The EU report on trafficking in human beings

Last month saw the publication of the EU’s first Trafficking in Human Beings report, which is billed as an attempt to “collect comparable data on trafficking in human beings at the EU level”. The report is (properly) littered with disclaimers, such as this one in screaming bold type on page 30:

More reported cases do not necessarily mean an increase in the actual number of victims. This may indicate an improvement in the reporting rate of the phenomenon or a change in the recording system

Unfortunately, however, when it came to the press release, the European Commission decided to go for the handy soundbite – and so we’ve been deluged with headlines like “Human trafficking increased by 18%” when of course, the report doesn’t show it did any such thing.

For all the faith put into that 18% figure, though, most of the newspaper and NGO reactions don’t evince much interest in the rest of the data. I’m going to put this down to simple laziness – it’s a whole 94 pages they’d have to wade through, poor dears – but it’s also the case that if all the statistics were accepted as readily as the “18% increase” has been, it would be a little bit inconvenient for some of those with an interest in this area. This table on page 31 shows why:

Contrary to what we’re constantly told by the anti-trafficking movement, the most recent figures make it hard to discern any link between trafficking and the legal status of sex work. The Dutch rate is very high, but the Cyprus rate is higher – and Cyprus has much stricter laws than the Netherlands (brothels are illegal, for one thing). Romania, where sex work itself is illegal, is nearly as high. Hungary (legal), Portugal (legal) and Lithuania (illegal) are tied for last. Austria and Germany are also relatively low – in fact, Germany and Sweden are tied, at 0.8 per 100,000.  And the German rate has remained more or less constant over the three years surveyed, while Sweden’s has quadrupled.

Don’t think we’ll be reading that in the next Turn Off the Red Light press release.

I’m being mischievous, of course, because as I’ve already said, the data don’t – and can’t – show the actual number of trafficked persons in each country. So we can’t really say that Germany, Austria and Sweden have more or less the same trafficking rate. But let’s be honest here – had the table shown Germany and Austria to be right up there with the Netherlands, and Sweden way down on its own, is there any doubt we’d have heard all about it from the prohibitionist groups?

But sex workers’ rights advocates shouldn’t leap on those figures, either, because truthfully the whole report is pretty hopelessly undermined by its methodological weaknesses. These include the following:

Different countries provided different sets of data. On page 30 we are told:

Ten EU Member States provided data on identified and presumed victims, 15 EU Member States provided data on identified victims only and two EU Member States provided data on presumed victims only. The participating non-EU countries provided data only on identified victims.

This alone would make it impossible to get a cohesive account of trafficking, or even just of detected trafficking, across Europe. It certainly renders a comprehensive cross-country comparison untenable. (In case you were wondering, though, all the countries I’ve mentioned so far provided data only on “identified victims” – except for the Netherlands, which only gave “presumed victim” data. More about this later.)

Different countries use different definitions. The Commission seems to have attempted to get around this, by reminding countries of the international definition of “trafficking” when sending them the forms to complete, but the methodology problem is unavoidable. An “identified victim” is, according to page 22,

a person who has been formally identified as a victim of trafficking in human beings according to the relevant formal authority in Member States

and that inevitably introduces an element of subjectivity into the data, because not all member states strictly apply the international definition when deciding whom to identify as a victim of trafficking. In fact, I doubt whether any of them do – and they are inconsistent in the manner and degree by which they diverge from the international definition. So when we look at that table and see that Ireland has more than twice the rate of “identified victims” as Greece, for example, we have no way of knowing whether that’s because Irish police genuinely detected twice the rate of people who fit the international definition, or whether the Irish police identified people outside the international definition, or whether the Greek police declined to identify people inside it.

This problem is magnified for those countries that include data for “presumed victims”, that is:

a person who has met the criteria of EU regulations and international Conventions but has not been formally identified by the relevant authorities (police) as a trafficking victim or who has declined to be formally or legally identified as trafficked.

Here, member states are essentially being asked to submit data on people who were trafficked but were not identified as trafficked. To identify unidentified trafficking victims. There’s something a bit Schrödinger’s Cat-like about this category; it just doesn’t seem to lend itself to any real scientific measurement, at all.

If the data had been collected on a forward-looking basis, it might almost be workable. That is, if countries had been told that, for the future, they would be asked to record details of not only the people who they formally identified as victims of trafficking, but also of people who they would have identified as trafficked but for some insurmountable obstacle (such as that the person themselves told them where to stick their “trafficking victim” status). You’d still have the subjectivity problem, but at least you would know that the records were being kept on an ongoing basis, with the member states knowing exactly what they were to look out for.

But that’s not how this was done. Instead, as explained on page 17,

The questionnaire was sent via Eurostat to the National Statistical Offices of the EU Member States, EU Candidate and Potential Candidate countries and to the EFTA/EEA (European Free Trade Association/European Economic Association) countries in September 2011. It included the appropriate tables, a list of common indicators, definitions and guidelines for collecting the statistical data as well as the country codes to be used and a template for providing metadata.

So countries were asked to collate retrospectively their data on people who they had identified as unidentified trafficked persons. And then to pick out of that data only the people who fit the “common indicators, definitions and guidelines”. But who’s to say that the data were initially recorded in such a way as to make that possible? If the authorities weren’t already aware they were going to be asked to provide data on people who they’d declined to identify as “trafficked”, you have to wonder whether they’d be quite so diligent in their record keeping about those people.

Did I say “authorities”? Oh – here’s the next problem:

“Presumed victim” statistics were supplied by a wide variety of sources. According to page 23:

Data on ‘presumed’ victims on trafficking in human beings may be available from national rapporteurs (or equivalent mechanisms which tend to act as national coordinating bodies), victims assistance services, immigration services, labour inspections and border guards.

I don’t think I can really overstate this point: there is no possible way to get anything like a reliable, consistent overview when you throw out to your data collectors that they can get their information from pretty much anywhere they can find it. And as it happens, some of the key data sources are pretty questionable. For example, in the Netherlands the source is CoMensha, a national agency, which according to page 39:

does not have a formal assessment based on specific criteria by which the registered person’s status as a victim can be verified.

In other words – as I discussed in this post – CoMensha simply records the alleged cases that are referred to them, without actually investigating whether there’s any substance to the allegations. The Dutch figures in this report, therefore, are based on little more than rumour. No wonder they’re so high.

Also problematic is the Latvian information, which according to page 38 is

provided by the NGO Resource Centre for Women “Marta”, which provides assistance for presumed victims using alternative financing.

Marta is a prohibitionist organisation, and one that therefore has an interest in finding trafficking victims. In fairness to them, there’s nothing in the data to suggest they’re inflating the numbers; Latvia’s overall rate is the same as Germany’s and Sweden’s and only a minority of these fall into the “presumed victim” category. Nonetheless, their objectivity as a source has to be doubted.

To take a final example, page 39 states that Finland’s number of presumed victims

includes all cases who were directed to reception centers on the basis of suspicion of human trafficking.

Does that sound like a rigorous effort to make sure only the cases that fit the Commission’s specified criteria were reported? Not to me it doesn’t.

I could go on, but I think I’ve made the point. There’s just too much variation across countries, and too loose standards within countries, to consider this a reliable measure of the amount of trafficking in Europe. The report is of (limited) interest for what it reveals about the countries’ data collection processes, but overall you really get the sense of this being a box-ticking exercise for Brussels bureaucrats more than anything else.

I’ll just finish on a couple other matters I found worthy of comment. Firstly, on page 24 it states:

Data is to be collected on the total number of female and male victims

The binary (and I’m guessing cis-centric) approach is unfortunate, as even though the figures aren’t reliable it would still be interesting to see how many trans* persons were recorded. But the only reference I can find within any of the country data is in the British records, on page 39, where it says that one adult trans* woman was identified. Unfortunately, it manages to say this in a grossly offensive way.

There are also figures on “suspected traffickers” and “prosecuted traffickers”. It’s not clear exactly what threshold was required for the first category, but one thing I found interesting was that non-EU citizens accounted for 55% of the former but only 24% of the latter. The reasons for the discrepancy aren’t explained. It could be that non-EU citizens are more likely to go missing before they can be prosecuted, of course, but there could also be a greater tendency to suspect them without any real proof. This is something that merits further research.

And finally, a word about citizenship and labour market status. On page 52 there’s a table (Table 6) that breaks down all the identified and presumed victims, by year and by citizenship. So I did a little numbers exercise. First I counted all the identified victims by country (I didn’t count the presumed victims, because for all the reasons explained above, I think those numbers are too messy to be of any real use). Then I subtracted the internal trafficking victims (as listed on page 53, Table 7). And then I totalled up the remainder, put them into geographical categories and worked out the percentages of each.

The disclaimers first. The numbers don’t quite add up, because I came up with a grand total of 12,261 identified victims in Table 6, whereas the Table 2 combined figure for identified victims is 13,424. I’m not sure where that missing 1,163 went. There also seems to be a minor discrepancy in the Table 7 figures, since I end up with –2 cross-border EU-15 citizen victims in 2009. I couldn’t subtract the child victims, as there’s no breakdown by “identified” and “presumed” nor any indication of how many were working-age children. And of course, all the problems with the data I’ve already discussed still apply. But bearing all that in mind, I still think the results are stark enough to be worth pointing out. Of the identified victims of cross-border trafficking in the three year period:

• 37% were Romanian or Bulgarian
• 58% were from outside the EU/EEA
• 5% were from the 2004 accession countries
• <1% were from the EU-15

Thus, 95% were from countries barred from all or most of the EU labour market during the data collection period – and nearly everyone else was from a country barred from much of the EU labour market during the data collection period. This isn’t quite as dramatic as it seems, because it still only accounts for just over half the identified victims – the internal trafficking rate is surprisingly high, which is something else that merits investigation (though as the number is very high in some countries and negligible in others, I suspect it mainly comes down to definitions). Notwithstanding those limitations, though, I think it’s safe to conclude that cross-border migrants in the EU are far, far, far more likely to experience trafficking if they don’t have labour market access in the country they’re migrating to.

You’d think this would be a no-brainer, wouldn’t you? But EU states just don’t seem to get it. They consistently wring their hands about the trafficking problem, but stubbornly insist they just can’t relax their labour migration policies – without seeming to make the connection between the two. At some point they’re going to have to admit that if they really want to address the issue, they’ll have to stop making it all but impossible for accession state citizens and non-EU citizens to legally work in their countries. Or else admit that they aren’t really that bothered about trafficking after all.