The objective of this dissertation is to examine the ways in which households recover from and adapt to changing conditions as a result of natural hazards. I use Indonesia as a case study. The dissertation is divided into four chapters. In the first chapter, I give an overview on the policy relevance of studying natural disasters as well as the broad literature on the topic.

In the second chapter, I estimate the short- and medium-run economic returns to capital and inherent ability by studying the recovery of fishermen in Aceh from the 2004 Indian Ocean tsunami. Since the natural disaster wiped out a significant portion of productive physical capital among fishermen, the subsequent quasi-random infusion of aid boats generates a natural experiment. Using panel data from fishing households, I investigate whether fishermen who were relatively more productive pre-tsunami retain their productive edge ex-post. Results suggest that (i) returns to inherent ability, measured by pre-tsunami productivity, become more important over time while returns to physical capital, measured by aid boats, become less important in the medium run, and (ii) the redistributive effects of boat aid on productivity are small and temporary.

In the third chapter, I explore the short-run (one year) behavioral changes in terms of market labor, voluntary labor, as well as borrowing through formal and informal sources among Indonesian households in the aftermath of natural hazards. I estimate the predicted number of hazards, including earthquakes, floods, landslides and storms, in each of the sampled districts in each survey year using historical data from 1980 to 2008. I then match household data with the residuals from these regressions as the unexpected number of hazards. I find that women from districts with more unexpected disasters work fewer weeks on the market. Unexpected disasters are also associated with higher probabilities of borrowing and larger loans. These results suggest that the substitution effect dominates the income effect in the short run.

In the fourth chapter, I and co-author, Philip H. Brown, investigate the link between poverty and vulnerability with respect to natural disasters by applying a utility measure of vulnerability to household panel data from Indonesia that brackets the 1997 forest fires. Using the decomposition method pioneered by Ligon and Schechter (2003), we find that households who live in areas unaffected by smoke from the fires were less vulnerable in total consumption, but they were no less vulnerable or likely to face poverty, aggregate risk, or idiosyncratic risk in food consumption than those who live in areas that were affected.