(PDF) The EU's 'return rate' with third countries: Why EU readmission agreements do not make much difference

| | Received: 7 July 2020    Revised: 18 March 2021    Accepted: 25 June 2021 DOI: 10.1111/imig.12901 ORIGINAL ARTICLE The EU's ‘return rate’ with third countries: Why EU readmission agreements do not make much difference Philipp Stutz  | Florian Trauner Institute for European Studies, Vrije Universiteit Brussel, Brussels, Belgium Abstract The signing of EU readmission agreements is commonly be- Correspondence Philipp Stutz, Institute for European Studies, lieved to lead to a higher rate of returns of migrants found Vrije Universiteit Brussel, Brussels, Belgium. ineligible to stay in Europe. By analysing the EU’s ‘return Email:

rate’ with the whole world over a period of 11 years (2008–­ 18), this article argues that (formal or informal) EU read- mission arrangements may have less impact than widely assumed. They often lead only to temporary increases in the return rates of third countries –­if at all. Their relevance may have been overstated, even in Eastern Europe and the Western Balkans, when looking at the return rates of neighbouring states without agreement. The return rates of third countries tend to follow regional dynamics. Most African states have converged at lower levels, irrespective of readmission cooperation with the EU. These regional trends imply a widening gap between regions with high and low EU return rates. KEYWORDS EU migration policy, EU readmission agreements, EU return policy, EU return rate, irregular migration This is an open access article under the terms of the Creat​ive Commo​ns Attri​butio​n-­NonCo​mmerc​ial-­NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-­commercial and no modifications or adaptations are made. © 2021 The Authors. International Migration published by John Wiley & Sons Ltd on behalf of International Organization for Migration International Migration. 2021;00:1–19.  wileyonlinelibrary.com/journal/imig  |  1 | 2      STUTZ and TRAUNER I NTRO D U C TI O N ‘People who have no right to stay in Europe must be returned to their countries of origin. When only 36% of irregular migrants are returned, it is clear we need to significantly step up our work. This is the only way Europe will be able to show solidarity with refugees in real need of protection’, (former) European Commission President Jean-­Claude Juncker (2017). How to measure ‘effectiveness’ in the field of European Union (EU) return policy? A policymaker is likely to re- spond by referring to the ‘return rate’. The return rate is the ratio between the number of persons ordered to leave and those who actually returned to a third country after an order to leave. The ‘EU return rate’ (as referred to by Jean-­Claude Juncker in the statement above) is the share of all third country nationals who left the EU’s territory after an order to leave. In 2018, the EU return rate was 35.6 per cent. The return rate has become a ubiquitous reference point in political and academic debates on irregular migra- tion in Europe. At national level, this rate has been used by politicians, most vocally by those from the far-­right, to accuse an incumbent government of ‘failing’ or being too ‘lenient’ in the migration field. A case in point is the Alternative for Germany, which has campaigned for a ‘minimum deportation rate’ (‘Mindestabschiebequote’) to highlight its zero-­tolerance approach to irregular migration (e.g. Der Tagesspiegel, 2017). At EU level, the return rate has been used for making policy decisions. A return rate perceived as too low was brought into play as an argument to justify the recast of the EU Return Directive in 2018, which sets out to harmonise return procedures and standards in the EU (European Commission, 2018b: 2). The approval of visa applications may also get more difficult for citizens of countries whose governments refrain from cooperating on readmission, with the return rate acting as one indicator of the assessment (Article 25a, EU Regulation 2019/1155 of 20 June 2019). A low percentage of effectuated returns was seen as a reason to start negotiations on a readmission agreement with particular third countries (European Commission, 2015: 14). With those states that have a readmission agreement with the EU, the dialogues in the Joint Readmission Committees should be consistently used ‘to further improve the rate of approved readmission requests and of effective returns’ (European Commission, 2011: 5). Yet, to what extent does the conclusion or the presence of a readmission agreement actually lead to a higher EU return rate? This article aims at understanding how the EU’s (formal and informal) readmission cooperation impacts the EU’s return rate. Policymakers often assume that a closer cooperation on readmission will lead to higher return rates. This article critically reflects upon and discusses this assumption. To that end, we have established a data set analysing the EU’s return rate over a period of 11 years (2008–­18) with all third countries (implying the whole world). We compare the data across different regions and countries and over time. To the best of our knowledge, no one has yet developed such a worldwide and longitudinal view. It is important to highlight that we do not aim at explaining why the EU has specific patterns of return coop- eration with third countries. We do also not suggest, implicitly or explicitly, that higher return rates should be an objective, or would imply more ‘effectiveness’ in the EU’s efforts to control irregular migration (for a critical dis- cussion of ‘effectiveness’ in the field of returns, see Carrera & Allsopp, 2018; Schuster & Majidi, 2013). Instead, we seek to shed light on the often-­made link between readmission cooperation and a third country's return rate. This makes us also highlight that the return rate may be an inadequate tool for assessing the readmission cooperation of a third country. The article is structured into three parts. It starts with a conceptual discussion why the EU’s readmission cooperation has often been connected to the EU’s return rate. The next section presents the data challenges in measuring the EU return rate. The main empirical part presents different statistics and data estab- lishing the extent to which the EU’s formal return cooperation impacts the return rate. FRO M EU R E A D M I S S I O N CO O PE R ATI O N TO A H I G H E R R E T U R N R ATE ? This section conceptualises the link between the EU’s return rate and the EU’s readmission cooperation. Negotiations on EU readmission agreements belong to the most contested ones in the Union's external relations THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       3 with third countries (see e.g. Icduygu & Aksel, 2014; Wolff, 2014). EU readmission agreements are overwhelm- ingly in the interest of the EU as the primary direction of return is from Europe to the outside world (Coleman, 2009; Kruse, 2006). The conclusion of negotiations on readmission often depends on factors such as the strength of the incentives offered by the EU, the geographical proximity of the third country in question or the salience of the migration theme in this country (Cassarino, 2010a, 2010b). In Africa, a (too) close cooperation with the EU on return issues is often seen as a betrayal of the own citizens, many of whom depend on the remittances of migrants who made it to Europe (Adam et al., 2020). In view of these sensitivities, the EU has increasingly refrained from concluding formal EU readmission agree- ments but asked for more informal, non-­legalised readmission arrangements or, simply, return deals (Cassarino, 2018; Slominski & Trauner, 2020). In the Commission's view (European Commission, 2017: 1f), these arrangements improve ‘practical cooperation’ on the identification and return of irregular migrants and ‘represent a first step in establishing mutual trust as well as towards the launch of formal negotiations for fully fledged readmission agree- ments’. While they are legally non-­binding, informal readmission arrangements are hence supposed to lead to higher return numbers too (albeit the Commission has higher expectations for formal EU readmission agreements). Thus, we look at both informal and formal readmission cooperation in our analysis, given their shared objective of raising return numbers. The establishment of a closer cooperation on readmission between the EU and a third country tends to involve complex, tedious and long negotiations. It is therefore no surprise that many policymakers and observers believe such an output (i.e. the establishment of readmission cooperation) will lead to changes in the outcome (i.e. a higher return rate between the EU and the third country under question). However, there are two important caveats to such an assumption. First, EU readmission cooperation and the EU’s return rate are determined by different dy- namics, at least to some extent. As can be seen in Table 1, the EU readmission cooperation reflects the outcomes of intergovernmental negotiations. The EU’s return rate refers to the number of migrants who left after an order to leave. Aside from the cooperation on readmission, the EU’s return rate also depends on how return numbers are counted (see next section, also Mananashvili, 2017; Singelton, 2016) and, importantly, on the behaviour of migrants. The EU’s return rate counts migrants leaving voluntarily after a removal order, so it depends on what migrants may do (or are compelled to do) after an order to leave (on the issue of ‘voluntariness’ of voluntary return, see Kuschminder, 2017; Webber, 2011).1. The return rate can be also high or low for other reasons. For instance, states may have no formal readmission cooperation with the EU but still accept the return of irregular migrants, thereby contributing to higher return rates. There is also a second caveat. Even if the EU has a formal or informal readmission arrangement in place, this does not automatically imply that a third country will take back more migrants found ineligible to stay in the Union. The actual removal procedures may be impeded by a range of domestic and international factors (see Ellermann, 2005; Trauner, 2018). Within the EU, some returns are prevented by ad hoc, local ‘grassroot’ protests TA B L E 1   Conceptualising EU readmission cooperation and the EU return rate EU readmission cooperation EU return rate Reflects… …intergovernmental relations on readmission …the number of migrants who left after an and return order to leave Focuses on… …negotiation and implementation behaviour …return numbers and statistics –­not the of state actors (migrants as ‘recipients’ of reasons for return policies) Depends on… …factors such as EU incentives and pressures; …factors such as EU return cooperation; geographical proximity; domestic migrant behaviour (voluntary return); opposition (driven by e.g. the strength non-­refoulement considerations; data of the diaspora and the salience of the reliability and calculation method migration theme) | 4      STUTZ and TRAUNER (Rosenberger & Winkler, 2014) or judicial challenges (Acosta & Geddes, 2013). Some states such as Portugal or Spain have never displayed a similar eagerness as their colleagues from northern EU states to follow up on removal orders with a forced return procedure (Finotelli, 2018). Not all member states have concluded or have sought to conclude the same amount of ‘implementing protocols’ with a third country, which are of relevance for the actual application of EU readmission agreements (Carrera, 2016). Moreover, existing human rights standards (e.g. principle of non-­refoulement; prohibition of collective expulsion) also make it difficult for member states to return migrants to third countries with a dismal human rights record (Coleman, 2009; Slominski, 2012). Having concluded a readmission agreement or not, unrest, civil war or authoritarianism in third countries can lead to a halt of return operations. This article acknowledges these scholarly insights yet goes one step further. By applying a worldwide view, it demonstrates just how little the EU’s readmission cooperation actually matters for any change in the return rate with a third country. Only in Eastern Europe and the Western Balkans, a closer cooperation on readmission seems to have led to steadily high(er) return rates. Yet, even in these regions, a word of caution is needed. Once several Eastern European and Western Balkan states signed an EU readmission agreement and had increasing return rates with the EU, the remaining neighbours moved in a similar direction. States such as Belarus (which back then had not yet had a readmission agreement with the EU) or Kosovo (no EU readmission agreement) started to display re- turn rates that were already comparable to their neighbours (which back then had an EU readmission agreement). This has not changed over time. Therefore, the question is whether it was indeed the EU’s institutionalisation of readmission cooperation that triggered higher return rates –­or whether return cooperation has followed a wider regional dynamic. In the case of Eastern Europe and the Western Balkans, having closer political relations with the EU has become an objective widely shared by these states, with cooperation on migration issues being one component of the overall process of rapprochement (or accession). Accepting the return of irregular migrants may have become an expression of political goodwill to achieve closer relations with the EU. The same dynamic is also visible in the opposite direction. The EU’s signing of a (formal or informal) readmis- sion agreement with Asian or African countries has most often led to no or only a temporary increase of the part- ner countries’ return rates. Outside Europe, therefore, we can also trace regional trends. If a majority of states in a region had a low return rate, those states that used to have higher ones converged with the lower levels of their neighbours. A formal or an informal EU readmission agreement tends to alter these dynamics only temporarily –­if at all. These findings question the EU’s strategy of seeking to sign more formal or informal readmission arrange- ments by using, if deemed necessary, more negative conditionalities (European Commission, 2020). Readmission cooperation may lead to higher return rates only in very particular circumstances. The following sections will substantiate these arguments with the available data. TH E R E T U R N DATA C H A LLE N G E The EU’s statistical office Eurostat publishes different data sets relating to the enforcement of immigration legisla- tion. They include data on persons ordered to leave and returned to a third country after an order to leave. The ‘return rate’ is established by dividing the latter by the former. 2. This ratio is not calculated by Eurostat. Yet, EU institutions such as the Commission's Directorate General for Migration and Home Affairs (DG Home) calculate the return rate in a similar way, albeit they rarely publish full statistics under this title. It is important to highlight that the EU’s Eurostat data on return is problematic for the following reasons: Data quality and reliability: The rate may be misleading as many people quit a territory after an order to leave without notifying the local authorities. Also, individuals may receive an order to leave by one member state but have already moved to another (this is referred to as ‘secondary movements’ in the EU context). This can inflate the number of orders. The EU has not yet a reliable mechanism to find out whether a person has actually THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       5 left the EU’s territory. Structural problems: Eurostat data on migrants, asylum seekers, visa recipients and readmitted persons are collected nationally. Member states still have diverging definitions and methods regarding the collection of data on the legal status of migrants. Many member states do not always use the same statistical baseline, they distinguish on which grounds a person has left a territory and how the person has been returned or readmit- ted. Even though the data has improved over the years and has become increasingly comparable, there is not yet a full harmonisation as to how different national statistical institutes collect the migration data and share it with Eurostat (Singelton, 2016). Technical issues: These structural problems may lead to technical problems. Member states often share their data with Eurostat in a delayed way, distorting some yearly numbers. Another issue relates to backlogs, nota- bly between the issuance of an order to leave and the actual removal. This backlog can lead to years in which the number of returned people exceeds the number of orders to leave, resulting in a return rate of over 100 per cent. Implications of choosing a method: Different calculation methods lead to varying results. Crucially, the EU return rate differs from the average of the return rates of all countries. Why is this so? The EU return rate is calculated by accumulatively comparing all persons that have been returned after an order to leave with all orders to leave issued by member states. This leads to an average return rate of 37.7 per cent for 2008–­18. Similar to the EU, we mostly use this method for this article. However, if we calculate the return rate for each country and then take the average mean between these values, we may get a different –­and higher –­total return rate, namely 43.2 per cent for the same period. Problems of interpretation: The dominant method of calculating the EU return rate also hides the fact that changes in one region may impact the overall return rate. For instance, the EU is now referring to ‘a consid- erable decrease in the rate of return from 45.8% in 2016 to merely 36.6% in 2017’ (European Commission, 2018a: 15).3. Yet, the average return rate for period 2008–­18 was 37.7 per cent, making the year 2016 the only real outlier. The higher return percentage of this year was primarily caused by a larger number of migrants sent back from the EU to the Western Balkans. Most of them were Kosovars and Albanians, entering the EU in the tens of thousands between 2014 and 2016. Their asylum applications were often rejected in accelerated pro- cedures. If 46 per cent is now promoted as a new benchmark while the return co-­operation to the candidate countries of the Western Balkans has returned to prior (lower) numbers, this may enhance the pressure on all other countries to co-­operate more with the EU. These issues impact the quality and reliability of the data. However, a view on longitudinal trends and a com- parison of all regions in the world allows to mitigate some of the technical problems, which are likely to be of more relevance in specific years and/or with particular partners. Concretely, we have established the EU return rate for the total of the EU’s external relations for the period from 2008 until 2018. A B I R D ’ S - ­E Y E V I E W O N TH E EU R E T U R N R ATE In this section, we will present a worldwide view of the return rate. We also show absolute numbers of irregular migrants and statistics highlighting dynamics across and within regions. Figure 1 presents the persons ordered to leave and returned in relation to irregular migration in the EU (for a complete list of all countries’ average removal orders and numbers of persons returned see Table A2 in the annex). The number of asylum seekers rejected in first instance more than doubled from 2014 to 2017. Between the years 2014 and 2017, a total of about 2.5 million individuals were illegally found present in the EU. These two groups of persons are potentially subject to the return policies of member states. In some countries, there is even a direct link (cf. Mananashvili, 2017). The rejections for asylum applications are sent out together with orders to | 6      STUTZ and TRAUNER leave. With this substantial increase of persons deemed ineligible to stay, we may expect an increase of returned persons, in absolute figures, given the EU’s prioritisation of return operations. However, the number of persons being returned increased from 2014 to 2016 by only one-­third (34 per cent), reaching the levels of 2008–­2010. It was almost back to previous levels in 2017. Thus, the return rate, being the ratio of persons returned to removal orders, has been decreasing in the last years (see also EU-­average in Figure 3). In 2018 it decreased to 35.6 per cent, the level of 2008 (35%) when the figures for EU return were published first. 2,250,000 2,000,000 1,750,000 1,500,000 1,250,000 1,000,000 750,000 500,000 250,000 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 First instance asylum rejection Illegally found present TCN Ordered to Leave TCN Returned to TC F I G U R E 1   Irregular migration and return: absolute numbers 2008–­2018. F I G U R E 2   The EU’s return rate with third countries, average 2008–­2018. THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       7 Return Rate 2008-2018, in % 45 Standard Deviation 2008-2018, in % 90 Region S. Eastern Africa 80 Europe 40 Western MENA 70 Balkans 35 60 Region S. 30 Africa EU average 50 25 East&Cent. 40 MENA 20 Africa EU average Western 30 15 Balkans Eastern East&Cent. Europe 20 Africa 10 10 West West 5 Africa Africa 0 0 2008 2013 2018 2008 2013 2018 F I G U R E 3   Comparing dynamics across and within selected regions. In Figure 2, we see a worldwide picture of the EU return rate. The darker a country is, the higher is the return rate. The return rate is highest with South-­E astern and Eastern neighbours, Northern American states, Southern Africa (mostly due to low removal order figures), and Australia. Asia has higher rates compared to sub-­S aharan Africa (and also South America). Regionally, the Western Balkans and Eastern Europe have increased their return rate since 2014 (77% in 2018 respectively). They have now the highest rates, even before other (non-­EU) Organisation for Economic Cooperation and Development (OECD) countries (71% in 2018). The lowest return rates are visible with regard to sub-­S aharan African countries (West Africa 10%; Eastern and Central African countries 17% in 2018), countries in Central and South Asia (38%) and the MENA region (41%). Split into two parts, Figure 3 highlights the dynamics across and within regions. Based on Eurostat data, we calculated the average mean of the countries that make up a region4. and focused on regions in the EU’s neigh- bourhood and Africa (priority regions for the EU in the field of return). On the left side, we can see that the regions that had comparatively high return rates (notably the Western Balkans and Eastern Europe) kept and even in- creased them by 2018. In the MENA region, the regional average decreased by almost 10 percentage points since 2013. Countries in West, East and Central Africa have all had decreasing regional averages since 2008. Overall, the figure shows that regions have become more different from one another, especially regions where the return rate has been high or low. On the right side, we focus on the dynamics within each region by looking at the standard deviation (i.e. the dispersion of countries in a region from the regional average for the years 2008, 2013 and 2018). If a standard deviation increases, the return rates of countries that make up a region may vary more. However, a low standard deviation does not automatically imply that countries vary less. In the region of West Africa, the standard devia- tion for 2018 is ‘only’ 7.8 per cent. This has to be seen in view of the regional average of ‘only’ 9.6 per cent. In other | 8      STUTZ and TRAUNER words, the variation is almost as high as the actual average value. Overall, the figure highlights a divergence trend in several regions, notably West Africa, East and Central Africa and Eastern Europe and the Western Balkans. In Eastern Europe and the Western Balkans, practically all states have started to have relatively high return rates with the EU, thus in contrast showing less variance. The trend is reverse in the other regions (MENA, Southern, East and Central and West Africa). Even though in these regions, states have started to have similar (often decreasing) return rates (and often at a low level), there are regional ‘exceptions’ in the sense that a few countries had higher return rates with the EU (though less frequently). For instance, all of the 16 countries that make up the MENA region developed lower and decreasing return rates, with the exceptions of Israel, Lebanon and Kuwait. This explains the actual increase in the standard variation in comparison to the regional averages. Put differently, most countries in the regions converge at a low level yet remain in contrast to very few outliers that have stable return rates (but rarely increased ones). Overall, regions in Africa provide a stark contrast to the Western Balkans and Eastern Europe, where countries converged at higher rates and became more homogeneous within their regions. TH E R E L ATI O N S H I P O F EU R E A D M I S S I O N CO O PE R ATI O N A N D TH E R E T U R N R ATE This section will elaborate in more detail on how the cooperation on readmission (be it through readmission agree- ments or informal return deals) may actually impact the return rate. We will start with the states in Eastern Europe and the Western Balkans. Long-­term EU readmission cooperation: Eastern Europe and Western Balkans The first readmission agreements that the EU ever signed were with the Western Balkan states, Moldova (2008), Ukraine (2008) and Russia (2007). Albania concluded the agreement in 2006; Bosnia and Herzegovina, Serbia, Montenegro and North Macedonia concluded their EU readmission agreements in 2008 (black vertical line when in 2008). 5. Our data covers the period from 2008 to 2018. For these states, therefore, we cannot compare the time periods pre-­and post-­EU readmission agreement but only interpret the data post-­2008 (see Figure 4). In our period of examination, the Eastern European and Western Balkan states had the highest return rates with the EU of all countries in the world. At certain years, the number went beyond 100 per cent, implying that the rate of people being returned or having left is higher than the number of removal orders. The return rates fluc- tuated also in these two regions, with Russia having the lowest rate. In Eastern Europe and the Western Balkans, it seems obvious to assume that there is a link between readmission cooperation and the EU’s return rate. The EU readmission cooperation has been backed with strong EU incentives. With the exception of Russia,6. all these states were offered visa-­free travel with the EU in exchange for close cooperation on readmission (and other areas). This was a yearlong process in which the EU defined a long list of demands which the Eastern European and Western Balkan states had to fulfil to get visa-­free travel (Trauner & Manigrassi, 2014). If cooperation on re- admission were found to be lacking, the EU would have had the option to re-­introduce visa requirements through a ‘visa suspension mechanism’. The Eastern European and Western Balkan states therefore constitute a particular (if not unique) setting in the EU’s external migration cooperation. They have been geographically close, keen to have close relations (if not full EU membership) and the EU has provided them with strong and tangible incentives (full visa-­free travel). However, the next section shows that other states in these regions also had increasing return rates, even before they formalised their readmission cooperation. THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       9 Albania Bosnia• Herzegovina Moldova 120 70 80 100 60 70 80 50 60 60 40 50 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 Montenegro North Macedonia Russia Return Rate in % 160 70 120 120 100 60 80 80 60 50 40 40 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 Serbia Ukraine 100 80 80 70 60 60 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 Year F I G U R E 4   Return rates for countries in the Western Balkans and Eastern Europe. Note: the y-­axes are different for each country. Countries with formal or informal EU readmission arrangements The next figure includes all third countries that signed a formal EU readmission agreement or an informal EU read- mission arrangement in the period from 2008 to 2018.The year of conclusion is indicated by the vertical line. The figure allows to identify three scenarios. The first is the ‘temporary peak’-­scenario, most visible for Afghanistan, Pakistan and Turkey (indicated by ‘s1’ before the name of the country in Figure 5). It implies that the return rate has temporarily gone up but has fallen back to previous numbers in a relatively short time period. Neither an agreement nor an arrangement (Afghanistan's Joint Way Forward is an informal deal signed in October 2016; Turkey and Pakistan have a formal EU readmission agreement) have a lasting impact on the return rate and have not permanently increased the return rate. The second is the ‘no-­impact’-­scenario (‘s2’-­countries). This implies that the effect of a readmission agreement has been close to non-­existent from scratch. For instance, Cape Verde has a very low return rate before and after the readmission agreement was concluded. Other African countries like Ethiopia, the Gambia and Guinea show little change over the last years, albeit it is too early to say whether their in- formal arrangements have an impact on the return rates. They were concluded only towards the end of our period of analysis. The third scenario is an increase of the return rate (‘s3’). This scenario is visible primarily for countries in Eastern Europe (Georgia, Armenia and Azerbaijan). In absolute figures, the number of people having received an order to leave has remained rather stable (from 6,900 in 2010 to 7,300 in 2017; 2018: 9,700 for Georgia; from 1,300 to 1,600 for Azerbaijan). However, the number of people being returned almost tripled for both countries (Georgia ca. 2,500 in 2010 and 6,400 in 2018; Azerbaijan: 500 in 2010 and 1,500 in 2018). 10      | STUTZ and TRAUNER s1_Afghanistan s1_Pakistan s1_Turkey s2_Bangladesh 50 45 35 40 20 40 30 35 25 10 30 30 20 25 15 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 s2_Cape Verde s2_Cote d'Ivoire s2_Ethiopia s2_Gambia 12.5 12.5 30 10.0 20 10.0 16 20 7.5 Return Rate in % 7.5 5.0 12 10 5.0 8 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 s2_Guinea s3_Armenia s3_Azerbaijan s3_Belarus 25 90 90 20 40 80 80 15 70 30 60 70 10 50 5 20 40 60 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 s3_Georgia 60 50 40 30 20 2008 2010 2012 2014 2016 2018 Year F I G U R E 5   Return rate for return deal countries with an EU readmission arrangement. Note: the y-­axes are different for each country. As Figure 5 highlights, an increase of the return rate has only been visible in the case of Eastern European states. It is, however, uncertain whether the increase of these countries’ return rate is directly related to the EU readmission agreement. Belarus, which only concluded an EU readmission agreement in 2020 (hence, after the period of our investigation) and Georgia already had increasing return rates before the conclusion of the readmis- sion agreement, indicating higher political and operational efforts to accommodate EU migration priorities and allowing the EU to ‘create an umbrella framework on migration in the region’ (Makaryan & Chobanyan, 2014: 62). This stands in contrast to the context in which the recent informal EU return deals with Bangladesh, Gambia, Ivory Coast, Ethiopia and Guinea have been signed (on the informalisation of EU migration management, see Cardwell, 2016; Cassarino, 2018). Their return rate with the EU had been steadily falling before they agreed on an EU read- mission arrangement in 2017 or 2018 (and have since remained stable at a low level or even further decreased). West Africa: a region with a weakly institutionalised cooperation on readmission In Figure 6, we focus on West Africa. Several EU migration initiatives such as the ‘Migration Partnership Framework’ have considered this region as a priority (European Commission, 2016). The co-­operation on readmis- sion remains weakly institutionalised, however. Only Cape Verde has a formal EU readmission agreement; Cote d’Ivoire, Gambia, Guinea have recently signed informal EU readmission arrangements (these cases are discussed in Figure 5). We focus now on nine other West African countries (Niger, Burkina Faso, Nigeria, Ghana, Senegal, THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       11 F I G U R E 6   Return rates from 2008 to 2018 for selected West African countries. Note: both axes are fixed. Guinea-­Bissau, Sierra Leone, Mali and Togo) and compare their ‘no-­cooperation’-­background with their four re- gional neighbours cooperating with the EU on readmission. In 2008, two countries (Ghana and Nigeria) were close to the average of the EU return rate. In 2011 and 2012, Nigeria was above the EU’s average and close to 40 per cent. The return rates of the other countries were between 15 and 20 per cent in 2008. For practically all these countries, the return rates have declined over the years. Ghana's numbers dropped from 34 per cent in 2012 to less than 19 per cent in 2017, only to climb again to 31 per cent in 2018. Most other states including Mali, Sierra Leone, Togo and Niger were between roughly 2 and 12 per cent in 2018. Figure 6 also shows that there is a growing difference between Nigeria and Ghana, on the one hand, and the rest of the region, on the other. With the exception of these two states, the other countries converged at a lower level (as shown in Figure 3). The four states that have agreed with the EU on a formal or informal readmission arrangement (Cape Verde, Cote d’Ivoire, Gambia and Guinea) fit into this regional trend. Their 2018 return rates were lower than in 2008 and ranked between 2 and 10 per cent. Therefore, the formal or informal EU readmission arrangements do not make a big difference if viewed from a regional perspective. Other countries with high or low return rates In this last Figure 7, we look at outliers and (substantial) changes over time. The y-­axis shows the return rate for the year 2018. This helps to locate a country's return rate vis-­à-­vis all other third countries. The x-­axis seeks to highlight the dimension of change. It reflects changes in the return rate over the period from 2013 to 2018 in percentage points. By comparing the return rates of 2013 and 2018, we account for more recent changes. For instance, Zambia had a return rate of over 90 per cent (y-­axis) in 2018. From 2013 to 2018, Zambia's rate increased by 30 percentage points of the return rate (x-­axis). The figure allows us to identify substantial changes (towards higher or lower return rates) outside the regions and cases that we have analysed beforehand. 12      | STUTZ and TRAUNER 100 South Africa United States Israel Zambia Saudi Arabia Canada Mexico 90 Australia Chile 80 Paraguay South Korea Jordan 70 Jamaica Return Rate 2018 60 50 40 Japan 30 Tunisia North Korea 20 Libya Gambia Senegal Yemen 10 Syria Somalia Eritrea 0 -50 -40 -30 -20 -10 0 10 20 30 40 Return Rate %-Pointage Change 2013-2018 F I G U R E 7   Changes in the return rate 2013–­2018 for selected countries. The figure highlights several factors of influence that are central to understanding the EU’s return rate. First, the rates of some countries have almost entirely been determined by the number of voluntary returns. For in- stance, Saudi Arabia had a return rate of 100 per cent throughout the period from 2013 to 2018 (it never had a formal or informal readmission arrangement with the EU). The high return rate was caused by the fact that com- paratively few Saudi citizens received an order to leave from EU member states (2,250 persons in the 11 years under examination here). Practically all returned voluntarily to Saudi Arabia.7. In general, voluntary returns have become more important in recent years regardless of the fact that the EU’s policies have focused on achieving more returns primarily by means of more restrictive practices (Eisele et al., 2019; Sundberg Diez, 2019). They explain, for instance, also the increase of the return rate of North Korea or why the EU return rate with other autocratic regimes such as Eritrea, Somalia and Syria is higher than ‘0’. According to our data, the number of volun- tary returns is roughly equal with the number of forcible returns. Based on data reported by 22 member states,8. 210,260 persons returned forcibly and 217,095 persons voluntarily in the period from 2014 to 2018. Second, the state of democracy/autocracy or a democratisation process seem to be of relevance (although future research may be needed to substantiate such an assumption). The return rates of Yemen and Libya fell substantially in view of their internal instability and conflicts. The principle of non-­refoulment has also set limits to EU member states even if these human right safeguards have already become weaker in many of these cases (Giuffré, 2020). The cases of Tunisia, Senegal and the Gambia are likely to indicate that countries becoming more democratic, especially in Africa, may also face more pressures from the electorate and the diaspora. This might be reflected by their stagnating or declining return rates with the EU. THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       13 The third take-­away message of Figure 7 is a more technical one. Changes can quickly seem to be significant if the overall number of removal orders and returns is low. A case in point is Zambia. In 2013, the EU issued 160 removal or- ders for Zambians, with 65 migrants leaving or being returned. In 2018, these numbers were 100 and 60 respectively. They made the return rate nominally increase by 30 points. By comparison, Jordan had a decrease of 21 per cent from 2013 to 2018. In absolute figures, the country had 340 removal orders in 2013 and 435 in 2018. The returns almost stayed the same with 295 and 285 respectively. The cases of Jamaica and Paraguay are similar: the removal orders decreased after 2013 but the number of returns remained very stable. We can see this with quite a few countries in other regions as well. In doing so, the return rate gradually increases, although the number of actual returns is not significantly altering. The non-­EU OECD countries (US, Australia, Canada, South Africa, South Korea, Israel, Chile and Mexico) also had comparatively high return rates and low numbers of removal order (none of these countries has a readmission agreement with the EU). Increases –­or, in the case of Japan, the steep decrease –­may seem rather big due to these small numbers. With many countries in the world, therefore, the return rate can be subject to (seemingly substantial) changes, for example, when one or few member states start to issue more or less removal orders for a given nationality. The example of the OECD countries (outside the EU) also indicates that, in general, other factors may need to be further investigated such as trade relations or development cooperation. CO N C LU S I O N This article has provided a critical analysis of the link between EU readmission cooperation and the EU return rate. It has done so by looking at the return rate of all countries in the world over the period from 2008 to 2018. The article explicitly acknowledges that the return rate can be influenced by factors other than the EU’s readmission cooperation (for instance by migrants’ voluntary returns or the method of measurement). However, the article's focus has been particularly on the impact of formal and informal readmission arrangements. It therefore focused on changes in the return rate of third countries following (or preceding) their conclusion. While scholars have already demonstrated that the implementation of EU readmission agreements may be challenged by a range of domestic and international factors (e.g. Carrera, 2016; Ellermann, 2005), this article goes one step further and highlights just how little a formal EU readmission agreement or an informal EU readmission arrangement may matter for the return rate of a third country. Some states have high or low ones regardless of whether or not they have a readmission arrangement with the EU. We present three scenarios (‘temporary peak’-­, ‘no impact’-­, ‘increase’-­scenario) on the impact of EU readmission arrangements on the return rate. The increase-­ scenario is primarily visible with states that have closer relations with the EU and/or are interested in integrat- ing into the Euro-­Atlantic structures (such as Eastern European countries). In general, the Western Balkans and Eastern European states are the only regions in which the EU’s readmission agreements seem to have led to higher return rates. In fact, the increasing return rates of these Eastern and South-­Eastern neighbours were a key reason why the EU managed to keep the average return rate at roughly the same level between 2008 and 2018. Many other regions in the world –­most pronouncedly so in Africa –­had declining return rates with the EU. However, even in Eastern Europe and the Western Balkans, the importance of formalising readmission co- operation is questionable. Those Eastern European states that had not yet a formal EU readmission agreement moved towards higher return rates, once their neighbours did so. In Africa, this dynamic worked in the opposite direction. If all states in a region had decreasing return rates, the signing of EU readmission arrangements with a few third countries was not likely to alter a wider regional trend. The states that signed such an arrangement were likely to converge soon again with their neighbours (i.e. they mostly moved towards their lower return rates). Why was this the case? It would go beyond the scope of this article to elaborate on and compare the regional dynamics in more detail. However, the explanation is likely to relate to how migration is organised and framed between the EU and a given region. Travelling to and entering the EU has become easier for citizens of Eastern Europe and the Western Balkans. Visa liberalisation has been a psychological boost for these countries, whose | 14      STUTZ and TRAUNER citizens long felt to live behind a ‘Schengen paper wall’ (e.g. ICG, 2005). It was an objective to which most of the wider European region started to subscribe. It has also contributed to a less politicised migration cooperation and an acceptance of the underlying bargain (visa liberalisation in exchange for pan-­European migration management). In contrast, the decreasing return rates of the African regions seem to coincide with the increasing polarisation of African migration into the EU (e.g. de Haas, 2008). The return rates of West or East African countries were almost all higher in 2008 compared to 2018. This contrasts with the considerable effort of the EU to have closer readmis- sion cooperation with these states during the same period of time. If these assumptions are valid, more conditionalities and a ‘punchier’ EU approach vis-­à-­vis third countries will be unlikely to lead to higher return rates (for a call of such an approach, see European Commission, 2020). The opposite may be more effective. Higher return rates may come from making access to the EU more tangible for the population of a third country. AC K N OW L E D G E M E N T S The authors would like to thank Ilke Adam and two anonymous reviewers for their constructive comments on the article. ORCID Philipp Stutz  https://orcid.org/0000-0001-6341-476X E N D N OT E S 1. They may leave on their own initiative or take up an offer of an assisted voluntary return programme run by the International Organization for Migration (IOM). The EU seeks to influence the cost-­benefit calculi of affected migrants by increasing pressures and relying on coercive practices, see Sundberg Diez, 2019; Eisele, Muir, Molinari, Fernandes and Galea, 2019. 2. We do this separately for each year but also for the whole period between 2008 and 2018 and for each country with the average mean. When there is no data available for a year or no person is ordered to leave or has been returned, we treat the case with a return rate of ‘0’. Countries that consistently have less than 25 removal orders per year are excluded from the analysis. While we do show values above 100 per cent in the regional analysis for Figure 3, in the other parts of the paper we cut off the return rate at 100 per cent. 3. These numbers were calculated based on provisional Eurostat data. The actual return rates for 2016 and 2017 were 47.1% and 37.5% respectively, as presented in this article and as shown by more recent DG Home documents (e.g. DG Home 2020: 11). 4. We look at countries for which more than 25 removal orders were issued per year and cut off by 100 per cent for the years 2008, 2013 and 2018. 5. Sri Lanka, Hong Kong and Macao also signed an EU readmission agreement pre-­2008. They are, however, quite special cases (special administrative regions of China and an island country in South Asia). It would go beyond the scope of this article to deal with them in detail. Belarus only signed its EU readmission agreement in 2020. 6. The Visa Free Dialogue was frozen as part of the sanctions the EU imposed on Russia after the annexation of Crimea. 7. Based on Eurostat data of 15 member states, no citizen of Saudi Arabia has been returned forcefully; all returns have been conducted ‘voluntarily’. Confirmed by Eurostat upon request by the authors. 8. Reporting countries: AT (since 2017), BE, BG, CZ, DK, EE, IE, ES, FR, GR (since 2018), HR, IT, LV, LU, HU, MT, PO, PT, RO, SI (until 2017), SK, SE. REFERENCES Acosta, D. & Geddes, A. (2013) The development, application and implications of an EU rule of law in the area of migration policy. Journal of Common Market Studies, 51(2), 179–­193. Adam, I., Trauner, F., Jegen, L. & Roos, C. (2020) West African interests in (EU) migration policy. Balancing Domestic Priorities with External Incentives, Journal of Ethnic and Migration Studies, https://doi.org/10.1080/13691​ 83X.2020.1750354. Cardwell, P.J. (2016) Rethinking law and new governance in the European Union: the case of migration management. European Law Review, 41(3), 362–­278. THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES |       15 Carrera, S. (2016) Implementation of EU Readmission Agreements. (ebook: Springer). Carrera, S. & Allsopp, J. (2018) The irregular immigration policy conundrum. Problematizing ‘effectiveness’ as a frame for EU criminalisation and expulsion policies. In: Ripoll Servent, A. & Trauner, F. (Eds.) Routledge handbook of justice and home affairs research. London: Routledge. Cassarino, J.-­P. (2010a) Dealing with unbalanced reciprocities: Cooperation on Readmission and Implications. In: Cassarino, J.-­P. (Ed.) Unbalanced reciprocities: Cooperation on readmission in the Euro-­Mediterranean area. Washington: Middle East Institute, pp. 1–­3 0. Cassarino, J.-­P. (2010b) Readmission Policy in the European Union. Brussels: European Parliament. Cassarino, J.-­P. (2018) Informalising EU readmission policy, In: Ripoll Servent, A. & Trauner, F. (eds.) Routledge handbook of justice and home affairs research, London: Routledge, pp. 83–­98, forthcoming. Coleman, N. (2009) European readmission policy. Third country interests and refugee rights. Leiden: Martinus Nijhoff Publishers. de Haas, H. (2008) The Myth of Invasion: the inconvenient realities of African migration to Europe. Third World Quaterly, 29(7), 1305–­1322. Der Tagesspiegel. (2017) Alice Weidel, die neue Rechte. Eisele, K., Muir, E., Molinari, C., Fernandes, M. & Galea, A. (2019) The proposed Return Directive (recast). Substitute impact assessement. Brussels: European Parliamentary Research Service. Ellermann, A. (2005) Coercive capacity and the politics of implementation. Deportation in Germany and the United States, Comparative Political Studies, 38(10), 1219–­1244. European Commission. (2011) Evaluation of EU Readmission Agreements. EU Readmission Agreements: Brief Overview of state of play SEC(2011) 209, 23 February 2011. European Commission. (2015) EU Action Plan on Return, Brussels: COM(2015)453 final. European Commission. (2016) Communication on establishing a new Partnership Framework with third countries under the European Agenda on Migration'COM (2016)385 final, 7 July 2016 European Commission. (2017) EU readmission developments -­State of play October 2017, Letter to Mr. Claude Moraes, MEP, Chair of the Committee on Civil Liberties, Justice and Home Affairs. European Commission. (2018a) Progress report on the implementation of the European Agenda on Migration COM(2018) 301 final. European Commission. (2018b) Proposal for a Directive of the European Parliament and of the Council on common standards and procedures in Member States for returning illegally staying third-­country nationals. (recast)'COM(2018)634 final, 12 September 2018. European Commission. (2020) Communication … on a New Pact on Migration and Asylum, Brussels, 23 September 2020: COM(2020) 609 final Finotelli, C. (2018) Southern Europe: twenty-­five years of immigration control on the waterfront. In: Ripoll Servent, A. & Trauner, F. (Eds.) Routledge handbook of justice and home affairs research. London: Routledge, pp. 240–­252. Giuffré, M. (2020) The Readmission of Asylum Seekers under International Law. Oxford: Hart Publishing. Icduygu, A. & Aksel, D.B. (2014) Two-­to-­t ango in migration diplomacy: negotiating readmission agreement between the EU and Turkey. European Journal of Migration and Law, 16(3), 337–­363. ICG (2005) EU Visas and the Western Balkans, Europe Report Nr 168–­29 November 2005. Brussels: International Crisis Group. Junker, J.-­C . (2017) State of the Union Address 2017. Brussels. Kruse, I. (2006) EU readmission policy and its effects on transit countries -­ the case of Albania. European Journal of Migration and Law, 8(2), 115–­142. Kuschminder, K. (2017) Taking stock of assisted voluntary return from Europe: decision making, reintegration and sustainable return –­Time for a paradigm shift. EUI Working Papers RSCAS 2017/31. Florence. Makaryan, S. & Chobanyan, H. (2014) Institutionalization of migration policy frameworks in Armenia. Azerbaijan and Georgia, International Migration, 52(5), 52–­67. Mananashvili, S. (2017) EU’s return policy: Mission accomplished in 2016? Reading between the lines of the latest EUROSTAT return statistics. Vienna: International Centre for Migration Policy Developement Policy Brief. Rosenberger, S. & Winkler, J. (2014) Com/passionate protests –­Fighting the deportation of asylum seekers. Mobilization, 19(2), 489–­510. Schuster, L. & Majidi, N. (2013) What happens post-­deportation? The experience of deported Afghans, Migration Studies, 1(2), 221–­240. Singelton, A. (2016) Migration and Asylum Data for Policymaking in the European Union -­The problems with numbers, CEPS Paper No 89. Brussels: Centre for European Policy Studies. Slominski, P. (2012) The ambiguities of legalization and the EU’s strategy of extra-­territorial border control, European Foreign Affairs Review, 17(Special Issue 2012), 183–­198. | 16      STUTZ and TRAUNER Slominski, P. & Trauner, F. (2020) Reforming me softly -­ how soft law has changed EU return policy since the migration crisis. West European Politics, 44(1), 93–­113. https://doi.org/10.1080/01402​382.2020.1745500 Sundberg Diez, O. (2019) Diminishing safeguards, increasing returns: Non-­refoulement gaps in the EU return and readmission system, EPC Discussion Paper Brussels. Trauner, F. (2018) Return and readmission in Europe –­ understanding negotiation and implementation dynamics. In: Weinar, A., Bonjour, S. & Zhyznomirska, L. (Eds.) The Routledge handbook on the politics of migration in Europe. London: Routledge, pp. 251–­260. Trauner, F., & Manigrassi, E. (2014) When visa-­free travel becomes difficult to achieve and easy to lose: The EU visa free dialogues after the EU’s experience with the Western Balkans. European Journal of Migration and Law, 16 (1), 125–­145. http://doi.org/10.1163/15718​166-­0 0002051 Webber, F. (2011) How voluntary are voluntary returns? Race & Class, 52(4), 98–­107. Wolff, S. (2014) The politics of negotiating EU readmission agreements: Insights from Morocco and Turkey. European Journal of Migration and Law, 16(1), 69–­95. How to cite this article: Stutz, P.&Trauner, F. (2021) The EU's ‘return rate’ with third countries: Why EU readmission agreements do not make much difference. International Migration, 00, 1–­19. https://doi. org/10.1111/imig.12901 ANNEX TA B L E A 2   Annual average of persons returned and ordered to leave (2008–­2018). avg. avg. Removal avg. avg. Removal avg. avg. Removal Country Returned Orders Country Returned Orders Country Returned Orders EU-­Total 191643 510162 East and Central Africa North and Central America Western Balkans Burundi 62 352 Costa Rica 43 78 Albania 33940 38744 Cameroon 425 3718 Cuba 93 799 Bosnia/Herz. 2244 3756 Cent. Afr. Rep. 16 275 Dominica 42 315 Kosovo 5749 8671 Chad 29 480 Domin. Rep. 316 1580 Montenegro 630 782 Congo 130 1995 El Salvador 189 405 N.Macedonia 3682 3943 Djibouti 19 97 Guatemala 110 215 THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES Serbia 8835 11538 DR Congo 237 4584 Haiti 39 630 Eastern Europe Equ. Guinea 42 346 Honduras 451 1054 Armenia 1196 4498 Eritrea 176 5146 Jamaica 682 1280 Azerbaijan 638 1341 Ethiopia 172 1200 Nicaragua 264 672 Belarus 1655 2135 Gabon 60 723 Panama 35 50 Georgia 3031 6609 Kenya 333 654 Saint Lucia 95 56 Moldova 2554 4029 Rwanda 51 433 Trinidad/T. 180 200 Russia 6155 10911 Sao Tomé/P. 10 103 East and South-­East Asia Ukraine 13645 18060 Somalia 402 7809 Cambodia 33 103 MENA Tanzania 185 299 China 5206 12658 Algeria 4650 21311 Uganda 271 636 Indonesia 303 515 Egypt 1783 6638 Region South Africa Laos 16 37 Iraq 5916 20482 Angola 250 1451 Malaysia 903 734 Jordan 285 375 Botswana 66 50 Mongolia 1046 1831 Kuwait 215 293 Comoros 54 740 Myanmar 53 659 Lebanon 530 1697 Madagascar 65 350 North Korea 17 138 | (Continues)       17 TA B L E A 2   (Continued) | avg. avg. Removal avg. avg. Removal avg. avg. Removal 18      Country Returned Orders Country Returned Orders Country Returned Orders Libya 643 2003 Malawi 211 340 Philippines 1412 1923 Morocco 11209 35725 Mauritius 510 687 Singapore 78 63 Oman 37 31 Mozambique 16 64 Taiwan 120 169 Palestine 279 4755 Namibia 93 78 Thailand 499 807 Saudi Arabia 411 205 South Africa 690 687 Vietnam 1920 4561 Sudan 265 3539 Zambia 105 146 Central and South Asia Syria 1405 15772 Zimbabwe 300 1590 Afghanistan 3466 30546 Tunisia 3083 14828 non-­EU OECD Bangladesh 2910 11273 U.A.E. 94 77 Australia 380 314 Bhutan 4 76 Yemen 74 276 Canada 564 413 India 8080 15779 West Africa Chile 534 903 Iran 1905 7855 Benin 63 514 Israel 382 465 Kazakhstan 505 769 Burkina Faso 76 720 Japan 197 427 Kyrgyzstan 395 606 Cabo Verde 107 1269 Mexico 399 490 Nepal 840 1524 Côte d'Ivoire 250 4091 New Zealand 135 113 Pakistan 7433 24358 Gambia 370 2875 South Korea 265 393 Sri Lanka 1360 4081 Ghana 1231 4295 Turkey 3624 10795 Tajikistan 235 320 Guinea 341 5515 United States 2356 1661 Turkmenistan 42 156 Guinea-­Bis. 80 1265 South America Uzbekistan 424 796 Liberia 50 460 Argentina 512 1060 Mali 424 5556 Bolivia 1430 5893 Mauritania 128 1696 Brazil 5042 10075 Niger 44 428 Colombia 1124 2962 Nigeria 4188 13773 Ecuador 730 2089 (Continues) STUTZ and TRAUNER TA B L E A 2   (Continued) avg. avg. Removal avg. avg. Removal avg. avg. Removal THE EU'S ‘RETURN RATE’ WITH THIRD COUNTRIES Country Returned Orders Country Returned Orders Country Returned Orders Senegal 829 6945 Guyana 47 67 Sierra Leone 160 1114 Paraguay 605 2522 Togo 70 740 Peru 361 1195 Suriname 367 838 Uruguay 172 359 Venezuela 410 1114 Note: Only the countries with an average of more than 25 removal orders per year are included. Own calculations based on Eurostat data. |       19