Information Cascades and Refugee Crises
In recent times, it is not uncommon to hear about yet another refugee crisis. Wars such as ones affecting Bosnia, Rwanda, Kosovo, Afghanistan, Iraq, and South Sudan cause civilians to make life-or-death situations of staying in their home country or leaving everything behind and fleeing to other neighboring countries. Benjamin Laughlin investigates a theory using information cascades as a way to explain how sudden refugee crises arise.
A sudden large flow of refugees can occur since in warzones there is often scarce information for the civilians to base their decisions on. Therefore, it may be beneficial for them to base their decisions on whether other people are fleeing or not, and the assumption that those fleeing know something that they do not. In particular, it’s better to be safe and flee than to stay in possible danger since the stakes are so high–a wrong decision could result in death. This paper focuses on one war and the resulting flow of refugees from it. Laughlin constructed a geocoded dataset on the Kosovo war, that displays day to day data on refugee flows, violence against civilians, and the actions of armed groups. In the end, the model seems to be consistent with the theory of information cascade affecting the flow of refugees–Laughlin found that exposure to refugees from neighboring villages significantly increases the number of refugees that flee. In fact, it’s found that on average, every refugee that flees causes more than one other refugee to follow and flee.
In class we discussed the example of information cascades with drawing marbles from an urn with majority-red marbles. The first person to draw makes a decision solely on their draw. The second person uses both the first person’s draw and his own draw to make a decision–effectively evaluating two draws from the urn. If the first two person’s guesses are the same, then for the third person, regardless of what they actually see, they should go with the guesses from the previous people instead. Then, from the fourth person onwards, everyone else is in the same situation as the third person, and should go with the guesses of the previous people. Thus, an information cascade is started. Once many people start fleeing, it influences others to start fleeing as well since the pooled information is deemed better than the scarce private information an individual can have. One trait of an information cascade that is relevant to refugee crises is that it can be based on very little information. From the third person onwards, people ignore their private information. Only the first few people’s information matter, and if a cascade starts quickly in a country, most people’s private information isn’t being used. This also means theoretically, refugee cascades should also be easy to end if somewhere in the middle people receive better information and make decisions not to flee–this should result in less people fleeing than previously.