Abstract:
This paper studies peer effects on the use of essential health care services offered by a microfinance institution in impoverished neighborhoods in the Philippines. I apply a novel IV identification strategy to overcome the well-known challenges in the estimation of peer effects in non-experimental, cross-sectional settings. The strategy uses structural information from social networks and the existence of overlapping peer groups for an unbiased estimation. I find positive and substantial peer effects in the communities. An increase in program uptake of 10% in the peer group leads to a 6.6% increase in individual health care utilization. I estimate hazard models to further explore underlying mechanisms. Peer effects are found to be strongest immediately after first exposure to the intervention and to fade out over time. While the strength of the relationship with the peer does not seem to matter for the adoption decision, the peers' structural position in the network does. Interestingly, peers with fewer connections seem to have a particularly strong influence on individuals with a central position in the network.