Estimating the impact of a food security program by propensity-score matching

Reducing poverty and improving household food security remains an important policy objective for rural development in the semi-arid areas of many countries in Africa. Many development programs have been introduced in efforts to bring the cycle of poverty and food insecurity to an end. This paper investigates the impact of a food security package (FSP) program in improving rural household’s food consumption in Tigray region, Northern Ethiopia. An empirical analysis based on a propensity score matching (PSM) method, which is a popular approach to estimate causal treatment effects, is employed. Using kernel-matching estimation technique, program beneficiaries were matched with non-beneficiaries. The results show that the program has had a significant effect on improving household food calorie intake. The findings indicated that the program raised the food calorie intake of beneficiary households by 41.8% above that of individuals not involved in the program. Sensitivity analysis also indicated that the observed estimate of impact is not vulnerable to hidden bias or selection on unobservables.   Key words: Propensity score, matching, selectivity bias, average treatment effect, impact, evaluation.