Association for Public Policy Analysis & Management

On December 26, 2004 a 9.2 earthquake with an epicenter just off the west coast of the Indonesian island of Sumatra spawned a tsunami that affected countries around the entire Indian Ocean. Aceh, on the northern end of Sumatra, was hardest hit. It is estimated that five percent of the population was killed, and that about 15 percent was displaced because of the massive destruction to housing, roads and infrastructure. Damage was estimated to exceed US$4.5 billion in 2004 dollars.

Using data from the Study of the Tsunami Aftermath and Recovery (STAR), a large-scale population-representative longitudinal survey, we address the complex interlinkages among post-disaster relocation decisions five years after one of the largest natural disasters in recent history. A limitation of many studies of post-disaster migration is that they tend to focus on documenting population movements during relatively short periods soon after the disaster, and amongst those most affected. Relatively few studies of post-disaster migration examine how determinants of migration change over time or across populations with different exposure to the disaster. Using a combination of household location data and detailed migration histories compiled into a five-year panel of post-tsunami moves we are able to address this limitation.

First, we compare individual migration patterns for two distinct population subgroups. We consider moves for people who were living (before the tsunami) in heavily damaged areas, distinguishing among those who stayed put despite the heavy damage, those who left their pre-tsunami communities and (for this group), considering who stays away and who eventually moves back to their original location. We then compare these migration patterns to those observed in areas where community-level impacts of the disaster were minimal. This analysis provides an overview of migration levels as a function of exposure to tsunami damage and of how migration changes over the five years after the tsunami.

Second we explore key correlates of short- and long-term migration decisions. Motivations behind migration are varied and can change over time. In the short-term some people may need to relocate to address basic needs for shelter, but as the reconstruction effort progresses many may choose to move back to their damaged communities, while others may decide to move from non-damaged to damaged areas to seek new opportunities. We explore how determinants of migration change over time to understand: (i) who is more likely to move away and return to damaged areas; (ii) who is more likely to move away from non-damaged areas; and (iii) what are the characteristics of the communities that make them more likely to send migrants and attract returned migrants. With respect to community features, we focus mainly on understanding the role that the nature and timing of the reconstruction effort may have played in attracting migrants.

The results presented in this paper contribute to our understanding of the complexity of the relationships between individual determinants of post-disaster relocations and the reconstruction process. These results highlight the importance of having population representative data that tracks individuals across to understand and inform policy decisions.