The effect of aid is a subject of considerable academic and policy interest. In this question we examine the relationship between foreign aid and civil conflict using a recent paper by Crost et al. (2014) in the Philippines. The authors collect data on civil conflict related incidents at the municipality level as well as information on aid receipts at the municipality level. The conflict data was collected from the Philippine military and provides casualty counts (as well as the party responsible) by municipality. Aid in this context refers to a large anti-poverty program (the KALAHI-CIDSS) funded by the World Bank and run from 2003-2008. The program provided cash grants to municipalities to build infrastructure such as roads, water-systems, clinics and schools. The grants were fairly large and accounted for approximately 15% of the annual budget for the average municipality. (a) Let Yi denote the number of conflict casualties in municipality i in a given timeperiod after the introduction of the KALAHI-CIDSS program. Let Di be a binary variable equal to 1 if municipality i received the KALAHI-CIDSS program. The researchers collect data on casualties and program presence in a sample of n counties {Yi , Di} n i=1. Consider the following regression specification Yi = α + βDi + i . If we are interested in learning about the causal effect of the program on conflict casualties (i.e. we wish to interpret β as the causal effect of the aid program) what is the interpretation of the error term i in the regression? (b) The government of the Philippines (in consultation with the World Bank) decided which municipalities would receive the program. Suppose (for this part only) that the rule they used was to target the program to the poorest municipalities and that these municipalities also tended to have high rates of violence. Given this information, do you think that estimating the above equation using Ordinary Least Squares will yield a good estimate of the causal effect β? Justify your answer. (c) The authors found that due to budgetary reasons the aid program was only provided to the poorest 25% of municipalities. A municipality that was below the 25th percentile of a poverty rank index received the program while a municipality that was above the 25th percentile did not receive the program. Use this knowledge to suggest an empirical strategy to estimate the causal effect of the program on violence. What assumption(s) are needed for your proposed strategy to be credible? In particular, how does your approach get around potential criticisms of omitted variable bias. Be brief in your answer.
The effect of aid is a subject of considerable academic and policy interest. In this question we examine the relationship between foreign aid and civil conflict using a recent paper by Crost et al. (2014) in the Philippines. The authors collect data on civil conflict related incidents at the municipality level as well as information on aid receipts at the municipality level. The conflict data was collected from the Philippine military and provides casualty counts (as well as the party responsible) by municipality. Aid in this context refers to a large anti-poverty program (the KALAHI-CIDSS) funded by the World Bank and run from 2003-2008. The program provided cash grants to municipalities to build infrastructure such as roads, water-systems, clinics and schools. The grants were fairly large and accounted for approximately 15% of the annual budget for the average municipality.
(a) Let Yi denote the number of conflict casualties in municipality i in a given timeperiod after the introduction of the KALAHI-CIDSS program. Let Di be a binary variable equal to 1 if municipality i received the KALAHI-CIDSS program. The researchers collect data on casualties and program presence in a sample of n counties {Yi , Di} n i=1. Consider the following regression specification Yi = α + βDi + i . If we are interested in learning about the causal effect of the program on conflict casualties (i.e. we wish to interpret β as the causal effect of the aid program) what is the interpretation of the error term i in the regression?
(b) The government of the Philippines (in consultation with the World Bank) decided which municipalities would receive the program. Suppose (for this part only) that the rule they used was to target the program to the poorest municipalities and that these municipalities also tended to have high rates of violence. Given this information, do you think that estimating the above equation using Ordinary Least Squares will yield a good estimate of the causal effect β? Justify your answer.
(c) The authors found that due to budgetary reasons the aid program was only provided to the poorest 25% of municipalities. A municipality that was below the 25th percentile of a poverty rank index received the program while a municipality that was above the 25th percentile did not receive the program. Use this knowledge to suggest an empirical strategy to estimate the causal effect of the program on violence. What assumption(s) are needed for your proposed strategy to be credible? In particular, how does your approach get around potential criticisms of omitted variable bias. Be brief in your answer.
Step by step
Solved in 2 steps
