Resumen

In this paper I study how individuals' choices of housing investments are affected by social norms of housing quality governing their relevant network. I use a field experiment that randomly assigns the provision of improved housing to slum dwellers in Estado de Mexico. The program was randomly assigned at household-level within each slum. Given that the density of treated neighbors varies across households' locations within each slum, the experimental design provides not only random variation in the offer of the program, but also on the average housing quality (the social norm of housing quality) surrounding treated and untreated slum dwellers. This allows me to test whether average treatment effects of the program on subsequent housing investments vary with the degree of treatment density surrounding households' locations. I find that untreated households exposed to higher density of treated neighbors invest less on housing quality than untreated households living in areas with lower treatment density, after 1 to 2 years of the intervention. Further evidence suggests that this negative spillover effect is driven by differences in the valuation of potential material improvements across untreated groups. I interpret this as a ``demoralization" effect created by a sudden and significant shock of neighbors with better housing, which resulted in a reduction of their willingness to conform to a higher housing norm while untreated. In contrast, treated households exposed to a higher treatment density increased substantially their housing quality, as well as their water and sewerage connection compared to treatments surrounded by a lower density of treated peers, an effect that is potentially channeled by their stronger willingness to conform to a higher housing norm. I also find that these housing, water, and sewerage investments resulted in a 52% reduction on diarrhea episodes for children below 5 years old, evidencing large positive spillovers on health outcomes of the housing program. All results hold after considering potential endogeneity of individuals' location choices. Finally, I provide robust evidence that these negative and positive spillover effects are increasing in the number of surrounding treated neighbors and that these are mostly driven through conformity effects, ruling out alternative mechanisms.