion, as such an approach enables us to examine both the scale and substance of the use of collectivized discretion in the mass denial of young Iraqi asylum-seekers in Finland. To investigate the reasons for the drop in recognition rates for this group, we analyze, first, whether there were any quantitative changes from 2015 to 2017 in the asylum applicant pool or in which aspects of asylum applicants' statements Migri deemed credible. Next, we examine whether there were any qualitative changes in how Migri justified whether to grant international protection and whether Migri's reasoning changed between the periods.
To contextualize our arguments, we, first, discuss the discursive and policy shifts in Finland that paved the way for the use of collectivized discretion during and after the peak of the perceived refugee crisis in Fall 2015. To provide further background for our analysis, we, then, outline Finland's legal framework for international protection. From there, we describe our data and methods and present our results and conclusions. As we argue, by employing the discretionary margin afforded by existing law, asylum officers in Finland were able to restrict access to asylum in a highly scalable, yet almost inconspicuous, manner, thus contributing to the mass denial of asylum applicants. In doing so, we illustrate how collectivized discretion can be used to restrict populations deemed eligible for asylum and to shift borders.

Shifts in Political Discourse and Asylum Policy: Delegitimizing Migration
Since 2000, Finland has typically received 1,500-6,000 asylum applications annually, 9 but in 2015, that number jumped to more than 32,000. 10 The Finnish government's immediate response to this increase seemed to be characterized by solidarity: in September 2015, the Prime Minister Juha Sipilä urged everyone to consider what 9 Finland, Ministry of the Interior. 2020. "Refugees flee persecution in their home countries," intermin.fi/en/areas-of-expertise/migration/refugees-and-asylum-seekers [accessed: January 5, 2021]. 10 Migri. 2020. "Statistics," tilastot.migri.fi/#applications/23330/49?l=en&start=540&end= 551 [accessed: January 5, 2021]. they could do to help refugees who came from distressing conditions. 11 Soon, however, the Finnish government assumed a different tone, publishing one action plan on immigration and another on asylum policy with numerous proposals to curb immigration within weeks of Sipilä's statement about helping refugees. 12 The asylum policy action plan's opening line, cited at the beginning of this article, stated that Finland's objective was to stop the "uncontrolled flow" of asylumseekers. The water metaphor of "flow," like other naturalized metaphors, draws a parallel between asylum-seekers and an uncontrolled force of nature, serving to dehumanize, create panic, delegitimize migration, and externalize responsibility (Malkki 1996;Horst 2018;Petersson and Kainz 2017;Schapendonk 2012). As Kainz (2016) suggests, when migratory movements are portrayed as "unstoppable and thus threatening," restrictive policies are more easily perceived as necessary for regaining control.
Finland is a part of the EU, which, since the 1990s, has attempted to harmonize the legislative framework pertaining to refugee protection. 13 As Wahlbeck (2018) notes, Finland has long supported international cooperation in refugee protection. Yet during the perceived refugee crisis in Fall 2015, Finland was reluctant to support EU-wide approaches to asylum policies (Wahlbeck 2018), seeming to pursue national approaches instead. For instance, in October 2015, Migri reassessed two significant origin countries of those seeking asylum, Iraq, and Somalia, 14 concluding that the security situation in both countries had improved and that, in future international protection decisions concerning applicants from these countries, individual grounds for asylum would be emphasized. 15 The Ministry for Foreign Affairs and Migri also funded Facebook campaigns to inform potential asylum-seekers, in Arabic and Somali, of Finland's increasingly restrictive migration control policies. 16 In October 2015, the Finnish Ministry of the Interior issued a press release lauding how "Finland's determined action to manage migration has proved successful," 17 as the number of asylum applications had declined by 35 percent from September to October 2015. 18 In the same period, the number of applications filed by Iraqi asylum-seekers had dropped by 43 percent. 19 Beyond these prominent policy changes and an anticipated decrease in asylum applications, which Finnish authorities openly pursued, there is a more surprising shift: the substantial decrease in the share of positive international protection decisions. Figure 1 illustrates the changes in international protection status for Iraqi asylum-seekers in Finland and its neighboring country, Sweden, in relation to the European average. Before the crisis, from 2008 to 2014, the annual share of positive decisions varied between 50 percent and 68 percent in Finland and between 40 percent and 70 percent in all EU member states.
In 2015, Iraqi applicants' recognition rate increased to 83 percent in Finland and in the EU (average). The following year, however, witnessed a general decrease in the proportion of positive decisions in Europe, but this decline was particularly dramatic in Finland. In 2016, positive decisions were granted to a mere 19 percent of Iraqi applicants in Finland and, in 2017, to 25 percent of Iraqi applicants. Such a drastic change did not occur in all EU member-states, although the overall recognition rate decreased from 83 percent in 2015 to 57 percent in 2016. In Sweden, the recognition rate remained rather stable over a ten-year period in 2008-2017, varying between 16 percent and 36 percent. 16 Helsinki Times. 2015. "Ministry for Foreign Affairs to continue its communication campaign for asylum seekers," news report, December 17, 2015, helsinkitimes.fi/finland/finland-news/ politics/13664-ministry-for-foreign-affairs-to-continue-its-communication-campaignfor-asylum-seekers.html [accessed: January 5, 2021]. 17 As the Ministry noted, however, the desired decrease in the number of asylum applications was not solely the result of Finland's actions but, instead, the cumulative result of several factors, many of which were beyond the country's borders (Finland, Ministry of the Interior (2015), "Drop in number of asylum seekers-new estimate for this year," press release, October 14, 2015, intermin In the press releases issued after the perceived refugee crisis, Migri and the Finnish Ministry of the Interior provided three main reasons for the drop in asylum recognition rates. First, in a March 2018 press release, Migri suggested that the asylum-seeker profile had changed in Finland in Fall 2015 and was different from that seen in Sweden and Norway because a large number of young Iraqi men had arrived in Finland. 20 Although a greater share of asylum applicants in neighboring countries were allegedly Iraqi families with children, Migri suggested in a September 2017 press release that many newcomers to Finland were unaccompanied. 21 At times, discussion around these arguments in the Finnish media had a moralizing, gendered tone, deploying the narrative of the "undeserving economic   Migri. 2018. "All circumstances affecting the matter must be taken into account in asylum decisions," press release, March 22, 2018, migri.fi/en/artikkeli/-/asset_publisher/ turvapaikkapaatoksia-arvioitaessa-huomioitava-kaikki-asiaan-vaikuttavat-seikat [accessed: January 5, 2021]. 21 Migri. 2017. "Two years since the autumn of 2015-what is the asylum seeker situation now?," press release, September 29, 2017, migri.fi/en/artikkeli/-/asset_publisher/kaksivuotta-turvapaikanhakijasyksysta-2015-missa-mennaan-nyt-[accessed: January 5, 2021]. migrant" and seeing "young, healthy, working-age men" as less deserving of international protection. 22 Second, in a June 2018 press release, Migri and the Finnish Ministry of the Interior suggested that more applicants were coming from more peaceful regions of Iraq or from regions deemed more secure, especially southern Iraq. Finnish immigration authorities also determined that the security situation in Baghdad had improved to the degree that it was "considered a possible area of internal flight" (i.e., a relocation alternative within the country). 23 Both explanations delegitimized a particular population's right to asylum, as the explanations framed the newcomers as migrants without a valid need for international protection.
Third, in a March 2018 press release, the Ministry of the Interior asserted that Migri had sought to harmonize Finland's asylum decision-making practices with other European countries, particularly Sweden. 24 This explanation justified mass denial of asylum applicants with the desire to align domestic asylum decision-making with that in other countries, implying a level of indifference for the individual assessment of each applicant's case. In what follows, we analyze whether the first and second reasons (i.e., changes in the applicant profile and changes relating to the regional security situation in Iraq) are reflected in our data of written asylum determinations concerning young Iraqi applicants (the third reason falling outside our data's scope). Having reviewed the shifts in policy and discourse, we now turn to examine the international protection statutes in force at the time studied.

Legal Framework for International Protection
The Finnish legal framework regarding people seeking international protection is a combination of domestic, European, and international input (Kuosma 2016, 1). Although the Constitution of Finland does not include a specific right to asylum, it does provide protection against non-refoulement. 25 As an EU member-state, Finland 22 Suomen Kuvalehti. 2016. "Näin perussuomalaiset sanelivat Suomelle uuden pakolaispolitiikan-kahdessa tunnissa paperi oli valmis," news report, February 12, 2016, suomenkuvalehti. fi/jutut/kotimaa/nain-perussuomalaiset-sanelivat-suomelle-uuden-pakolaispolitiikan/?shared= 309331-23cde557-500 [accessed: January 5, 2021]. 23 Finland, Ministry of the Interior. 2018. "Report: Still room for improvement in the asylum procedure but no systematic errors," press release, June 14, 2018, valtioneuvosto is also bound by the EU Charter of Fundamental Rights, 26 which establishes both the right to asylum and the principle of non-refoulement as part of the EU legal order. These obligations are further binding for Finland through its international obligations, which are incorporated into the domestic legal order through Acts of Parliament. 27 After incorporation, treaties can be applied in domestic courts, like any Acts of Parliament. 28 As far as international protection is concerned, the most relevant international obligations have been implemented by the Aliens Act (301/2004), which has also transposed the relevant EU legislation regarding the protection of foreign nationals, in particular the Qualification Directive (Directive 2011/95/EU), into domestic law. According to Section 87 of the Aliens Act, foreign nationals residing in Finland "are granted asylum if they reside outside their home country or country of permanent residence owing to a well-founded fear of being persecuted for reasons of ethnic origin, religion, nationality, membership of a particular social group or political opinion and if they, because of this fear, are unwilling to avail themselves of the protection of that country." Furthermore, as provided by Article 15 of the Qualification Directive and Section 88 of the Aliens Act, a foreign national may be granted "a residence permit on grounds of subsidiary protection if the requirements for granting asylum under section 87 are not met, but substantial grounds have been shown for believing that the person, if returned to his or her country of origin or country of former habitual residence, would face a real risk of being subjected to serious harm." In addition to the aforementioned grounds for international protection, the Finnish Aliens Act includes a provision on issuing a residence permit on compassionate grounds. 29 During the study period, no legislative amendments were made to the provisions concerning asylum or subsidiary protection, but the category of humanitarian protection was erased from the Aliens Act in April 2016. Previously, humanitarian protection was granted to applicants who did not qualify for asylum or subsidiary protection but who could not be returned to their home countries due to a poor security situation, poor human rights record, or an environmental catastrophe (Aliens Act 301/2004).
Decisions on international protection in Finland are made at the first instance by Migri. 30 As part of its decision-making procedure, Migri conducts an asylum interview to establish the grounds given by the applicant for the persecution they have faced in their home country or country of permanent residence or for other violations of their rights or related threats. 31 All Migri's residence permit decisions may be appealed to an administrative court. 32 Subsequent appeals to the Supreme Administrative Court are possible, provided that the Court grants leave to appeal. 33

Data and Methods
The data analyzed in this article consist of positive (i.e., awarding asylum, subsidiary protection, or residence permit on compassionate grounds) and negative (i.e., denying residence permits in any of the aforementioned categories) international protection decisions concerning 18-to 34-year-old Iraqi citizens (n = 243). We use a total of 125 decisions made from April to August 2015 (census, complete enumeration). From 2017, we use a total of 118 decisions made from June to August (simple random sample, partial enumeration). 34 The two time periods in 2015 and 2017 were chosen to compare the decisions before and after the perceived refugee crisis that peaked in Fall 2015.
In May 2017, the University of Turku Faculty of Law and the Institute for Human Rights at Åbo Akademi University began collaboration on this study. 35 The Non-Discrimination Ombudsman in Finland granted a research permit allowing the research group to use confidential written asylum decisions as data. We have published one report in Finnish (Saarikkomäki et al. 2018) and a summary in English. 36 In this article, we extend the analysis and theoretical interpretations from the descriptive report and its summary (Saarikkomäki et al. 2018; see also footnote 36).
From the international protection decisions, we coded information regarding our research aims into an SPSS data matrix. Variables consisted of, for example, the 31 Finland, Aliens Act (301/2004), Sections 97 and 97a. 32 Finland, Aliens Act (301/2004), Section 190. 33 Finland, Aliens Act (301/2004), Section 196. 34 The analysis and coding phase started in September 2017. Therefore, the data period ends in August 2017. We chose to focus on the 2017 summer period from June to August, during which Migri made 322 decisions. As coding the decisions was time consuming, we did not code them all. Instead, we used a simple random sample of 118 decisions. In Summer 2015, the number of decisions was not enough to reach the 120 decisions indicated by power analysis; thus, we also included April and May, which enabled us to reach a total of 125 decisions. 35 The collaboration was carried out in the context of three Academy of Finland research projects: Migri's assessment of the case's facts, the caseworker's justification for the decision, and whether the caseworker believed the applicant. We created most variables based on the conditions for providing international protection to asylum applicants, as established by Chapter 6 of the Aliens Act (e.g., acts of persecution specified in Section 87a, reasons for persecution specified in Section 87b, and criteria for subsidiary protection provided by Section 88 of the Aliens Act). However, while reading the asylum decisions, we found that some additional categories were needed to provide more detailed information about phenomena that are not explicitly mentioned in the abovementioned provisions of the Aliens Act (e.g., kidnapping). We used cross tables as a statistical analysis method and Pearson's chi-squared test to test for statistical significance. We conducted a power analysis to estimate the smallest sample size suitable for the statistical analysis and concluded that a sample of around 120 (for each period) would be suitable for our analysis (cross tables). 37 However, regarding a few variables, there were not enough cases to present statistical significance, as we indicate in the tables ("+ too few cases to present statistical significance"). We then carried out a qualitative content analysis to examine the types of argumentation used in the different periods. Our quantitative and qualitative findings are described, respectively, in the next two sections.

Results of Quantitative Analysis Recognition Rates
Our data indicate that in the studied 2017 period (hereafter "in 2017"), decisionmakers granted international protection significantly less often than in the studied 2015 period before the perceived refugee crisis (hereafter "in 2015"). In 2015, decision-makers rejected 14 percent of international protection applications and issued a residence permit to the vast majority (86 percent) of applicants. By contrast, in 2017, decision-makers rejected the majority of international protection applications (79 percent) and granted a residence permit to only 21 percent of applicants (observed significance of change p < .001). The number of those receiving international protection dwindled in all remaining 38 categories of international protection in 2017, as, in addition to fewer applicants receiving asylum, virtually no applicants 37 Power analysis can be used to estimate a sample size which reveals an effect if one is present (Uttley 2019.) The analysis requires random sampling. 38 The criteria for humanitarian protection was erased from the Aliens Act in April 2016 (for more information, see section Legal framework for international protection), however, even before the category was removed, the humanitarian protection criteria were not applied in any case in 2015.
received subsidiary protection (Table 1). In this article, we explore potential reasons for Finland's sudden mass denial of young Iraqi asylum applicants. Accordingly, in the following sections, we present quantitative findings as to whether there were differences between 2015 and 2017 in the asylum applicant pool or asylum credibility assessment.

Applicants' Backgrounds
What explains the dramatic drop in the percentage of positive asylum decisions from 2015 to 2017? To address this question, we first studied whether there were quantitative changes in applicants' socio-demographic characteristics between the periods, in terms of gender, family ties, and religion. We included these variables to examine the veracity of the Finnish authorities' suggestion that the composition of asylum applicants changed in the 2015 crisis (i.e., more unaccompanied young men and more applicants from safer areas of Iraq). However, as we explain below, we were only able to conduct the analysis in part. We discovered no statistically significant differences between the two periods in terms of asylum applicants' gender or family structure. The shares of men, women, and applicants with families (spouses and underage children included in the application) were the same in both periods. In 2015, 87 percent (n = 109) of applicants were male, and 13 percent (n = 16) were female. In 2017, 89 percent (n = 105) of applicants were male, and 11 percent (n = 13) were female. In 2015, 12 percent (n = 15) and in 2017, 20 percent (n = 23) of applications included a spouse. A child or children under 18 years of age were included in 13 percent (n = 16) of 2015 applications and in 21 percent (n = 25) of 2017 applications. As discussed above, top authorities in Finland suggested that the drop in recognition rates was due to an increase in the share of young unaccompanied men (gendered discourse of "undeserving economic migrants"). We did not find support for this reasoning, as the proportions of men, women, and families remained unchanged.
As regards religion, in both periods, Sunni Muslims were the largest religious group. In 2015, Sunni applicants formed a large majority (76 percent), and the share of other religious groups (Shia 4 percent, Muslims not specified 5 percent, Christians 0 percent, and other, applicants with missing, unclear information 13 percent) and atheists (2 percent) was much smaller. In 2017, Sunni applicants were still the largest religious group (44 percent), and the second largest group was Shia Muslims (27 percent). The other groups remained small: Christians (4 percent), Muslims not specified (1 percent), atheists (4 percent), and other, unclear or missing information (20 percent). Concerning all religious groups, there were so few observations in certain groups that we could not demonstrate statistical significance. In fact, the decrease in the share of Sunni applicants from 76 percent in 2015 to 44 percent in 2017 (p < .001) and increase in the share of Shia applicants from 4 percent to 27 percent (p < .001) constituted the only change in the asylum applicant pool we observed. Whether this decline in the share of Sunni Muslims had some impact on the drop in recognition rates is beyond the scope of our analysis, and more research is needed on the potential role of asylum applicants' religious background in Finland's drop in recognition rates after the perceived refugee crisis in 2015. However, asylum applicants' religious background is unlikely to be the key explanation for the drop, as recognition rates plummeted in a similar fashion for all three of the largest asylum applicant groups in Finland from different origin countries (in 2016, citizens of Iraq, Afghanistan, and Somalia, as specified in footnote 6).
In Migri's asylum decisions, the geographical area used as the basis for decision making is typically indicated as "Your application will be assessed regarding [a given region]" [authors' translation]. We intended to investigate whether asylum applicants in 2017 originated from areas deemed "more peaceful parts of Iraq" or "areas of improved security," as Finnish authorities suggested, by comparing the areas indicated in the decisions against Finnish policy guidelines for different parts of Iraq. However, in both periods, we found that decision makers did not indicate any part of Iraq in the vast majority of asylum decisions and were, therefore, unable to carry out an analysis that would yield reliable results. In 2015, decision makers clearly indicated the area in 14 percent of decisions (n = 17). In ten decisions, the area indicated was "Baghdad," with no specification as to whether "Baghdad" referred to the capital or the surrounding region. In 78 percent of the 2015 cases (n = 98), the area indicated was simply "Iraq" (n = 39) or no region at all (n = 59). In 2017, decision-makers indicated the area clearly in 24 percent of decisions (n = 28). In 22 of the 2017 decisions, decision-makers indicated the area merely as "Baghdad," despite the capital and surrounding region being assessed differently in the 2017 policy guidelines regarding the level of violence. In 68 decisions, accounting for almost 60 percent of the data, decision-makers indicated the area as "Iraq" (n = 42), or listed no region (n = 26).
Even though the percentage of negative decisions increased considerably from 2015 to 2017, decision-makers' precision in specifying applicants' origin area increased only moderately (from 14 percent in 2015 to 24 percent in 2017). Because such a large portion of the decisions had missing information regarding the area, it was practically impossible to examine whether the regional security situation contributed to the drop in recognition rates. We cannot rule out the influence of geographical variation in applicants' origins or of increased safety in a particular region on asylum status determinations, which may have implicitly factored in the decisions. Our findings suggest, however, that this facet of decision-making was not made explicit to most asylum applicants. Such a large share of missing information is problematic and calls into question the transparency of asylum decision-making.
To recapitulate, applicants' socio-demographic background appeared to be similar in both periods, and there were no clear differences in the variables we measured that could explain such a major drop in recognition rates. This finding is interesting, as Finnish authorities claimed that, in the perceived refugee crisis, the asylum applicant pool had changed and that more newcomers were unaccompanied young men. Gender and family status, however, stayed the same, and the age and citizenship were controlled for as the data concerned 18-to 34-year-old Iraqi citizens. It is worth noting that the decisions were not transparent on applicants' origin areas, rendering it impossible to assess the authorities' claims of applicants' "safer" regional origins. Could factors other than changes in applicants' background, then, explain the drop in recognition rates? We attempt to answer this question next by exploring potential changes in asylum applicants' claims and how Migri received such claims.

Reasons for Persecution
We now examine whether there are quantitative differences between 2015 and 2017 in what reasons applicants reported for persecution and fear of infringements and in how Migri assessed those claims. According to Section 87b of the Aliens Act, reasons for persecution include "ethnic origin, religion, nationality, membership in a particular social group or political opinion." Regarding the most common reasons for persecution applicants provided, we did not find very large differences between the time periods (Table 2). In both periods, religion and political opinion were the most common reasons applicants provided, and membership in a particular social group and ethnic origin were less common. Examining 2017, we discovered that applying for asylum for religious reasons had decreased since 2015 and that applying for asylum for political reasons had increased. Providing other reasons was rare in both periods, and we observed no changes over time. We then looked at how often Migri deemed the reasons credible and observed that Migri accepted the reason for persecution significantly more often in 2015 than in 2017 ( Table 2). The most notable change was how often Migri accepted religion as the reason for persecution. In 2015, Migri accepted religion as the reason in almost all applications in which it was provided (90 percent), while in 2017, it did so in only one-fifth of applications.

Fear of Infringements
We now take a more specific look at whether asylum applicants had experienced previous infringements, what applicants feared, and whether there were changes between the time periods in how often Migri accepted their claims. Typical acts of persecution, specified in Section 87a of the Aliens Act, include various forms of violence, such as mental and physical violence, and public authorities' discriminatory practices. We explored whether applicants reported having been detained, imprisoned, kidnapped, tortured, or subjected to other forms of violence or public authorities' discriminatory practices and whether Migri considered these claims plausible.
Another key issue is whether Migri deemed that past infringements could indicate future threat. The well-established starting point in asylum law, reinforced by Article 4(4) of the Qualification Directive (Directive 2011/95/EU), emphasizes that an applicant's having previously been subjected to persecution or serious harm is a serious indication of their fear of persecution being well-founded or of a real risk of suffering serious harm in the future. 39 Thus, we examined whether there were changes from 2015 to 2017 in to what extent Migri considered previous infringements against the applicant or their family members as potentially indicative of a future threat. As Table 3 shows, in 2015, 58 percent of applicants reported past persecution experiences as a reason for fearing future persecution if they were to return to their origin country. In 2017, that figure increased to 73 percent. In both periods, nearly all applicants reported fearing future physical or mental violence if they were to return to their origin country (Table 3). In 2015, 3 percent of applicants did not report any fears, and in 2017, 5 percent. Between 2015 and 2017, we did not observe a decline in whether applicants reported past persecution as a reason for fearing future persecution or a decline in applicants' fears of future violence. What is interesting here is that we did observe statistically significant differences as to whether Migri considered applicants' fear of physical or mental violence to be objectively justified. In 2015, decision-makers considered applicants' fears to be objectively justified in 85 percent of cases, whereas in 2017, this figure Table 3. Reasons for Fear of Infringements Provided by Applicants (%) and Reasons Accepted by the Finnish Immigration Service (%) in 2015 and 2017. plunged to 19 percent (Table 3). Overall, authorities considered applicants' fears to be objectively justified in 77 percent of the 2015 cases, while in 2017, this proportion dropped to 20 percent (not shown in Table 3, p < .001, decisions reported 2015: n = 121, 2017: n = 112). Again, we did not observe changes in the reasons applicants provided for their fear, but we did observe changes in whether Migri believed applicants. This shift in percentages indicates that asylum decision-makers' wide discretionary power can lead to immense changes in credibility assessment without a significant change in the applications assessed.
As regards previous infringements, Table 4 shows a similar trend to what we observed regarding other types of claims: there was no change in the share of alleged past infringements, but there was a change in how often Migri deemed these claims plausible. For example, in 2015, decision-makers accepted claims of kidnapping in nearly all instances in which they were made (91 percent), whereas in 2017, decision-makers accepted claims of kidnapping in only one-third of instances. Here, infringements refer to acts specified in Sections 87a and 88 of the Aliens Act: physically abusive or restrictive acts such as detention or imprisonment, torture, or other forms of physical violence or restriction. Because kidnapping appeared in the data, we included it in the coding as a separate infringement category, even though kidnapping is not specifically mentioned in the Aliens Act.
In 2015, only 11 percent of decisions involved an assessment of previous infringements as "isolated incidents with no connection to future threat," Observed significance of change from 2015 to 2017: *p < .05; ***p < .001; + too few cases to present statistical significance. Percentage is calculated column wise for each time period, separately for different grounds of persecution. The table shows only the share of "yes" column; "no" (including also missing cases) is not shown.
whereas in 2017, nearly all decisions involved such an assessment (95 percent). 40 In other words, in 2017, past infringements were nearly always deemed singular, isolated events, not considered to indicate future threat for applicants. This reversed trend in percentages (from 11 percent in 2015 to 95 percent in 2017) again reflects the breadth of discretionary power authorities have as to whether they define past infringements as indicating future threat or as isolated events with no likely link to the future. Thus, what clearly changed between 2015 and 2017 was the percentage of applicants whom Migri believed and whose claims Migri deemed credible. As these findings highlight, asylum officers had the ability to make vast interpretive shifts without substantial changes in legislation or jurisprudence. But was there also a shift in how asylum officers justified and reasoned their decisions? Next, we look at potential differences in Migri's argumentation between 2015 and 2017 decisions.

Results of Qualitative Analysis
As our quantitative analysis showed that Migri did not believe applicants' claims so often in 2017 as it did in 2015, our qualitative content analysis set out to more specifically study the argumentation decision-makers used to justify decisions and whether their reasoning differed between the two periods. Regarding 2015, we were able to explore both positive and negative decisions, but for 2017, we were only able to examine negative decisions, since positive decisions no longer included any explanations or justifications. 41 For decisions in which asylum authorities did not consider the applicant's fear objectively justified (either fully, as in the case of decisions in which international protection had been entirely refused, or partly, as in the case of positive decisions in which some doubts regarding the justification of the applicant's fear had remained), we examined the justifications used for this assessment.
As there were substantially fewer negative decisions in 2015 than in 2017, the volume of argumentation and the array of argument types were equally small in 2015. In 2015, the argumentation against granting asylum focused on the credibility of the applicant's account and on concluding that persecutors did not specifically target the applicant. Overall, in 2015, the threshold for making a negative assessment based on lack of credibility was rather high. Typically, decision makers only refused international protection in instances where the applicant's account was deemed 40 According to Section 45 of the Finnish Administrative Procedure Act (434/2003), authorities do not have to state the reasons for a positive administrative decision if the decision does not concern any other party and if others do not have the right to request a review of the decision. 41 Facing the increasing number of asylum applications, Migri adopted this practice around the turn of the year 2015-2016 to make asylum determinations more efficient (information obtained from Migri by email on March 3, 2020). manifestly unfounded and where the applicant also neglected to provide the requested documentation or evidence. The fact that decisions included expressions of minor doubt regarding the credibility of an applicant's account did not often lead to refusing protection. Moreover, the 2015 data include several decisions in which decision makers considered some parts of the applicant's statement to lack plausibility but did not conclude that the applicant would not be at risk of persecution.
For example, Migri granted asylum in the following case, even though the caseworker noted that the applicant was unable to name any agents of persecution: You are unable to say why you were kidnapped, but you told us that they hate [members of your profession 42 ]. On the other hand, you told us that you think this group kidnapped you because they just wanted money. When the officials asked you after the kidnapping whether you suspected some particular agent of this act, you responded that you could not think of any in particular. 43 Despite concluding that the applicant's account contained these deficiencies and several other inconsistencies, the caseworker decided in favor of the applicant. Photographs allegedly depicting the applicant at their workplace, the applicant's description of the workplace, and the official's conclusion that the origin-country information supported the applicant's account helped secure asylum for the applicant.
In another 2015 case, the applicant did not refer to any infringements but claimed that they had heard from someone that an armed group was looking to recruit them. The applicant claimed that those who refused to be recruited by the armed group were punished by death. The applicant asserted that the armed group wanted to recruit them because of their profession and language skills but failed to provide any documentary evidence of being a professional in their claimed field. Relying solely on the applicant's own account of their studies in that field, the caseworker granted asylum.
It is important to note that the assessment of the threat of persecution in the aforementioned case was based on a potential threat looming in the future, rather than on concrete infringements in the past. In many 2015 cases, decision-makers made positive asylum decisions based on future threat, but in the 2017 data, future threat alone was insufficient, and concrete and robustly verifiable infringements in the past were required for the granting of international protection. In addition, assertions, such as those of the 2015 applicant who claimed to have heard about an armed group's possible recruitment intentions, were often dismissed as "hearsay" in the 2017 data. In 2017, it seems decision-makers typically dismissed elements in an applicant's account that were not directly and personally lived.
In 2017, the argumentation in negative decisions also became highly diverse. In comparison to 2015, the burden of proof seemed to have been reversed. In many 2015 decisions, decision-makers granted international protection unless they could demonstrate unequivocally that the applicant was not in danger. By contrast, in 2017, decision-makers often refused protection if the applicant failed to provide exact details of the persecutors threatening them and why and how the infringements either they or their family members had experienced were linked to the persecutors.
For example, Migri refused international protection in a case in which the applicant claimed that they and their family had been persecuted by a particular armed group. Migri stated: " [Migri] considers it to be your own assumption that the letter your neighbor brought to your family was sent by [a particular armed group]." Regarding the kidnapping of the applicant's family member, the officials stated: [Migri] considers it to be your own assumption that [a particular armed group] was behind the kidnapping. Your [other family member] has supplied eight photographs picturing [a person]. [Migri] cannot verify the origin of these pictures or who is pictured in them. In addition, [Migri] cannot verify how, when and under what circumstances the injuries on the person pictured in the photograph occurred.
In another case in which Migri refused international protection, the applicant had presented a medical report stating that it was possible that the scars on their body were inflicted by torture. Migri concluded that although the applicant's account of their imprisonment and torture had been consistent at different stages of the asylum application process, it has remained superficial and without detail. According to the medical report you provided on [date], the scars observed on your body could have been inflicted by torture in the way you have described. The observed injuries alone do not connect the emergence of the injuries to the torture you claim was inflicted on you by [an armed group]. Your description of your capture, imprisonment and torture [by the armed group] lacks personal elements. [Migri] considers it a fact that you have been assaulted [this way] but concludes, on the aforementioned grounds, that the perpetrator of the assault and its motive remain unknown.
In many 2017 cases, decision-makers interpreted the lack of documentary evidence corroborating an applicant's account as indicating a lack of credibility. Interestingly, however, in many cases, decision makers also dismissed documentary and photographic evidence as unverifiable and, therefore, not substantiating the applicant's case. Decision-makers deemed unverifiable not only, for example, document copies but also originals, as demonstrated by the following statement from a 2017 case: You have presented a copy of a document regarding [a particular matter]. You have told us that you can supply the original document to [Migri], if needed. [Migri] states that the original document is not needed because its origin cannot be verified. For the same reason, no decisive importance can be attributed to the document.
Overall, in 2017, decision-makers typically argued that the applicant's account was "superficial," "inaccurate," or "contradictory." Several decisions included arguments to the effect that the connection between the infringement experienced by the applicant or their family member and the persecutor named by the applicant was based on the applicant's own assumptions. In these cases, decision-makers often argued that the applicant was not specifically targeted by persecutors and, therefore, not in any personal danger. In 2017, it appears, applicants were required to provide detailed information on who was threatening them, for what reason, and why the threats were directed at them specifically, before the decision-maker deemed it plausible that the applicant needed protection.

Conclusion
Young Iraqi asylum applicants faced a sudden mass denial in Finland in the context of the 2015 perceived refugee crisis in Europe. In this article, we explored the reasons for this dramatic drop in asylum recognition rates and Finnish immigration control authorities' use of discretion in asylum credibility assessment. Between the time periods we compared (before the crisis in 2015 and after the crisis in 2017), we observed no shift in the asylum applicant pool that could explain the drop. The share of men, women, and applicants with families remained the same in both periods, challenging Finnish authorities' moralizing, gendered notions that while neighboring countries received families with children, many newcomers to Finland were unaccompanied young men. We also did not observe a decline in how often applicants reported past persecution or fear of future violence. The only change observed in the applicant pool was that the share of Sunni applicants, as compared to Shia applicants, dropped (32 percentage points) in 2017, yet the Sunni continued to be the largest applicant group in both periods. Whether this decline had some impact on the drop in recognition rates is beyond our study's scope. However, asylum applicants' religious background is unlikely to be the key explanation for the decline in recognition rates, as recognition rates dropped in a similar way for all three largest, arguably heterogeneous, asylum applicant groups in Finland from different origin countries. Rather, the mass denial of young Iraqi asylum applicants in Finland shows how asylum officers are able to "shift the border" (Gorman 2017) on a grand scale through collectivized discretion.
We demonstrated that immigration officers' use of collectivized discretion after the perceived refugee crisis (in 2017) was manifested in their lack of faith in the veracity of applicants' claims. Collectivized discretion was visible in asylum credibility assessments in two ways. First, in quantitative terms, in 2017, asylum officers were less likely to believe applicants' reasons for seeking asylum or their claims regarding past infringements and future risks than in 2015. Second, in qualitative terms, in 2017, asylum officers assessed similar facts inconsistently, as compared to 2015, and demonstrated a lack of faith in the veracity of applicants' claims. Our findings, thus, indicate that both quantitative and qualitative changes in asylum credibility assessments were essential to the drop in recognition rates.
Our article shows the importance of mixed-method approaches in research on asylum decision-making. Unlike case law and statutes, which can alter legal outcomes in a single judgment or act, initial interpretive shifts in asylum decisionmaking, we argue, are made en masse, and their full extent may be observed by combining quantitative and qualitative methods. Quantitative approaches alone do not address exactly how decision-makers use collectivized discretion, but qualitative approaches alone cannot address the scale of interpretive shifts. We maintain that a mixed-method approach can offer key insights on large-scale, collective discretionary shifts in the application of asylum law and in initial-level decision-making more generally and that there is a need for a deeper and wider understanding of this type of "bordering work" (Gorman 2017).
As our analysis is limited to one country and one particular group of asylum applicants, we acknowledge the need for further research on recognition rates in the context of crisis. More research is warranted, for example, on possible differences between applicant groups from different origin countries and on the situation in other refugee-receiving countries. Different approaches and data, such as multivariate analysis or interviews with asylum applicants and asylum decision-makers, could help provide an even more nuanced picture of collectivized discretion.
Many scholars (e.g., Alpes and Spire 2014;Tucker et al. 2016;Dörrenbächer 2017) have argued that initial decision-makers can crucially influence policies' final outcomes, and our analysis reinforces, as well as develops, this notion. In particular, we look beyond individual asylum decision-makers and their motivations for approaching asylum credibility assessment in a particular way and consider discretion (also) as a collective phenomenon. Discretion in asylum decision-making is often seen and analyzed as a phenomenon pertaining to an individual decisionmaker's subjective response (e.g., Kagan 2003;Herlihy, Gleeson and Turner 2010;Gill et al. 2018). Our article enhances this work by arguing that asylum decisionmakers are also able to use discretion collectively as a tool for mass denial of asylum applicant populations. Simply put, asylum decision-makers can change the overall or collective outcomes of their decisions without meaningful changes in law or the applicant pool by, for instance, discrediting applicants' claims. This collectivized discretion, we argue, merits attention from migration scholars and decision makers. Gorman (2019, 490) points out that "the refugee definition is not static but rather a site of ongoing struggle over asylum protection, evolving in response to changing human rights norms and domestic priorities." We assert that concrete changes in Finland's international protection practices closely match the objectives and content of the Finnish government's 2015 asylum policy action plan and the series of crisis-based policies aiming to shrink the population of asylum-seekers. A perceived refugee crisis or "too many" asylum applicants from a particular origin country can potentially impact asylum recognition rates, as previous quantitative research has demonstrated (e.g., Rottman, Fariss and Poe 2009;Neumayer 2005). Even amid a perceived crisis, though, it is crucial that asylum status determinations be anchored in the individual assessment of each applicant's case, not in collectivized discretion or in efforts to ensure uniformity with political goals, which can lead to arbitrary results in the application of asylum law and potentially force those in need of refugee protection to face deportation.