Zero, Some, or Zero-Sum: Exploring Trade-Offs in Identifying Human Trafficking Among Migration Flows

Conferences CoMSES 2017
1

Presenter: Kyle Ballard

Part of the Session on Migration

Models

2

Thanks for the presentation. It is interesting how you approach such a complex topic. There are of course many ways to extend such a model. I wondered whether you thought about bias of the “sampling”. Now it seems that there are equal probabilities for each type of agent to be detected, but agents with certain traits are more likely to be checked. If there was agency in trafficking a bias in checking agents with attribute X could lead to traffickers to recruit more agents without attribute X. In fact the model might be considered a kind of predator prey model, with two types of predators. Just a suggestion.

3

Thanks so much for taking a look and for your comments! One of the biggest flaws in the model is that attributes are distributed evenly (albeit randomly) across the population, as are “victims.” The previous model mentioned in the presentation was much more dynamic in terms of how attributed shifted as timesteps advanced. However, neither model includes trafficker agents, which is a great suggestion. Traffickers do evolve along with enforcement, and that dynamic would be great to explore. I just might extend this model in that way. I will have to integrate attributes that are more likely to manifest in victims, as that does not exist in this model; victims in this model simply draw from the same atributes at the same rate as other agents . A set of victim profiles would be needed, and there is existing research to draw from. My hope with this very simple model was to show how difficult it is to balance immigration policies and anti-trafficking efforts even in ideal circumstances… in this case ideal means victims cooperate when engaged and officials have the capacity and will to seek victims out. The results suggest some immigration policies might yield better anti-trafficking results, but of course many more factors must be considered.

4

I will also note another major element I abstracted away in this model: the model does not account for the effort required to enforce a given immigration policy. That is to say, enforcement officials accept or deny a migrant according to the policy and never have to expend resources to seek undocumented immigrants. These resources could very well detract from anti-trafficking efforts and would also factor heavily into the behaviors and decisions of government officials, migrants, and trafficker agents. I would have to factor these tradeoffs as well as the impact they would have on a more dynamic model that has traffickers and officials evolving based on their interactions. This opens up a whole host of new emergent qualities for exploration.