Skip to content

Where do our numbers come from?

September 25, 2006

One of the challenges in promoting constructive discussion of UNHCR’s refugee status determination practices is simple lack of information. UNHCR does not publish comprehensive reports about its RSD, and until recently the entire issue was rarely mentioned. Independent information is spotty because rights-based advocacy organizations concerned with the issue exist in just a handful of countries, although their numbers are growing. RSDWatch was founded in part to collect and focus the available data in one place.

One part of the RSD story can be told through numbers. Although they can’t tell us everything, statistics can at least tell us where UNHCR is conducting RSD, approximately how many cases are at stake, and they can show trends.

UNHCR publishes a general statistical portrait of its RSD activities every year, but it is buried within much larger pool of data in annual reports. Each June, UNHCR publishes provisional global statistics about refugees and asylum-seekers around the world for the previous calendar year. At the same time, UNHCR publishes a Statistical Yearbook for the year that ended 18 month earlier. In other words, in June 2005, UNHCR published provisional data for 2004, and a final Statistical Yearbook for 2003 (revising the provisional data published the previous June).

In these reports, UNHCR provides a global portrait on refugee status determination, and its statistical tables distinguish countries where governments perform RSD from those where UNHCR makes refugee status decisions. In order to provide a portrait of just UNHCR RSD, RSDWatch enters data from these tables into a spreadsheet, which is then used to produce the information you see on the website.


The main variable that RSDWatch focuses on is the number of people who are subject to UNHCR RSD, in other words, the number of new cases submitted for the first time to a UNHCR office that conducts RSD. This is a relatively simple index of how “big” UNHCR status determination is in any given place.

Until September 2006, RSDWatch listed its statistical data as numbers of applications, not people, in reliance on the way the data is presented in public UNHCR statistical documents. A single application can include multiple family members; if there are 10 applications in country X, there might be 25 people affected, depending on family size. However, in September 2006 UNHCR informed RSDWatch that the figures actually correspond to people, with an average of approximately 2.39 people included in each individual application. RSDWatch’s statistical pages for 2005 onwards have been corrected accordingly.

The number of new applications submitted in a given year is different from the number of pending cases waiting to be decided, and it is different from the actual number of decisions. We try to record only new applications for each year, but one ambiguity in UNHCR statistical reports is that first instance applications are not always distinguished from appeals. Whenever UNHCR separates them, we count only the first instance cases. But if they are not separated, we count whatever number UNHCR publishes.

We occasionally report the recognition rate at a UNHCR office. This rate is reached by dividing the number of recognized (accepted) cases in a given year by the number of decisions (not applications) during the same period.


The statistics presented on RSDWatch should be seen as the best available portrait UNHCR RSD around the world. For a number of reasons, our statistics are better at showing the forest as a whole than each individual tree, so to speak. There are a number of reasons for this caution.

First, raw numbers can’t show the human factors of RSD — the profile of the asylum-seekers in a particular place, the attitudes and training of decision-makers, the services available from civil society, the attitude of the host society and government toward refugees, the political climate, etc. These factors determine the fates of most refugees, but they often can only be observed up close. For instance, we do not emphasize recognition rates because they are difficult to interpret without separate knowledge of the type of applications in a particular country. A 5 percent recognition rate looks low, but if the refugee claims are generally weak it may be quite reasonable. A 75 percent rate looks generous, but it may be quite strict if all of the asylum-seekers are fleeing genuine persecution.

Second, our data is only as good as UNHCR’s data. We know, for example, that UNHCR says that it conducted RSD in 80 countries in 2004, but its provisional statistics listed only 56 countries. Harder to see, not all UNHCR offices are equally consistent in compiling and reporting their statistics, which means that numbers from some countries may be more accurate than others. For instance, if a UNHCR office double counts appeals as new applications, then UNHCR’s RSD operation in that country will appear relatively larger in terms of the number of applications than it really is. (Likewise, governments also do not necessarily report their data in the same way, which makes it a challenge for UNHCR to compile all of this information in a single report.)

Third, RSDWatch is human. From entering UNHCR data in a spreadsheet to analyzing it to writing explanatory texts, there’s lots of room for error, much as we try to avoid it. We ask our readers to e-mail us if you think you have found a mistake. The goal is to provide accurate, reliable and insightful information.

Comments are closed.

%d bloggers like this: