Experimental Design
All eligible neighborhoods in the municipalities participating in the program were part of the sample. A random fraction of eligible neighborhoods in each municipality were assigned to the treatment group. Treatment was a block grant to be spent on improving basic infrastructure in low-income urban neighborhoods. This included introducing or expanding things such as water access, sewerage, road pavement, street lights, community sports fields, etc. The researchers conducted a baseline and a follow up survey to create a panel dataset; however, panel-tracking was performed at the house-level, not the household level, as the researchers were interested in changes in structure attributes.
Since the level of treatment has an endogenous component due to the number of eligible neighborhoods in a city, the researchers instrument for the treatment with the randomly assigned fraction of eligible neighborhoods. The researchers then examined evidence of flypaper effects, the impact of the treatment at the neighborhood level, evidence for spillover effects, impacts of private investment, impacts on real estate values, and impacts on political behavior.