Skip to content

The Statistics Are Coming (so here are some caveats)

September 10, 2014

Over the next few weeks I will provide statistical data and analysis about UNHCR RSD in 2012 and 2013. But before we get to that, let me provide some general notes of caution about the data.

The data I will be presenting come from two UNHCR reports, the 2012 Statistical Yearbook and Global Trends 2013 (the data I use are are in the statistical annexes, which you can download here). These are the best reports we have about what is going on in RSD globally, both with governments and UNHCR. I take the data and re-analyze it so as to produce a clearer picture about UNHCR RSD.

But the data isn’t perfect. For starters, UNHCR gets government RSD data mainly from governments themselves, and they might make a mistake or two. Second, even assuming there are no basic errors in the data, there’s no standard agreement on how to categorize the data. Even UNHCR offices are inconsistent. And this creates some messiness.

Person v. Cases

One of the things I am most interested in is how many people apply to individual RSD systems. But getting this number is not as simple as it seems. For one, there is a difference between reporting the number of persons who apply versus reporting the number of cases, since a single case could include multiple family members. UNHCR has improved its reports by listing whether the data from a certain country is reporting persons or cases. But that simply makes the inconsistency more transparent.

Most UNHCR offices report persons, but one (South Sudan) reports cases. Now, South Sudan is a very small RSD operation, so this does not throw things off too much. But there are a number of governments that report cases, not persons. If we knew with confidence the average number of persons per case in each country, we could correct for this. But I don’t know that data, and my intuition is that it is likely to vary according to differing migration patters and different RSD systems. The nut of this is that the aggregate government RSD data is likely an under-count, but I can’t say by how much.

New Applications v. Appeals

Another problem is that RSD systems typically have first instance and appeal levels (they should, anyway). But if we want to look at recognition rates or the number of new applications, we don’t want these mixed together. Unfortunately, a lot of governments and UNHCR offices do mix them together. This means that for some large UNHCR RSD offices (including Egypt, Jordan, and Pakistan), the data includes both first instance and appeals cases. This is also true for some governments, such as Austria, Burundi, Ecuador, Israel,  and Montenegro, among others.

This presents me with two unappealing choices. I could exclude these countries from the data, but then we would be completely ignoring some very significant RSD systems. Or, I could include them, knowing that I am in effect double counting applications (i.e when a person appeals, they will be counted twice). I have chosen the latter, since I would rather over-count something important than ignore it. But understand that this is an error either way.

Oh, the USA

Let’s just agree that the data reported for the United States is rough. It includes both an over-count (or two) and an under-count. We could assume that they cancel each other out, but you know what folks say about assumptions (and they would be right).

A person can start an asylum application in the United States either through the Asylum Office or in Immigration Court, which is part of the Executive Office of Immigration Review (EOIR). But if they start at the Asylum Office and lose they are usually referred to the Immigration Court, where they can  start their application again. But some people start their asylum cases in Immigration Court, skipping the Asylum Office part.

UNHCR reports two lines of data for the United States. One is for the Asylum Office, and the other for EOIR. But the EOIR data will naturally include both brand new applications, as well as people referred by the Asylum Office — creating an over-count.

Also, decisions by the Immigration Court can be appealed to the Board of Immigration Appeals, which is also part of EOIR. Thus, the EOIR data in UNHCR’s report might include appeals to the Board of Immigration Appeals; it is not clear if it does. So, that’s an additional possible over-count.

To make this even messier, the data from the EOIR is for persons, but for the Asylum Office it’s cases. Thus, we have an undercount.

What good are messy statistics?

In my opinion, it’s better to have a messy, somewhat distorted picture of what is happening in RSD than to leave it completely in a black box. But it is essential to know that the data is messy, and that what we are getting it is a rough snapshot only. It’s a bit like the famous map of the London Tube. It’s not to scale, and it is distorted in certain ways. But it’s still extremely useful in portraying the entire system in one comprehensible image, and it can get you where you want to go. I’m hoping for something similar with the data I am about to report, but it is essential that everyone understand what we are looking at here. It’s not perfect, but it’s the best we have.

  1. Jeff Crisp permalink
    September 11, 2014 3:10 am

    Good to see you blogging again Michael, but we are still waiting for your promised analysis of the review of UNHCR’s efforts to transfer RSD responsibilities to states. Hope we can see it soon.

    • September 11, 2014 3:11 am

      Yes, yes. Good things come to those who wait. But I am doing the stats first because they provide a useful context for other issues.

Comments are closed.

%d bloggers like this: