The Asylum Lottery: Recognition Rates Vary Strongly within Germany
by Gerald Schneider, Professor of International Politics at University of Konstanz, editor of European Union Politics, co-editor of International Interactions and President of the European Political Science Association (2013-2015) & Lisa Riedel, Master’s degree student at University of Konstanz.
The German asylum law should, as any responsibility of the federal government, be implemented uniformly across the country. A recently published study shows that this is not the case in a comparison of the 16 German Länder. Recognition rates differ so strongly that a different treatment of asylum seekers with a comparable background and similar reasons to flee is highly likely.
One of the central benchmarks of a democracy is equality. It cannot play its role where citizens must count on their luck. The requirement for fairness also applies to the weakest members of a society, in particular refugees who seek safety and shelter.
Asylum seekers should not experience differences when Germany decides on their applications. It is exclusively a responsibility of the Federal Office for Migration and Refugees (Bundesamt für Migration und Flüchtlinge, BAMF), which determines the fate of the asylum seekers in first instance. This office makes the initial decision about asylum applications. One notices considerable differences if one extrapolates the decision to the länd where they are made.
While Saarland and Bremen had with 69% and 55,7 % respectively, the highest recognition rates over the period from 2010 to 2015, Berlin and Saxony could be found at the bottom of the ranking with 24,6% and 26,9%. As the following figure for four representative lander shows, these differences persist over this period despite a general increase in the rates.
These differences are equally drastic for the most important groups of asylum seekers. While 75,5% of Iraqi asylum seekers are recognised in Lower Saxony, only 37,5% are accepted in Saxony-Anshalt. Afghani asylum seekers had the highest chance of recognition in North Rhine-Westphalia with 34,4%, while only 10% of the same group are accepted in Brandenburg.
The figure below shows that the variation coefficient increases for the rejection ratios and does not change much in Germany over the six years under examination with regards to the recognition ratios. This means that asylum policy making did not converge in light of the “refugee crisis”. Furthermore, the level of divergence is not that much higher among EU Member States than the German lander as one could expect.
The differences in Germany continue to exist in relation with factors like the population size that play a role in the official allocation key, the so-called Königstein formula. Statistically, it makes no difference if regional governments are led by the Christian Democrats (CDU or, in case of Bavaria, CSU) or the Social Democrats (SPD). The proportion of foreigners who live in a land exerts, however, a systematic influence – the larger the share of people who do not possess a German passport, the higher is the chance that decisions on an asylum request is positive. We find it troublesome that the number of xenophobic attacks in a land lowers the recognition rate in the subsequent year. Note that we only report correlations, and not causal claims in our analysis. The quality of the data received upon a written request is simply too limited. We would need much longer time series or ideally information on the decision-making behavior of the case officers to make valid judgements. These results demonstrate that the decisions by the BAMF case officers are not only influenced by the credibility of individual requests. They rather also let the preferences and moods that prevail in the land guide their decisions.
We cannot say with our analysis of highly aggregated data what are the reasons which ultimately shape the behavior of individual decision makers. Switzerland and, of course, the EU exhibit similar patterns. The differences that asylum seekers have to endure in federal states and a Union like the EU should not come as a surprise in the light of the key work of a U.S. political scientist William Riker on federalism. He stated already in the 1960s that “(…) if in the United States one disapproves of racism, one should disapprove of federalism”.
We can explain such differences with the help of the so-called principal-agent models which play an important role in modern social sciences, especially economics and political science. These models suggest in the political realm that governments frequently delegate tasks to agents who do not completely share the ideological goals of the principal or who do not invest sufficiently in the completion of a task. Principals can, however, avoid the implementation of non-representative policy choices through a careful selection of the agents or increased control of the agent’s actions.
To estimate the risk of a negative decision, just assume that the BAMF decides like a good statistical model and recognizes 95% of all requests correctly and that a recognition ratio of 30% means that 30% of the refugee population deserve protection and 70% not. The risk of an incorrect decision resorts to Bayes´ rules. This conditional probability amounts to 0.05*0.3/(0.05*0.3+0.95*0.7)=0.02. The risk of a false rejection of 2% increases with an error probability of 10% to 5% and 20% to 10%. If we consider the large number of decisions to be made, we can easily anticipate that several thousands asylum seekers were wrongly rejected and a similar number wrongly admitted.
The BAMF reacted to the media attention that our article received by insisting that all decisions are made at the individual level. We never doubted that each request receives the attention it deserves. However, the recognition rates differ globally and also in the comparison of individual groups so much that the chance of recognition of similar requests also depends on where the decision is made. Moreover, our statistical analysis only considered groups of asylum seekers for which a sufficiently large number of requests had been made each year in the 16 lander. The large number of cases renders it, in our view, impossible that the differences are random. The German länder would furthermore oppose an allocation key other than the Königstein formula in which some of them receive a disproportionate number of troublesome applications.
We therefore conclude that the German system of decentralised decision making on asylum requests has in all likelihood a considerable discriminatory potential. This asylum lottery can only be prevented if the monitoring of the decision-makers in the BAMF is intensified and if German asylum decision-making becomes more transparent through the provision of more and better data. The frequently uttered demand that asylum seekers should demonstrate a willingness to integrate into the German society should be preceded by fair decision making of the responsible German institutions.