Essays on household behaviour at the intersection of conflict and natural disasters: the 2010 floods in Pakistan

This thesis examines household behaviour at the intersection of natural disasters and conflict. I structure this research around four distinct analytical chapters that use empirical microeconomic analysis to study household-level decisions and outcomes in the year following the 2010 floods in Pakistan. I first examine how does conflict affect household access to cash transfer programmes, and what mechanisms explain such effects. Using IV estimation to overcome endogeneity of conflict exposure and cash transfer receipts, I find that conflict reduces household and community level access to two large cash transfer programmes in Pakistan. The effects are driven by the likely presence of armed rebel groups who possibly resent state-led efforts to win legitimacy through social protection programmes. Next, I examine the effect of conflict on household access to remittances. I use IV estimation to overcome the endogeneity of conflict and remittance receipts and find that conflict exposure reduces household remittance receipts. This effect is driven by security threats associated with armed group presence, which threatens the operations of informal money transfer agents. Further, I find evidence for conflict negatively affecting investment-focused remittances as the effects of conflict are strongest among households more likely to use remittances for investment, than for consumption. These findings are in contrast to the macro literature that tend to view conflict as a factor that affects altruistic motives of remittances but has not examined investment motives in detail. In my third analytical chapter I examine the unintended effects of household aid receipts on violence through a mechanism that has not been studied in much detail: civilian militarisation through the purchase of guns. Using propensity score matching to overcome selection bias, I find that overall, flood relief cash transfers did not lead to any increases in household gun ownership. However recipients who own large tracts of land and live in conflict-affected areas were 8.3% more likely to acquire a gun, compared to a matched group of non-recipient households. The effects are driven by households that lived in displacement camps, which may have enhanced security concerns and the need for guns. This suggests that for groups that have low material but high security needs, exogenous increases in cash, through cash transfers, can increase the likelihood of acquiring guns for use, or for signalling, as a safety good. Finally, I examine the under-studied role of uncertainty of disasters in affecting post-disaster short-term migration decisions. I find that while flooding exposure increases the propensity to migrate, a higher level of uncertainty, represented by more anomalous floods (compared to recurring floods), decreases migration. I also develop a measure of flooding anomaly, based on the likely past exposure to floods at the community level, using satellite data on long term precipitation levels, and distance to the nearest rivers. My research examines important, but hitherto under-studied and challenging relationships that play out in complex emergencies, where many households simultaneously face flooding and violent conflict shocks. The findings are relevant for economic theory, empirical analysis and for policy.