Playing the Numbers: The spurious promise of global trafficking statistics

‘Playing the numbers,’ ‘the numbers game,’ ‘the policy racket’: for those unfamiliar with American illegal lotteries and some of the legendary gangsters like Bumpy Johnson and Dutch Schultz that turned them into a major revenue stream for organised crime that still flourishes today, the concept was simple. The odds were disproportionately long, but poor people could bet very small amounts. It was, as they used to say, ‘a mugs game’; the organisers did well, while the poor lived on hope.

The quest for global trafficking prevalence numbers strikes me as a similar triumph of hope over experience. It is all too easy to succumb to what Merry has recently called the 'seductions of quantification'. 1 Previously, I have analysed how 'information' on human trafficking comes to be constructed as 'data', and how those data do--or, more often, do not--inform policy. 2 I said at the time that 'the trafficking field is best characterised as one of numerical certainty and statistical doubt. Trafficking numbers provide the false precision of quantification, while lacking any of the supports of statistical rigor.' 3 In fact, little has changed. While ambitious (and, some might say, pretentious) projects like the Global Slavery Index (GSI) have appeared, claiming to produce a methodologically rigorous estimate of the total number of 'slaves' in the world, much of the rigour proves illusory on closer examination. 4 Anne Gallagher, for example, has provided a detailed critique of the shortcomings of the GSI 2016, both conceptually and methodologically. 5 She has also highlighted the reluctance of many of our colleagues to express in public the pointed critiques of the GSI that they make in private, questioning whether this might not be related to a reluctance to alienate a potential donor.
There is insufficient room here to address all of my own reservations regarding the anomalous approach to data extrapolation used by the GSI, or to analyse the incongruities in country rankings that result from their approach. I would note, however, that while the conflation of 'trafficking' with 'slavery' may be rhetorically appealing, it diminishes conceptual clarity, exacerbating one of the key problems in the field. 6 The term 'slave' is certainly emotive, but devalues or elides the experience of numerous victims. 7 It is part of a flight from complexity that hampers the field through a one-size-fits-all diagnosis. While undoubtedly useful for fundraising, reducing-or inflating-every form of exploitation to 'slavery' makes any sort of useful measurement fundamentally unfeasible and useless as a basis for policy. Frederick Douglass, who knew more about slavery than most, said, 'No term is more abused, or misapplied, than that of Slavery.' 8 He also noted, 'It is common in this country to distinguish every bad thing by the name of slavery.' 9 I believe that he would recognise the situation today.
I have long advocated for an epidemiological approach to trafficking. 10 This could involve 'incidence' studies (e.g., number of new people trafficked over a given time period) or 'prevalence' studies (e.g., number of people ever trafficked, or over a specified period). These sorts of studies can be highly useful in specified contexts (sectoral, geographic, etc.), where the object of study is clearly defined and the universe of inquiry clear. They are even more reliable and useful when they can be tracked through time using a sentinel surveillance system (similar to the approach used for tracking HIV and AIDS).
However, the desire for a global prevalence estimate raises three issues: First, is it presently well done? As noted above, the answer to date is clearly negative. Second, if not, is it possible to do well-or at least better? While it may be possible to improve on current efforts, it is unlikely that it will ever be possible to arrive at estimates that are at all consistent, accurate, or useful; especially, given limitations of time, resources, and basic data quality. Finally, even if it were possible to do better, is it worth doing? What do we know if we have a number-35.8 million slaves in 2015 vs. 45.8 million in 2016, 20 million before that? Does this change our policy in any way? Does it alter consumer behaviour? Does it alter the behaviour of victims or exploiters?
Over a decade ago, Kangaspunta correctly assessed the situation: The question of whether global estimates of the scale of trafficking in humans serve any serious policy purposes should be posed. For other serious crimes, such as homicide, assault, or rape, global estimates are usually not given even though there are considerable problems with the data in some regions of the world. The global estimates on the numbers of persons involved in trafficking are always vague and cannot serve as a reliable knowledge base for policy planning. Thus it remains questionable whether this type of information is needed at all. 11 The same can be said today. The global numbers game may be slicker than the more traditional variety, but I bet Bumpy Johnson would have recognised it.
David A. Feingold is the director of the Ophidian Research Institute and Ophidian Films Ltd. In the past he developed the regional UNESCO HIV/AIDS and Trafficking Programme, which he coordinated for fifteen years at UNESCO, Bangkok. He has represented UNESCO at numerous international fora on human trafficking, including as the representative to the UN Interagency Coordination Group against Trafficking in Persons (ICAT), and has served on the management board of the UN Inter-Agency Project on Human Trafficking (UNIAP). A research anthropologist and filmmaker, he is a Fellow of the Royal Anthropological Institute, and a Fellow of the Centre d'Anthropologie de la Chine du Sud et de la Peninsule Indochinoise (CACSPI