A favourite piece of research for Swedish model advocates throws up a few surprises

Guest post by @pastachips

“Does legalized prostitution increase human trafficking?“, a 2012 study by Cho, Dreher and Neumayer, is cited everywhere as evidence that ‘legalised prostitution’ increases trafficking into the sex industry. This article from last month is a recent example. The study’s conclusions have already been called into question, basically because the paper doesn’t distinguish between different meanings of the word trafficking, either in terms of taking account different countries’ laws – are we comparing like with like? – nor in terms of making a distinction between something-that-might-be-legally-trafficking-but is-essentially-undocumented-migration, and something more like cross-border kidnapping. Making that kind of distinction is pretty important if you’re trying to say something coherent! As is comparing like with like! The papers’ authors caution against treating its conclusions with too much weight, noting that “the quality of data is relatively low”, and that more research “will require the collection of more reliable data to establish firmer conclusions” (p26), but I think their data is actually way worse that they’re letting on.

They focus in on comparing Denmark, Germany and Sweden, and tell us: “in terms of human trafficking victims, the ILO estimated the stock of victims in Germany in 2004 to be approximately 32,800 – about 62 times more than in Sweden” (p25). I looked up their reference for that 32,800 figure, and found that the ILO paper cited as the source – Danailova-Trainor & Belser, 2006 – doesn’t even mention Germany. I discovered that by reading it, but you can also test it by clicking through and doing a command-f search for the words “German” or “Germany”, which you might reasonably expect to occur in a document that mentioned Germany.

The same 2006 paper is also cited as the source for the numbers on Denmark, where the claim is made that “… the ILO estimates the stock of human trafficking victims in Denmark in 2004 at approximately 2,250, while the estimated number in Sweden is about 500”, and the bracketed reference reads: “Global report data used in Danailova-Trainor and Belser, 2006” (p24). Again, I looked for those numbers (or any mention of Denmark) in vain in the Danailova-Trainor & Belser 2006 paper. I also checked out the “Global Report” mentioned as the source for Danailova-Trainor and Belser’s data, which I figured was probably a reference to the ILO’s 2005 report titled ‘A Global Alliance Against Forced Labour’ – there’s nothing else published that it plausibly could be; I checked. The 2005 Global Report doesn’t contain any country estimates, let alone numbers like those cited by Cho, Dreher and Neumayer.

I assumed that the 32,800 figure regarding Germany must have come from somewhere, so I dug around for ages on the ILO site, and found that the ILO seems to mostly resist giving country-specific numerical estimates (cf this 2005 report on trafficking in Germany, which really won’t be drawn on numbers). The only figures I could find for trafficking in Germany in 2004 were these official lists of identified victims that put the 2004 number at 972 (for all officially identified trafficking victims, not just sex trafficking victims). Obviously a list of ‘officially identified’ victims is unreliable – for instance, surely everyone knows these lists massively under-represent the number of men trafficked into agricultural work. But I feel like at least that list has an available methodology, which you could for instance replicate to see if you would get a similar number, or critique or challenge (as I would). How do you assess the validity of the process which produced an unreferenced 32,800, that appears to have come from nowhere?

It’s probably worth noting that I’m not invested in “defending” Germany’s record on ‘sex trafficking’ (scare-quoted because I think a lot of what is referred to in that phrase is more complicated that is generally allowed). I’ve focused in a bit on Germany because the study does. I don’t support Germany’s legal model in terms of sex work, and nor do any sex worker-led organisations in Ireland or the UK that I’m aware of; sex workers are perfectly capable of articulating why and how laws like those in Germany harm us, and disproportionately harm the more marginalised of us. I just feel like having references that go somewhere is quite a low bar in terms of the social sciences – especially when you’re citing very large numbers that apparently don’t appear anywhere else! – and I’m pretty surprised that the Cho/Dreher/Neumayer study seems not to clear that bar.

I also noticed that the study’s info on sex work laws around the world (167 countries! such big study wow!) is from, uh, 1995. (See p46.) The authors of the study are aware of some potential problems with this, noting: “for some countries, prostitution law changed during the 1996-2003 period: … Germany (2002), Denmark (1999) … Netherlands (2000), New Zealand (2003), and Sweden (1999). Our results are robust to the exclusion of these countries” [emphasis mine; some countries removed because if you want to see the full list you can follow up my reference] (p37). Norway implemented the sex purchase law in 2009; Iceland in 2007. In short, the effects of the ‘Nordic model’ are not actually included in the data of this study, and nor is the effect of the New Zealand model. “Our results are robust to the exclusion of these countries”. This might arguably make the study a not-totally-solid citation for you, if you’re looking to argue that the Swedish model is great and the New Zealand model is 💩. Here’s maybe the most interesting surprise. Cho, Dreher and Neumayer include a handy list of all the countries they’ve “looked at” (scare-quotes because hmmm), sorted into categories according to whether those countries have “very high”, “high”, “medium”, “low” or “very low” trafficking ‘inflows’. (See p44.) Sweden is listed in the “medium” category, along with … New Zealand.

Now, I don’t think that’s actually meaningful! Because I think that the data used to produce the conclusions of this study was 🌸 garbage 🌸. In general I think you get information that’s meaningful about sex work, trafficking, migration and exploitation by asking migrants who sell sex about their experiences and their policy suggestions. This study is a good example of that. But if you do think the conclusions of the study are meaningful – for example, if you’ve cited this paper as part of your argument in favour of laws like Sweden’s – then it should probably concern you that a study-you-apparently-consider-reliable ‘reveals’ Sweden is actually no ‘better’ at tackling trafficking than New Zealand. Whoops!

I know “the point” of this study is that it’s not “just about” individual countries; it’s trying to see patterns on a macro scale. But – that’s kind of a design problem with the study? In order to have relevance to policy debates, you have to organise your data in a way that is coherent with the terms of the debate – or at least, not egregiously incoherent. The global sex worker rights movement isn’t arguing for the (massively varied!) laws that this paper puts in the pile it calls “legalisation”; we’re not campaigning for “oh, laws that look something like the ones they have in Nevada, or Amsterdam, or Germany, or New Zealand; the details don’t matter, we don’t really mind”. We do mind! We’re trying to work towards (and improve on) the sex work laws they have in New Zealand. Sorry if this idea is complicated, but: aggregate data from Germany, Denmark, New Zealand and the Netherlands doesn’t make sense if no one is arguing in favour of the legal system in Germany, Denmark or the Netherlands, and when people who are pro-criminalisation refuse to understand this, they don’t derail us so much as make it obvious that they don’t care about the detail of the laws because they won’t be affected by them. Which isn’t that great an advocacy look, tbh. My focus on comparing New Zealand to Sweden in this study’s (broken and out-of-date) data isn’t because I’m scared of what will be ‘revealed’ by the aggregated global data (except in the sense that I’m finding this study scarily incompetent); but because I think it makes sense to talk about legal systems in a way that’s precise enough to be coherent.

Plenty of people have cited the Cho/Dreher/Neumayer paper as if it closes the argument, or as if it has some kind of weight or meaning. People who have been using it (presumably without reading it) probably need to decide whether the study – once read beyond the abstract – shows that New Zealand and Sweden have pretty much the same outcomes in terms of ‘sex trafficking’, or whether it’s actually so unreliable and badly put together as to be functionally useless (ding ding ding). While they decide which angle to take, I’m gonna write to the journal that published this paper and raise a few concerns.