Center for Effective Global Action

This paper reports on the results of a large infrastructure investment experiment in which $68 million in spending was randomly allocated across a set of low-income urban neighborhoods in Mexico. We show that the program resulted in meaningful improvements in the access of the average household to numerous forms of infrastructure, such as electric lighting, street lights, sidewalks, medians, and road paving. We exploit the study’s randomized saturation design to understand how spillovers at the municipal level might work to undermine causal inference when multiple levels of government provide forms of investments that are close substitutes. Our results suggest that the strategic response of municipal governments to federal spending is muted. The program increased the aggregate real estate value in program neighborhoods by two dollars for every dollar invested.