What new evidence from Somalia tells us about when civilians decided to flee war zones

The massive scale of the refugee crisis driven by five years of war in Syria highlights the urgency of the much broader challenge posed by population displacement to the international system. Global displacement has now reached a higher level than it has been since World War II. An effective response requires understanding not just why populations become displaced but also when. The experience of Somalia suggests that the expectation of changes of balance of power on the ground, not the actual territorial gains or the entry or exit of foreign powers, is the crucial trigger for population displacement.

There are actually very different types of refugee flows, which require quite different policy responses. Sometimes displacement occurs in massive floods, like when more than 200,000 people crossed from Rwanda into Tanzania within a 24-hour period  April 29, 1994. In other cases, we see the nature of refugee flows change over time. For instance, Syria saw displacement in relatively small trickles in 2011 and part of 2012, but then a massive flood of refugees later in the conflict.

What factors influence the timing of displacement? Why do certain crises prompt displacement floods while others only elicit a trickle? Unique daily internal displacement data from Somalia in the mid-2000s, which I published in a recent article, can offer insights into these critical policy questions. Since Ethiopia’s 2006 invasion, Somalia has produced some of the world’s largest displacement flows. As Anna Lindley observes, two-thirds of Mogadishu’s population fled between the end of 2006 and the end of 2008. Somali displacement flows furthermore exhibit substantial variation over time. Its lessons may be more broadly applicable, since Somalia contains characteristics that exist in many cases of civil conflict: a weak state, proliferation of armed groups and militias, protracted conflict, poverty, and environmental challenges.

The data derives from an UNHCR project called the Population Movement Tracking system. Begun in mid-2006, the project works with 48 local partners inside Somalia to track displacement on a daily basis. Analyzing this daily displacement data reveals that there were actually 15 distinct cycles of displacement in Somalia from 2008 to 2013. The structural conflict characteristics of geographic scope and balance of power are the most important drivers of displacement timing – and not individual events, as is commonly believed.

There are two possible drivers of displacement timing in Somalia. The first possibility is that civilians can respond quickly to violence, so we need daily violence and displacement data that offer the opportunity to detect quick responses to individual events. A second possibility is that civilians consider a range of factors in their displacement decisions, thereby requiring broader changes in conflict dynamics to drive their displacement. These changes may involve the geographic scope of the conflict and the balance of power. But such simple explanations do not do justice to the complex reality of Somali displacement.

Increases in the geographic scope of a conflict can increase the size of displacement flows, because it increases the proportion of a country’s population affected by conflict. This finding is consistent with previous cross-national analysis from Erik Melander and Magnus Oberg. While this point may appear trivial, researchers have noted that conflict usually does not engulf the entire territory of a country. This makes it important to monitor the proportion of a territory where armed conflict is being waged.

Balance of power has more nuanced effects on the size of displacement. The crucial factor for displacement appears to be not the actual changes in balance of power but the effect of anticipated changes in the balance of power.

In fact, when change in the balance of power results from a change in territorial control, displacement decreases. This may seem counterintuitive. The key reason is that civilians can often anticipate changes in territorial control. If they believe that an armed group they oppose will take control of their town, then they are likely to leave before that change in territorial control occurs; but once territorial control changes, it may be dangerous for civilians to flee. Therefore, anticipated changes in the balance of power increase displacement, while actual changes in the balance of power decrease displacement.

On the other hand, when change in the balance of power results from the entrance or exit of a foreign intervener, displacement increases. Somalia experienced substantial increases in displacement when Ethiopia entered in 2006, and even when it quietly re-entered later in 2010. Kenya’s entrance in 2011 also increased displacement in Somalia. When Ethiopia withdrew its forces from Somalia in January 2009, there was also an increase in displacement as Al Shabaab and other groups escalated their attacks.

This research helps us understand how and when to anticipate high levels of displacement. Existing research examines the likelihood of displacement floods at the beginning of a conflict, or when civilians anticipate mass killing. However, for a more comprehensive explanation of why displacement floods may occur at any stage of a conflict, we must pay close attention to geographic scope and balance of power dynamics. Careful monitoring of these conflict dynamics may enable more effective responses to displacement crises around the world and help ease their staggering human costs.

Justin Schon is a PhD candidate in political science at Indiana University, Bloomington.