New issues in refugee research

Statistically correct asylum data: prospects and limitations

Head of Unit Registration and Statistics Programme Coordination Section UNHCR
CP 2500, CH-1211 Geneva 2 Switzerland

E-mail: hovy@unhcr.ch

April 2001

These working papers provide a means for UNHCR staff, consultants, interns and associates to publish the preliminary results of their research on refugee- related issues. The papers do not represent the official views of UNHCR. They are also available online at http://www.unhcr.ch/refworld/pubs/pubon.htm

ISSN 1020-7473

Introduction

The subject of asylum seekers, traditionally of interest only to national and international refugee agencies, a handful of human rights lawyers and refugee advocates, has recently caught the attention of population statisticians and migration scholars. Significantly, the Revised UN Recommendations on International Migration Statistics2 devote, for the first time, explicit attention to refugees and asylum seekers.

This paper considers the issue of asylum statistics in Europe. The first part summarizes UNHCR’s role in the collection of asylum statistics, highlighting the organization’s experience in implementing the UN recommendations. The second part uses period-based asylum statistics to illustrate recent trends in asylum migration in Europe. In part three of the paper, the availability and usefulness of period-based statistics are considered by examining the various measures and indicators available to study the asylum process. A comparison is made between a period- and cohort-based analysis, using data from the United Kingdom. The paper concludes that while cohort-based data may provide some valuable insights in the asylum procedure, period-based information remains essential to monitor current international trends.

The international framework

Since its creation in 1950, UNHCR has collected, used and reported statistics on asylum seekers, refugees and other people of concern to the organization. Indeed, UNHCR’s Statute recognizes the key role of statistics: “The High Commissioner shall provide for the protection of refugees falling under the competence of his Office by… obtaining from Governments information concerning the number and conditions of refugees in their territories…”3

Coming under increased pressure to provide accurate and timely refugee figures, in the early 1990s UNHCR established a unit with specific responsibility for the collection, compilation, reporting and analysis of statistics.4 Since that time, the statistical function within UNHCR has been progressively professionalized by means of improved guidelines and training. The number of statistical publications produced by UNHCR has also risen sharply in recent years.5

UNHCR offices are located in more than 120 countries around the world and employ some 5,000 staff. Through its day-to-day contacts with official bodies as well as with the refugees themselves, UNHCR is in a unique position to collect and verify asylum and refugee statistics. In many developing countries, UNHCR is involved in refugee registration at the request of Governments. The UNHCR statistics are based on Government records for most industrialized countries, whereas UNHCR field offices

_______________________________________
1 This paper was orginally presented at the Joint ECE-EUROSTAT work session on Migration Statistics, Geneva, 8-10 May 2000.
2 Statistics Division, United Nations, New York, 1998, Statistical Papers Series M, No. 58, Rev. 1.
3 Statute of the Office of the United Nations High Commissioner for Refugees, Chapter II, Paragraph 8, General Assembly Resolution 428(V) of 14 December 195
4 For further discussion of UNHCR statistics, see: Jeff Crisp, “Who has counted the refugees? UNHCR and the politics of numbers”, New Issues in Refugee Research, Working Paper No. 12, June 1999.
5 Most statistical publications are available on http://www.unhcr.ch/statist/main.htm

are the source of much of the data from developing countries. Since 1950, UNHCR has compiled global refugee statistics. Since 1994, the Office has issued an annual statistical report.

The revised UN recommendations on international migration statistics

In 1998, the Statistics Division of the United Secretariat published the revised UN recommendations on international migration statistics. Recognizing the numerical importance of asylum flows during the 1990s, the 1998 Recommendations, unlike its predecessor, cover asylum and refugee statistics. UNHCR has been an important contributor to the chapter on asylum statistics.

UNHCR’s most recent Statistical Overview6 mirrors much of what is proposed in the UN Recommendations (see for instance page 78 of the Recommendations). The 1998 Overview provides detailed information on the number of applications, the type of decision (positive, negative, otherwise closed), pending cases and recognition rates by detailed origin for more than 130 asylum countries. Furthermore, it contains longitudinal (10-year) information on asylum for most industrialized countries. Whereas data on asylum applications, type of decision and recognition rates are available for virtually all asylum countries, the coverage of pending cases is less.

The UN recommendations also suggest distinguishing between first instance and appeal procedures. This distinction was systematically introduced during the 1999 round of UNHCR’s global data collection. The 1999 draft tables have a column stating “Type” which indicates (a) the actor responsible for the status determination process (Government or UNHCR) and (b) the administrative level of the procedure. It is expected that the coverage of this latter information will improve in following years.

UNHCR’s 1999 draft tables contain four columns of recognition rates. Whereas the 1998 Statistical Overview used only one denominator (the total number of asylum decisions), the 1999 draft tables uses two (total number of asylum decisions and total number of substantive decisions). The last two columns of the 1999 standard tabulation (“Excl. o/w cl.”) show the Convention and total recognition rates based on the UN Recommendation, that is, a period indicator based on the number of substantive decisions.

Data elements suggested in the UN Recommendations but which prove more difficult to obtain are:

• A systematic breakdown between the number of cases and persons by type of decision (UN Recommendations, table 10 through 13);

• A systematic breakdown between “newly arrived” and “from within country” by type of decision (UN Recommendations, table 12);

_______________________________________
6 Refugees and Others of Concern to UNHCR, 1998 Statistical Overview, Geneva, July 1999 (http:///www.unhcr.ch/statist/main.htm>).

• Information on “stay of deportation”. Information on rejected cases, which are granted stay of deportation, is obviously highly relevant. After all, asylum-seekers whose formal claim for refugee status have been denied may still not be able to return to their own country for refugee-like reasons. From the perspective of UNHCR, what matters it is not so much the granting of official refugee status, but whether any person in need of protection is not sent back against his or her own will. The most fundamental principle in international refugee protection is not the granting of asylum (there is no such right), but protection against “non- refoulement”. Although the data on rejected cases that are not deported are thus highly relevant, few countries produce such reports. And if they do, the extremely short-term nature of process (a removal may be postponed for 1 or 3 months for instance), makes any international comparison less feasible.

As more detail is being collected on asylum procedures, the differences between national procedures become more apparent. One of such differences is the way in which asylum claims are dealt with which have no chance in the actual determination process (“manifestly unfounded claims”). Whereas some countries reject such cases during a pre-screening process (Belgium and Canada for instance), other countries (e.g. United Kingdom) distinguish refusals in those after a full hearing and those on the basis of formal grounds. Clearly, these different asylum processes have a significant effect on some indicators. In the case of Belgium and Canada, does the number of applications concern those submitted to the eligibility procedure or only those which were given a full hearing? Nevertheless, calculating recognition rates on the basis of the number of substantive decisions only, as recommended by the UN, has the advantage that the rates are not “polluted” by formal rejections made either during the pre-screening or at the level of the actual determination process.

Information on the gender and age of asylum-seekers and refugees, as proposed by the UN Recommendations, has become increasingly available in UNHCR statistical publications.7 Whereas prior to 1999, the gender and age was only collected on refugees directly assisted by UNHCR, since 1999 the coverage has been extended to asylum-seekers, refugees and all other groups of concern to UNHCR. Consequently, the coverage of gender and age on asylum-seekers and refugees in Europe has significantly improved in 1999. As noted elsewhere, however, one of the main constraints in providing a genuinely comprehensive coverage of gender and age remains the inability of many industrialized countries to provide this information from their registration systems.

_____________________________________

Using asylum statistics during emergencies: the Kosovo crisis

The 1999 Kosovo refugee crisis provides an excellent framework to consider the relevance and use of asylum statistics. In which indicators were policy makers most interested in and what was feasible to provide? Not surprisingly, the indicator drawing most attention was the number of new applications submitted. In addition to reporting on prima facie refugee arrivals in countries in the region, UNHCR started a monthly reporting system on Kosovar asylum applications covering some 24 European countries, a system that was subsequently expanded to cover all nationalities on a monthly basis.8 Weekly asylum applications, however, appeared not feasible.

On several occasions, tables were prepared showing not only applications, but also decisions and pending cases concerning Kosovar asylum-seekers. The number of pending cases appeared an extremely useful indicator as it showed that most Governments, overwhelmed by the number of Kosovar asylum applications, were not able to determine most claims in a speedy fashion. The asylum-seeker stock estimate of pending cases has at least two more advantages. First, it provides a directly comparable indicator to the prima facie refugee stock in the countries neighbouring Kosovo. Second, the information is much more easily understood by the general public as it provides an answer to the question “How many Kosovar asylum-seekers are there?” Unfortunately, only a few countries were able to report regularly on the number of FRY applications pending in the procedure.

One of the main data limitations during the Kosovo crisis was origin. As most asylum countries record only the nationality (country of citizenship) of the applicant, few countries were able to distinguish Kosovar asylum-seekers from other citizens of the Federal Republic of Yugoslavia (FRY). Although UNHCR was able to estimate the number of Kosovar applications among the total number of FRY asylum-seekers submitted during the Kosovo crisis (almost 90%), information on the province of previous residence (Kosovo) as well on the “ethnic origin” (Kosovar) would have in fact been required. Whereas citizenship is generally a sufficient proxy for origin, the Kosovo crisis showed the need for more detail. (The same applies for other groups such as Kurds and gypsies). In view of the increased emphasis European states put on “internal flight alternative”, i.e. the possibility to find refuge within the country of origin in areas considered “safe”, the importance of sub-national information on origin is likely to increase.

A second problem was related to the actual asylum decision process. Most Kosovars whose claims were adjudicated were provided short-term residence status on the basis of humanitarian grounds. The short-term nature of some of these decisions lead to the situation that some cases, which had received a positive decision earlier during the year, were subsequently rejected.

____________________________
8 See for instance Asylum Applications in Europe, 1999. Trends in Monthly, Quarterly and Annual Applications by Country of Asylum and Origin, UNHCR, Geneva, 28 February 2000 (<http://www.unhcr.ch/statist/0002euro/text.htm>).

Recent asylum trends in asylum in Europe

Using period data, this section provides an overview of main asylum trends in Europe. During the period 1995-1999, the 25 European countries listed in table 1 received some 1.6 million asylum-seekers, 85 per cent of which were lodged in the European Union. During 1999, some 450,000 applications were lodged in Europe, 25 per cent more than in 1998. During the past five years, the largest relative increase was recorded in Eastern Europe as its share increased from 1.3 per cent in 1995 to 6 per cent in 1999.

In 1999, 28 per cent of all asylum applications originated from the Federal Republic of Yugoslavia. In Western Europe, Kosovars constituted 33 per cent all asylum- seekers, in Eastern Europe 25 per cent, in Northern Europe 17 per cent and in Southern Europe (excluding Italy) 6 per cent (see Box 1).

Some 200,000 asylum-seekers were granted Convention refugee status over the past five years in Europe. Although the 1999 decisions are not yet available for Austria and France, it appears that Convention recognition during 1999 in Europe will be considerably lower than during 1998. Table 2 indicates that, since 1995, the annual number of persons recognized under the 1951 UN Convention has fallen each year. This overall decline trend is mostly due to the sharp fall in recognition in Denmark, Germany and the Netherlands. Conversely, Eastern Europe, Northern Europe and Southern Europe all show an increase in persons granted Convention status. The United Kingdom registered the strongest increase in recognition under the Convention.

The number of asylum-seekers who were allowed to remain for humanitarian reasons in Europe during 1999 increased by some 50 per cent compared 1998 (Table 3). Among the main asylum countries, the largest increase was recorded in the United Kingdom (+240%) and Switzerland (+225%), whereas a sharp decrease was registered in Sweden (-55%) and the Netherlands (-35%).

Cohort-based asylum statistics

In the above analysis, the recognition rates were obtained by dividing the number of positive decisions made during year X by the total number of positive and negative decisions made during the same year. In this way, an indicator was obtained concerning the percentage positive decisions taken during a given period. Alternatively, a cohort-based approach looks at a given cohort of asylum-seekers and calculates the recognition rate for this entire cohort. “If all cases submitted over the course of a year are followed through to their completion, the (…) measure can be obtained on a cohort basis” (UN Recommendations, page 81). The interest in this approach is prompted by the significant “carry-over” of asylum decisions from one year to the other. If all applications would be decided during the year the application was made (i.e. no pending cases), one could simply divide the number of positive decisions by the number of applications to obtain the recognition rate during a given year.

While it may be useful to know how many of the asylum-seekers who applied during year X were eventually recognized (or rejected), the policy relevance of this indicator seems limited as the calculation can only be made once all applications from a given year have been decided. As illustrated below, it may take many years before an entire cohort of asylum applications receives a decision.

The length of the asylum procedure

While the interest of “cohort-based” recognition rates may thus be more of an academic nature, cohort-based information becomes absolute key in determining the length of asylum procedure – a critical indicator for the efficiency of the refugee status determination process. Speedy procedures are in the interest of both governments and of genuine refugees. As such, the recent increase in the number of pending cases in a number of countries is a worrying trend and of direct concern to all involved. The example below demonstrates that, whereas the number of cases pending in the procedure is an important indicator, one can only determine the actual length of the procedure by linking the case to the year the application was made.

Comparing period and cohort information: the case of the United Kingdom

The UK has been chosen as it is one of the very few countries which have published the data required for a period and cohort-based analysis. As indicated in Table 7, the UK developed a backlog of some 40,000 decisions in the early 1990s. Despite the fact that in subsequent years the number of decisions sometimes surpassed the number of applications, the situation has, up to this day, not yet been normalized. In fact, due to a sharp increase in applications during 1999, the number of pending increased to some 103,000 by the end of 1999.

Table 8 shows the distribution of asylum decision by the year the application was made. One-quarter or less of all applications lodged in any given year during 1990- 1997 received a decision during the same year (see also Table 9). The three columns of the right hand side of the table, calculated by the author, show the proportion of applications decided by the end of 1998 by year the application was made. For instance, some 20,300 applications lodged in 1990 had been decided by the end of 1998, some 77 per cent of the total number of applications lodged during 1990 (26,200).

Based on the crude assumption that both applications and decisions were distributed evenly throughout each year (no statistics are available by month), Table 10 shows that the average processing time for asylum applications in the UK in first instance was around 2 years for much of the 1990s. Although the situation somewhat improved in 1998 (1.9 compared to 2.4 years in 1997), the situation during 1999 will have undoubtedly worsened. The figures have not yet been published, however.

The granting of refugee status generally take the longest, whereas refusals (including rejections on formal grounds) need the least time in the United Kingdom (see Box 2). At the end of 1998, the average processing time for refugee status and exceptional leave to remain was some 3 years, whereas this was 1.5 years for refusals.

In Box 3, the cohort and period-based recognition rates for the UK have been summarized (see also Table 11 and 12). A number of observations can be made. First, recognition rates in the late 1980s were considerably higher than during the 1990s, a conclusion that can be drawn from the period and the cohort-based rates. Second, a positive relationship exists between recognition rates and the average processing time: the longer the processing time, the higher the rate. This observation cannot be derived from period-based calculations.

Evidence from Canada and Switzerland

Due to a lack of comparable data, it is difficult to ascertain whether the situation in other countries differs from the United Kingdom. However, data on total decisions by year of application from Canada and Switzerland indicate that the asylum processing in the UK is markedly slower than in the other two countries. In Switzerland, more than 60 per cent of the applications lodged in 1998 was decided during 1998, compared to 36 per cent in the UK and to only 23 per cent in Canada. (Note that the Canadian data refer only to the substantive procedure of the Immigration and Refugee Board. Furthermore, the data for the early nineties are difficult to compare with the later data due to a change in the procedure.) At the end of 1998, the estimated average processing time for asylum applications in Switzerland was one year, almost half the UK estimate, where the Canada it was estimated at 1.6 years. It should be stressed, however, that these are rough estimates. More precise information on the duration of the procedure can only established on the basis of monthly data, which were not available to the author. A second major problem for cohort analysis is the absence of data by nationality.

Discussion

The above analysis has shown some of the advantages and disadvantages associated with period- and cohort-based asylum data. First, in the calculation of recognition rates, a cohort approach does not seem to be particularly useful as these rates can only be calculated once all decisions have been taken on applications lodged during a given year. Cohort-based recognition rates for the more recent years are subject to significant change in the near future due to the high level of pending cases in Europe.

Cohort-based data is essential to determine the efficiency of asylum procedures. Unfortunately, very few countries are able to provide the information on the type of decision by month or year of application and by nationality.

The above analysis has shown that period-based recognition rates allow for an instant comparison between years and countries as soon as the year is over. [In this context, it should be recalled that period-based recognition rates measure something else than cohort-based rates.] Furthermore, their much wider availability guarantees a much wider coverage, an important consideration when compiling internationally comparable data. The period-based pending cases at the end of the period have proven a strong indicator of (in -)efficiencies in the asylum procedure.