DECISION-MAKING PROCESSES AND ASYLUM CLAIMS IN EUROPE : AN EMPIRICAL ANALYSIS OF REFUGEE CHARACTERISTICS AND ASYLUM APPLICATION OUTCOMES

Although refugees are offered the potential of asylum protection through the 1951 Convention Relating to the Status of Refugees, many states are attempting to curtail the number of asylum applications received each year. Several scholars have argued that even without such deterrence measures, some refugees are less likely to receive a positive asylum decision than others. Using country level data for 32 European states, this project examines the effects of various characteristics – including gender, age, and state of origin – on the likelihood of being granted asylum. This study is a fi rst attempt at empirically testing the asylum literature’s anecdotal evidence of bias in the decision-making processes in determining the outcomes of asylum claims.


PROCESY DECYZYJNE W SPRAWIE WNIOSKÓW O AZYL W EUROPIE: CHARAKTERYSTYKA UCHODŹCY ORAZ ANALIZA EMPIRYCZNA WYNIKÓW PRZYZNAWANIA AZYLU
Streszczenie: Chociaż uchodźcy otrzymują ofertę ochrony azylowej na podstawie Konwencji dotyczącej statusu uchodźców z 1951 roku (Konwencja genewska), wiele państw stara się ograniczyć liczbę wniosków o azyl otrzymywanych każdego roku.Niektórzy badacze sądzą, że nawet bez tych środków odstraszania, niektórzy uchodźcy mają mniejszą szansę na pozytywną decyzję o przyznaniu azylu niż inni.W niniejszej pracy, na podstawie danych z 32 państw europejskich, przeanalizowano wpływ różnych cech -w tym płci, wieku, pochodzenia i kraju pochodzenia -na prawdopodobieństwo przyznania azylu.To badanie protection is offered equally to all refugees.Specifically, many claim that political leaders in key destination states face competing national and regional interests.Governmental leaders struggle between their desire to offer protection to refugees seeking asylum and their serious political and social concerns about the consequences of having asylum seekers in their countries.More importantly, critics claim that a desire to satisfy both of these competing interests has created an asylum system in which not all asylum applicants have an equal chance of receiving asylum.However, while many variables have been suggested, there is little consensus on which factors of an asylum application affect asylum case decisions.Further, while many offer anecdotal evidence of disparities between asylum decisions, few studies have been done to empirically test these claims.
In this article, I take the first of many steps to tackle a large and overarching question that addresses these gaps in the literature.In particular, this article asks: what factors determine the outcomes of asylum claims filed in Europe?To answer this question, I examine variables identified in the literature as affecting recognition rates3 , including characteristics of the individual applicants, their states of origin, and the states in which their asylum applications were filed.Data was collected for 32 European states -the 28 members of the European Union as well as Iceland, Liechtenstein, Norway, andSwitzerland -between 2008 and2013.This analysis aims to simultaneously test the findings of earlier studies and to contribute to the limited set of studies that include individual applicant characteristics, specifically gender and age, as independent variables.
The estimation results of this analysis are mixed, but, like other empirical studies of asylum recognition rates in Europe, they provide cause for limited optimism against the charges of bias.The rates of full asylum -known as Geneva Convention statusand intermediate protection statuses as well as the rates of rejection in asylum cases are most consistently affected by the conditions of applicants' states of origin.The applicant characteristics have limited effect, with only certain age variables returning significant results.
The remainder of this article is organized as follows: the next section examines the key theories regarding the determinants of the decisions made in asylum cases identified by the literature.This section is followed by a brief discussion of the existing empirical studies of the determinants of asylum recognition rates.The third section outlines the research design of this project.In the fourth section, I present the results of the logistic regression analysis and a discussion of these results in light of my hypotheses.The article concludes with a discussion of the implications of the findings of this study and areas for future research.

Determinants of Asylum Decisions
Women and Gender-Based Discrimination One of the earliest and most extensive areas of research on the potential bias within asylum legal regimes examines the gender of asylum applicants, focusing specifically on female refugees.It should come as no surprise that research on women dominates much of the asylum and refugee literatures; women and children are strikingly overrepresented in refugee populations, accounting for as much as 80% of the world's refugees (Randall 2002).Despite this fact, the impact of an applicant's gender remains unclear.Some scholars note that women are seriously underrepresented in the population of successful asylum applicants and that being a woman contributes to the rejection of asylum claims (Randall 2002;Keith and Holmes 2009).Conversely, others argue that female refugees actually have an advantage over males in asylum cases (Bhabha 2004a).
The feminist and gender asylum literatures focus primarily on aspects of persecution that scholars argue are more common or particular to women, such as domestic abuse, female genital mutilation (FGM), and sexual violence (see e.g.Ankenbrand 2002;Anker 2000Anker -2001;;Anker, Gilbert, and Kelly 1997;Bahl 1997Bahl -1998;;Bloch, Galvin, and Harrell-Bond 2000;Crawley 2000;Gomez 2003Gomez -2004;;Heyman 2005;Heyman 2002Heyman -2003;;Musalo 2002Musalo -2003;;Randall 2002;Seith 1997;Sinha 2001).Two interrelated obstacles emerge for courts and administrators when considering asylum claims made as a result of these types of persecution.The first obstacle is that many of the instances of persecution considered to be of particular concern for women or that have historically been viewed as belonging to the 'domestic' sphere rather than the public sphere are also less clearly covered by international legal conventions (Anker 2000(Anker -2001;;Anker, Gilbert, and Kelly 1997;Randall 2002).Cases of refugees who suffered FGM, for example, often fall into this category.Second, the agents of persecution in many of these cases are non-state actors or are declared to be non-state actors during the adjudication of asylum claims due to what is often seen as aberrant rather than institutionalized or state-sanctioned conduct (Bahl 1997(Bahl -1998)).Victims of spousal abuse and rape by authority figures often struggle to prove the role of the state in presenting their cases for asylum.Within this framework, being a woman puts the individual at a disadvantage and results in fewer instances of full asylum protection.Human rights advocates have, therefore, been working to bring attention to the specific plights of women; although it remains unclear to what extent this advocacy has been successful in addressing potential sources of bias in asylum cases (Kelly 1993).It is possible that advocacy has increased the rate at which intermediate, rather than full asylum protection is granted.The perspective presented in this literature leads us to the first two hypotheses to be tested in this analysis: H1: As the number of female applicants increases, the total number of applications granted full asylum protection will decrease.H2: Higher rates of applications from females will correspond to increased rates of intermediate protection.
The majority of the literature discussed above emphasizes the negative impact of being female on an applicant's likelihood of gaining asylum.However, there are other issues to consider.First, there is a strong victimization narrative in the civil war literature that has traditionally viewed women and children as the primary victims in times of conflict (Alison 2007;Bhabha 2004a;Carpenter 2005).This overemphasis on women and children as the victims of conflict has two problematic consequences.First, the focus of this narrative on women and children has disempowered these groups and diminished their potential to be seen as autonomous political actors; this is especially true for adult women (Alison 2007;Carpenter 2005).This viewpoint supports the arguments outlined above in that it can make applications from women based on political persecution in civil conflict less compelling.Second, this narrative creates an environment in which adult male victims often do not get the protection they need; it is assumed that men can either take care of themselves if they are truly victims or that they are, in fact, active and willing participants of the conflict (Alison 2007;Carpenter 2005;Sivakumaran 2007;Zawati 2007).This would indicate, contrary to the feminist viewpoint, that adult women would be more likely to receive asylum than adult men because women are viewed as victims and men as potential perpetrators of violence.
H3: If women are viewed primarily as victims of conflict, then the rate of successful asylum claims will increase as the number of female applicants increases.

Age
Since children, along with women, are overrepresented among refugee populations, there is a growing literature related to the particular issues involved with refugee children and asylum applications made by minors (Bhabha 2004b;Taylor and Byford 2003;Stellinga-Boelen et al. 2007;Christie 2003;Dunkerley et al. 2005).At present, the bulk of this literature focuses on either the medical needs of refugee children once they arrive in the receiving state (Taylor and Byford 2003;Stellinga-Boelen et al. 2007) or the interactions of national social work agencies and individual social workers with children seeking asylum (Christie 2003;Dunkerley et al. 2005).However, research has started to emerge that focuses on the legal implications and obstacles for minor asylum applicants.In particular, scholars are parsing out the similarities and, more importantly, the differences between the experiences of women and child refugees, deconstructing assumptions that have collapsed them into a single category of vulnerable people (Bhabha 2004a).Scholars like Bhabha (2004a) argue that, much like female refugees, children -especially when unaccompanied -seem to have a more difficult time in successfully procuring asylum than adult applicants.However, unlike adult women, the primary source of difficulty for these young applicants comes directly from the structures and procedures of the legal system rather than ambiguity or bias toward their claims.In particular, because of their age, minor applicants have little to no agency within asylum courts; they must rely on the advocacy of others in their search for protection.
[Minor applicants] have difficulties getting adequate legal representation, their cases are more likely to be postponed and to drag on over time, and they have less chance of being granted refugee status.The outcome of an asylum application is often a troubling limbo of indeterminacy, rather than reassuring guarantee of permanent status.(Bhabha 2004b: 143).
Further, while the feminist and women's rights movements have systematically challenged the male-centered aspects of the asylum regime, there has been no comparable movement to challenge the adult-centered aspects of asylum law (Bhabha 2004a).Thus, there has been little movement made to improve the legal outlook of minor asylum seekers.
H4: As the number of applicants under the age of 18 increase, the rate of successful applications will decrease.
One final note about the age of asylum applicants should be made before moving forward.In examining the impact of age on asylum applications, it is striking that there is essentially no research that directly examines the impact of other points of the age spectrum.Some studies have included variables that may serve as partial proxies for later stages in life such as educational attainment or occupation.Keith and Holmes (2009), for example, found that marriage has a significant and negative effect on gaining asylum.There are studies within the migration literature that study the impact of age more fully, but the focus of many of these studies is on either the relationship between age and mobility (e.g.Sandefur and Scott 1981) or on the effect that one's age at the time of migration has on post-migration life (e.g.Angel et. al 1999;Angel and Angel 1992).Similarly, examinations of policies regarding the sustainability of the welfare state in the face of aging populations may denote a concern over admitting older asylum-seekers (Vincent 1996).However, while there is no clear indication of whether or how age might affect an applicant's chances of asylum, I posit two hypotheses based on the importance of economic factors, such as GDP, found in other studies (Neumayer 2005a) -discussed below -and the welfare state literature: H5: An increase in the number of applicants whose age -between 18 and 64 -could make them more to be able to support themselves economically will correspond to an increase in the rate of successful asylum applications.
H6: An increase in applicants over the age of 65, who are likely to rely increasingly on state services, will lead to a lower rate of asylum protection.

Destination States
In addition to applicant characteristics, aspects of the state in which the case is decided are also believed to impact rates of asylum.There are three specific areas of research within this area of the literature: politics, economics, and history of applications.The political characteristics of destination states have been a strong area of research in terms of asylum and immigration policy (Boswell 2007;Fekete 2005); however, to my knowledge political variables have not had much success in explaining asylum recognition rates (Boswell 2007;Neumayer 2005a).Conversely, both economic and historical variables have shown some promise as determinants of asylum claims.
When assessing the impact of the economic conditions of a destination state, the state's gross domestic product (GDP) and unemployment are often used within the asylum literature.Neumayer (2005a), for example, found that higher levels of GDP led to higher recognition rates in European states -perhaps because they can better afford to take on the financial cost of asylum seekers -while the percent of unemployed in the state had no significant effect.However, increasing concerns over the sustainability of welfare states (Vincent 1996) as well as rising unemployment in the wake of the Great Recession (Bell and Blachflower 2011;Cervany and van Ours 2013;Hatton 2012) may have changed or increased the influence of economic factors on asylum rates in recent years.Higher levels of GDP may still contribute to higher recognition rates, but high rates of unemployment may temper them, leading to a reliance on intermediate or temporary forms of protection, or even cause the overall rates of protection to decrease.H7: As levels of a receiving state's GDP increase, the rate at which the state grants asylum will also increase.H8: High levels of unemployment in receiving states will correspond to a decrease in the level of asylum protection granted, but an increase in the rate of intermediate protection.
In addition to economics, the asylum recognition rates of destination states may be affected by their history of received applications.Specifically, it is argued that higher numbers of applications received by a destination state in previous years lead to lower rates of recognition moving forward.There may be a certain degree of 'crisis fatigue' affecting the willingness of destination states to continue to accept refugees and asylum seekers (Holzer and Schneider 2001).Both Neumayer (2005a) and Holzer and Schneider (2001) tested this theory and found that higher total numbers of applications corresponds to lower recognition rates.Both studies used data from before 2000; therefore, it is necessary to re-examine the effect of application numbers using more recent data to see if the conclusions hold.
H9: As the number of applications received in recent years increases, the rate at which asylum claims are successful will decrease.

States of Origin
Another aspect of asylum applications that might be impacting applicants' chances of gaining asylum is from what state or part of the world the applicant originates.Much of the asylum literature has dealt with this feature in a different manner than it does other aspects of asylum cases, focusing particularly on the volume and direction of the migration of refugees from particular states of origin to states of destination (Bocker and Havinga 1998;Castles and Loughma 2003;Hatton 2004;Neumayaer 2005b).Another group of scholars has tried to test the impact of specific states of origin or characteristics of states of origin on the outcomes of asylum claims.In this analysis, I aim to contribute to the latter of these literatures.
There are two groups of characteristics related to an asylum applicant's state of origin.The first set of characteristics is associated with an applicant's specific state of citizenship.One of the most challenging areas of this literature deals with stateless refugees or refugees who lack citizenship.Stateless refugees form a particularly vulnerable population and often pose a challenge to states tasked with assessing their asylum claims.Stateless persons have no official nationality and are not considered a citizen by any state.While on the one hand this reality means these individuals are not under the legal protection of any state, it also makes claims based on state persecution difficult to prove.Further, the definition of statelessness is a legal one and does not take into account the nuances of lived realities; for example, the degree to which citizenship is granted can vary, making determining the point at which statelessness occurs difficult (Batchelor 1995;Settlage 1997Settlage -1998)).Indeed, as some have argued, the conditions in a person's original state of citizenship might be so dire or the persecution of their group so great, as in the case of the German Jews during the Third Reich, that the person chooses statelessness in order to escape it (Batchelor 1995).While international agreements have sought to prevent statelessness (UNHCR 2014a), they are less clear on how to protect stateless individuals who seek refugee status and asylum.Since the burden of proof in asylum cases falls most heavily on the applicants, we might expect that higher numbers of stateless applicants to lead to lower rates of asylum recognition; conversely, since there appears to be consensus that stateless individuals are at risk, the rates of intermediate protection might be higher.H10: As the rate of applications from stateless individuals increases, the rate of asylum protection will decrease, but the rate of intermediate protections granted will increase.
The second group of characteristics related to applicants' states of origin deal with the conditions -or the perception of conditions -in the state of origin.Several studies have used common measures of political freedom and state violence to try to explain the rates of successful asylum claims.In a way, measures such as the Freedom House scores of states' level of political rights and civil liberties (Freedom House 2014) and the Political Terror Scale, which measures violations of physical integrity in each state (Gibney, Cornett, Wood, and Haschke 2014), serve as proxies for the merit of individual asylum claims.This is especially true when using aggregate data where individual measures of merit are not available; these scores can mimic the effect of merit because they measure the potential for persecution in a state.Asylum applicants who are citizens of states with poor records on these scales would be expected to be more likely to receive asylum since they are less likely to receive protection in their state or origin.
H11: Low levels of political rights and civil liberties in a state of origin -as measured by the Freedom House -will increase the rate at which asylum is granted.H12: As the likelihood of violations of physical integrity -denoted by a high score on the Political Terror Scale -increases, the rate at which asylum protection is granted will also increase.
Other measures of the characteristics of a state of origin are meant to gauge the perceptions of states and the conditions in those states by refugee destination states (Akram 2000;Keith and Holmes 2009).For example, while scholarly research has shown that there is no clear relationship between becoming a party to an international human rights treaty and greater respect for human rights (Goodman and Jinks 2003;Hathaway 2007;Neumayer 2005c), it is possible that the decision whether or not to sign a treaty still sends a signal to other states about a government's intent to comply with the treaty's terms (Martin 2005).Not signing the Refugee Convention, therefore, may signal to other states a lack of intent to protect individuals who would qualify as DECYZJE NR 23/2015 DOI: 10.7206/DEC.1733-0092.45refugees outside of their state of origin and, therefore, could lead to higher rates of asylum for individuals from these non-signatory states.
H13: An increase in the applications received from citizens of states that are not signatories to the Refugee Convention will correspond to an increase in the rate of successful asylum applications.
Similarly, Akram (2000) argues that Western perspectives of Islamic governments, specifically the prevalence of neo-Orientalist views among both Western states and human rights advocates, has proved detrimental to applicants from these states.The stereotyping of Islamic governments as 'warmongering' or as 'terrorists' has only served to "[support and promote] the most repressive and extreme versions of Islamic interpretation currently being manipulated by fundamentalist regressive regimes… further [strengthening and entrenching] such regimes' efforts to distance 'Islam' from universal human rights" (Akram 2000: 9).However, rather than contributing to the success of claims made by applicants from Muslim states, these simplistic characterizations have made it much more difficult for these refugees to seek asylum.Extreme or simplistic characterizations of Islamic governments often obscures the types of persecution that are actually occurring and, more importantly, such grandiose claims about the nature of such states can easily be disproved through the governmental research and expert testimony that are customary in the adjudication of asylum cases (Akram 2000;Polk 1997Polk -1998)).In both instances, the case for asylum becomes more difficult for applicants to prove and may lead to lower instances of successful asylum claims.
H14: As the rate of applications from refugees from Muslim states increases, the rate at which asylum protection is granted will decrease.

Existing Empirical Studies
To my knowledge, there are only four existing empirical studies that directly examine the specific determinants of asylum recognition rates: Holzer and Schneider (2001) 4 , Holzer, Schneider, and Widmer (2000), Keith andHolmes (2009), andNeumayer (2005a).The focus of most of these studies is on Europe before 2000 and they include variables from either the destination state alone (Holzer and Schneider 2001;Holzer, Schneider and Widmer 2000) or on the destination and origin states (Neumayer 2005a).Only Keith and Holmes (2009) use data that extends into the new millennium; it is also the only study to include variables that measure the impact of characteristics of individual applicants.The present analysis builds on these previous 4 The article written by Holzer and Schneider (2001) is only available in German; therefore, my present knowledge of its contents comes directly from Neumayer's (2005a)  studies in two important ways.First, it builds on the available data temporally by using data for the years 2008 through 2013.Second, this study includes variables measuring the impact of characteristics of the destination state, of the applicant's state of origin, as well as qualities of the applicants.The examination of variables from each of these categories by the above authors has produced significant results; however, none of these four previous studies includes variables from all three categories into a single analysis.

Research Design
For this article, I used a data set compiled, primarily, from the European Union's Eurostat database (European Commission 2014).The observations in this dataset are country years for 32 European states -the 28 member states of the EU as well as Iceland, Liechtenstein, Norway, and Switzerland -for the years 2008 through 2013.In this section, I will outline each of the variables included in the models presented in the Results and Discussion section.Table 1 provides descriptive statistics for each of the variables discussed in this section.

Dependent Variables
In analyzing the determinants of asylum cases, the dependent variable is the outcome of an asylum case and, in general, there are three categories of potential outcomes: full asylum or Geneva Convention status, an intermediate decision -which includes statuses that provide some form of protection to the applicant, but fall short of full asylum status -and rejection.Most of the literature is focused on determining the variables that characterize claims that receive Geneva Convention status.The variable for Geneva Convention decisions is a log transformation of the proportions of decisions in which full asylum was granted each year.In addition to asylum, states have developed other protection statuses to address the needs of vulnerable populations that do not meet the requirements for asylum.Therefore, I created a variable for intermediate decisions in order to assess the similarities and difference between the determinants of these two groups of decision types.It is possible that if there is bias in the rate of full asylum decisions due to certain characteristics that it is addressed by destination states through the use of the intermediate statuses.The final dependent variable includes cases in which the asylum claim was rejected.I use this depended variable to assess whether the determinants of these cases are any different than those of positive decision types.None of the literature I have found to date directly addresses the determinants of rejection in asylum cases.Implicitly, we would expect the determinants of rejection to be the inverse of those of successful applications.However, since it does not appear to have been analyzed directly, it is important to test this assumption empirically.I used the log transformation of the proportion of each of the outcomes in order to address violations of the normality and heteroskedasticity assumptions of regression.
Before discussing the independent variables, an important observation must be made about these dependent variables.As Neumayer notes, "the theoretically correct recognition rate is the percentage of asylum claims recognized relative to the number of asylum claims lodged" (2005a: 51).However, this is rarely what our data actually captures.As noted above, the data from the Eurostat database is organized by year.Our analysis is hindered by the reality that not all asylum claims are adjudicated in the same year in which they were filed as well as the fact that data on the submission date of applications is generally unavailable.Thus, this analysis follows Neumayer in his adoption of the UNHCR's approach to assess the number of successful applications in a year compared to the total number of cases decided in that same year."In other words, [the] recognition rate does not measure the successful rate of applications but the rate of successful decisions" (Neumayer 2005a: 51).This is obviously an imperfect measure for the determinants of asylum.However, despite this limitation, the analysis should at least provide a general idea of what factors might impact asylum recognition rates.

Independent Variables
As discussed above, this analysis focuses on three primary categories of independent variables drawn from the asylum literature as possible factors determining asylum cases: characteristics of the applicant, of the destination state, and the state of origin.

Applicant Characteristics
There are two groups of variables that measure the effect of the characteristics of applicants on the outcomes of asylum claims.The first is gender or, as measured by the Eurostat database, sex.The database provides information on three categories of sex: female, male, and unknown.The literature is clearest on the theoretical impact of being female; therefore, the proportion of cases out of the total in which the applicant was female was included in the models presented below.Based on the literature discussed above, there are three potential hypotheses for the effect of being female.The first two posit that, due to the fact that the types of persecution experienced by women are not clearly covered by international conventions, we could expect a higher proportion of claims made by female applicants to lead to lower rates of full asylum decisions but to higher rates of intermediate decisions as a result of advocacy on behalf of female refugees.Conversely, the third hypothesis states that the belief that women are primarily victims during conflict will mean that higher rates of female asylum applicants will correspond to higher rates of asylum recognition.In addition to the variable for female applicants, I have included a variable for the proportion of cases in which the sex of the applicant was unknown.Since it is unclear exactly what the 'unknown' sex category contains, it would not be appropriate to use it as a reference category; the proportion of cases in which the applicant was male, therefore, fulfills this role and does not appear in any of the models presented in Table 2.
The second group of applicant characteristic variables measures the impact of various age categories.There are six age categories in this data set: applicants less than 14 years of age, from 14 to 17 years, from 18 to 34 years, from 35 to 64 years, 65 years of age and older, and cases in which the age of the applicant was unknown.The asylum literature only directly provides us with a hypothesis for minor asylum applicants: since the legal system presents unique obstacles for child refugees, higher proportions of minor applicants will lead to lower rates of asylum or other protection statuses.The literature is less clear about the effect of the other age categories.However, as noted above, two further hypotheses can be tested based on previous studies of the effects of economic variables on asylum outcomes as well as the welfare state literature.First, a higher rate of applications from adults of an age where they could feasibly join the work force and support themselves would lead to higher rates of asylum protection.Second, the proportion of older applicants -who may no longer be able to join the work force and would potentially add to the burdens of aging populations in various destination states -would lead to lower rates of asylum protection.As the proportion of applicants aged 65 or older rises, therefore, we would expect to see the rate of positive case decisions to decline if this perspective is true.Finally, the variable for the proportion of applicants between 35 and 64 is left out of the models and acts as the reference category.

Destination State Characteristics
As discussed above, the effects of the state in which an asylum claim is decided are often measured through economic and historical variables.For this analysis, I incorporated two economic variables and one historical variable.The first economic variable is a measure of the state's GDP in thousands of current US dollars per capita.The second economic variable measures the proportion of the destination state's total labor force that is unemployed in each year of the data set.The data for both of these economic variables came from the World Bank's database (World Bank 2014a;World Bank 2014b).According to the available literature, higher levels of GDP per capita should lead to higher recognition rates while we expect higher unemployment to lead to lower rates of asylum protection and higher rates of intermediate, or temporary, forms of protection.
To address the impact historical factors, namely the number of applications received by a state in the past, I use the average number of applications, in thousands, received by each European state in the data set over the previous 1 to 5 years. 5Taking an average of the previous applications is in line with earlier studies of this type (Neumayer 2005a) and is done for two reasons.First, it takes into account shortterm changes in the total number of applications received by any state.Second, it addresses, in part, the endogeneity problem created by the fact that not all asylum claims are adjudicated in the same year in which they are filed.Based on earlier studies, I expect that as average number of applications increases, the rate of asylum recognition will decrease.

Origin State Characteristics
The remaining variables listed in Table 1 address the impact of applicants' states of origin.The first of these variables addresses the issue stateless refugees and uses data from the Eurostat database.As the literature describes, stateless applicants face 5 The number of years that I was able to include in the average for this variable was determined by the availability of data.There were several years with missing data, which limited the number of years that could be included in the average for each state.challenges in proving asylum claims that are not necessarily faced by other refugees.For this reason, it is expected that as the proportion of applicants who are stateless increases the rate of full asylum protection will decrease.However, since stateless persons are also considered a vulnerable population, the increase in such applicants should correspond to an increase in the rate of intermediate decisions.
In order to address the neo-Orientalism that Akram (2000) argues can affect asylum outcomes, I have included a variable for the proportion of applicants from Muslim majority states.I define a Muslim majority state as one in which at least 50 percent of the population identifies as Muslim and no other single religious community accounts for 20 percent or more of the population.I determined the religious breakdown of the various states of origin using the Correlates of War's World Religion data set (Correlates of War 2014).If neo-Orientalism is affecting asylum recognition rates in the way that Akram contends, then as the proportion of applicants from Muslim states increases, the rate of rejection should increase and the rate of decisions granting full asylum protection should decrease.
The final four variables address the conditions in applicants' states of origin.The first variable is the measure of the proportion of applicants that come from states that are not signatories to the 1951 Geneva Convention.If destination states view not signing the convention as a signal that a state has less intention to protect refugees then as the proportion of applicants from these states increases, the rate of intermediate and Geneva Convention decisions should increase as well.The remaining variables are based on the scores assigned to states by Freedom House (Freedom House 2014) and the Political Terror Scale (Gibney, Cornett, Wood, and Haschkey 2014).Freedom House scores are based on the levels of political and civil liberties within a state and are grouped into three levels of freedom: free, partially free, and not free.The models below include variables for the percentages of applicants from free and not free states with the percentage of applicants from partially free states serving as the reference category.Applicants from free states would be less likely to be able to meet the burden of proof in asylum case than those from not free states.Therefore, as the proportion of applicants from free states increases, the rate of positive asylum decisions should decrease while the rate of rejections should increase.The inverse should be true for applicants from states deemed not free.The Political Terror Scale measures instances of state-sanctioned violence and assigns a score to each state on a 1 to 5 scale where 1 indicates states in which such violence is rare and a 5 indicates a state in which state-sanction violence is wide-spread and common.The Political Terror Scale data set only provides scores for the years 2008 through 2012 and for each of these years the average terror score for all of the European states included in the data set was less than 3. Therefore, I included a variable for the proportion of cases in which the DECYZJE NR 23/2015 DOI: 10.7206/DEC.1733-0092.45applicant originated from a state that scored a 3 or higher on the terror scale; in other words, these are the applicants from states considered to have worse than the average levels of state-sanctioned violence.Much like the Freedom House classifications, I expect that as the proportion of applicants from states with terror scores of 3 or higher increases, so too should the rate of positive asylum decisions.

Statistical Methodology
In order to test the hypotheses outlined in the previous sections, I used logistic regression analysis to assess the effect of each of the independent variables on the log transformed dependent variables, each of which represents one of the three types of outcomes in asylum cases: Geneva Convention status, intermediate protection, and rejection.In each of these regression models, I also included fixed effects for the country and year in which the case was decided in order to control for different effects across these two variables.Similarly, I also clustered the observations by the country in which the state was decided; the standard errors reported in the remainder of this article, therefore, are robust standard errors.

Results and Discussion
Table 2 presents the results of each of the three models tested as part of this analysis.The first model uses the intermediate decisions as the dependent variable, the second the percent of Geneva Convention or full asylum decisions, and the final model uses rejection decisions.The remainder of this section analyzes the results of each group of variables -the characteristics of the applicant, the destination state, and the state of origin -as well as the overall performance of the models.
The most interesting point to note about the variables measuring qualities of applicants is that neither being a female applicant or a minor asylum-seeker have statistically significant effects on any of the decision outcomes.The coefficient for female applicants is positive in the Geneva Convention model and negative in both the intermediate and rejection models.This means that neither H1 nor H2the hypotheses arguing that being a female negatively impacts the rate of asylum protection and that advocacy to address this issue might have increased the likelihood of intermediate protection -are supported.Indeed, the opposite appears to be true.This could indicate one of two things.Either the bias suggested in the literature did not exist or that it did exist and advocacy efforts over the past thirty years -or even some other unknown factor -have made a difference for female applicants.The third hypothesis representing the victimization narrative for women has better support in these models since the results are positive in the Geneva Convention model and negative in the rejection model.However, since none of these results are statistically significant, the hypothesis will need further testing.The results for the age categories included in these models are somewhat more complicated.The influence of the proportion of minor applicants differs between the two age categories for minor applicants.The variable for applicants under the DECYZJE NR 23/2015 DOI: 10.7206/DEC.1733-0092.45age of 14 is negative for both Geneva Convention and rejection decisions, but is positive for intermediate decisions.If the literature is correct and the legal system creates barriers for the youngest of applicants in gaining full asylum protection, it does not appear to prevent them from receiving some form of protection from the destination states.By contrast, applicants between the ages of 14 and 17 do not seem to receive the same level of consideration -the proportion of applicants in this age group have negative effects on the rates of intermediate and Geneva Convention decisions and a positive impact on the rate of rejections.The fact that the coefficients for these two variables are both negative in the Geneva Convention model provide tentative support for H4, which argued a higher proportion of minor applicants would correspond to a lower rate of successful asylum claims, but future research will need to see if these results hold.
Interestingly, although the asylum literature does not supply direct information on other age categories, the hypotheses posited above for adult applicants lead to significant results in this model.The fifth hypothesis, which holds that increases in applications from adults of working age will increase the rate of asylum is not supported by these results.Instead, the coefficient for this variable in the Geneva Convention model, which is statistically significant at the .05level, is negative.This result might indicate that the concern in receiving states might be that applicants between 18 and 34 years of age are seeking admittance to their countries for the purpose of work.This would primarily make them economic migrants rather than refugees and not entitle them to asylum protection.In contrast, the results for the proportion of applicants 65 years of age and older -with a negative coefficient in both the Intermediate and Geneva Convention models and statistically significant coefficient at the .01level in the Geneva Convention model-do support the sixth hypothesis and the notion that fears of overtaxing the welfare state could lead receiving states to offer protection to older applicants at lower rates.
The variables for the destination states perform differently in this analysis than they do in earlier studies (Holzer and Schneider 2001;Neumayer 2005a).Neither the average number of applications received by the destination state nor its GDP per capita have statistically significant effects on any of the asylum outcome types.However, the signs on the coefficients for GDP per capita are in line with previous research as well as with H7: higher levels of GDP per capita are related to higher rates of protection granted to refugees.Conversely, the average number of applications produces coefficients with signs in the opposite direction to that found in earlier studies and expected based on the ninth hypothesis.
The eighth hypothesis posits that higher rates of unemployment will negatively affect the rate of Geneva Convention decisions -which is supported by the sign of the coefficient for unemployment in the Geneva Convention model though the result is not statistically significant.Unemployment is statistically significant in the model for intermediate decisions.However, in that model, the coefficient is positive.This may indicate that high unemployment makes receiving states less likely to offer full asylum protection -which can lead to the applicant's permanent residence in the state -and that they rely instead on the temporary forms of intermediate protection to offer assistance to those who need it without potentially contributing to their unemployment in the long term.
The final group of variables -those for applicants' states of origin -performs the best in these models.The variables for the Freedom House scores, in particular, as well as the Political Terror Scale variable perform well and as expected according to the hypotheses outlined above for Geneva Convention decisions: applicants from free states decrease the rate of successful claims while applicants from not free states and states with high scores on the Political Terror Scale increase it.The proportion of applicants from free and not free states also have statistically significant effects on the rate of rejections, again in the direction we would expect if these variables serve as proxies for the merit of a claim.
Statelessness has a strong significant and positive effect on the rate of intermediate decision.Further, it is noteworthy that, although the coefficients are not significant, the relationship between statelessness and the rate of Geneva Conventions is positive.These results are mixed in light of the hypotheses offered earlier and the available literature.The literature would lead us to expect that despite being acknowledged as an especially vulnerable population, stateless refugees face greater obstacles in receiving asylum protection but could receive intermediate protection at higher rates to compensate for the obstacles they face.These results indicate that the relationship between statelessness and recognition rates could be more complicated.By contrast, the theory regarding the effect of neo-Orientalist perspectives on asylum outcomes does receive some support in these results.As suggested by H14, the effect of the proportion of applicants from Muslim majority states has a statistically significant and negative impact on the rate of Geneva Convention decisions.However, although not significant, the effect of this variable is positive on intermediate decisions.Therefore, this may be an instance where European states utilize intermediate forms of protection to assist refugees without being responsible for them indefinitely.The impact of not signing the Refugee Convention also has significant impact on the rate of full asylum decisions; however, the direction of the relationship is not in line with the theory discussed above, having a negative impact on the rate of Geneva Convention decisions.In this case, it is possible that aggregation of the data may be hiding the nuances of the relationship between the independent and dependent variables.

Concluding Observations
The results of this study, overall, show low levels of bias in the decision-making processes within asylum legal regimes.The key applicant variables that would indicate a bias in asylum outcomes according to the literature -those for female and minor applicants -were not statistically significant in these models.Further, the variables for the states of origin, which are measures of the condition from which an applicant has fled and is seeking protection, have the greatest explanatory power for the rates of asylum outcomes analyzed above.Some of the findings, however, do indicate the existence of certain types of bias affecting decisions made in asylum cases.Specifically, the negative impacts that the proportions of applicants between the ages of 18 and 34 and 65 years of age and older have on the rate of cases in which full asylum protection was granted is of concern, especially since the impact of age as a potential determinant of asylum claims has not been fully addressed within the asylum literature.
The incorporation of variables for the characteristics of the applicant, the destination state, and the state of origin seems to have provided important information regarding the determinants of asylum outcomes and indicates the need for future research that continues to examine the impact of all three of these aspects.As noted in the previous section, the characteristics of the state of origin appear to have the clearest relationship with asylum claim outcomes.Future research should attempt to disaggregate direct measures of a claim's merit from conditions of the origin state.Keith and Holmes (2009) included both types of variables in their analysis of cases filed in the United States and reported significant results.While, applicant characteristics are often difficult to include in the analysis of the determinants of asylum claims due to the rarity of individual-level data, the results in Keith and Holmes (2009) as well as the analysis presented here should make clear that individual characteristics do have an effect on whether or what type of protection an applicant receives.It is important for asylum scholars to continue to investigate the relationship between attributes of applicants, especially on an individual level, and the outcome of their asylum claims.
Conversely, the influence of characteristics of the destination state is less clear.It is possible that the effect of such variables changes over time or location.In the case of Europe, the variety of state-level asylum systems -despite the European Union's attempts at standardization through policies such as the Dublin Regulation -could be creating a selection bias problem in each state.Further, increasingly anti-immigration policies and the EU's stated intention to strengthen 'fortress Europe' could further affect the types of applications seen by individual states as well as their response to individual asylum claims.It will be important, therefore, to attempt to gain a clearer understanding of the relationship between receiving state characteristics and asylum claim outcomes.
The model used to explain the rate of Geneva Convention decisions produced the strongest results out of the three models presented in Table 2.However, the results also provide a foundation for future research on the other categories of case outcomes.The rate of rejection was most sensitive to the state of origin variables, which makes sense in light of the literature discussed above.However, the variables for applicants of unknown sex and unknown age also have strong statistically significant effects on the rate of cases that are rejected; it is important to find out more about what these variables could be measuring and how they impact the rates of rejection.The results for intermediate decisions are also intriguing.In many instances, variables that decrease the rate of Geneva Convention decisions -such as those for applicants under the age of 14 or those between 18 and 34 years of age -have a positive effect on the rate of cases in which an intermediate protection status was granted.This may support the case that some bias exists in the asylum legal regime as well as point to strategies destination states have employed to provide some protection while stopping short of granting full asylum to applicants.
Overall, the results of this analysis are decidedly mixed, as are the results of most studies within this literature.While there are certainly reasons for optimism about the results presented here, there is not enough evidence to prove or disprove the charges of bias within the asylum legal regime in Europe.As some scholars have already noted (Neumayer 2005a), even the perception of bias impacts how, where, when, and whether refugees file asylum claims.Providing more definitive information on whether any bias truly exists, therefore, will prove vital not only to refugees, but their legal representatives, members of the courts, and policymakers as well.There is clearly a great deal of ground yet to be covered in our attempts to uncover the factors that go into the decision-making process regarding asylum claims and it is to this uncovered ground that future research must turn.
assessment of it.

Table 2 Results for the Rate of Each Decision Type
In this table, * denotes statistical signifi cance at the .1 level, ** at the .05level, and *** at the .01level.The values provided in parentheses are the coeffi cient's standard error.