Food Security Imputation in and around Refugee Camps in Bangladesh

Food security is threatened for a growing number of households, especially within refugee contexts in developing countries. Most food aid programs need to make targeting decisions under tight budgetary constraints. However, data is often unavailable or blatantly limited, and possibilities of gathering it are most constrained in crisis times. This thesis offers a first proof-of-concept for using small-area-estimation techniques to impute five different measures of food security for refugee camp populations as well as their neighbouring host communities. Overall, imputed values give a good estimate of the actual situation, in most cases the difference between imputed and true values is not statistically significant. Furthermore, imputed values allow for identification of most food insecure areas. All this can be obtained with comparably small samples of about 1,250 households for model training.