Networks play an important role in the integration of migrants to a new economy, especially in the rapidly expanding urban environments of many developing economies. I posit the following hypothesis: as economic forces induce migration from rural to urban areas, migrants weaken their ties to their village network at origin, and join the urban network which offers the highest return, which in this case is the religious network. As a result, patterns of migration will be correlated with patterns of religious change both for the migrant and for households left at the origin. This may partially explain the rapid religious transition from ethnic religions towards Pentecostalism observed in many African nations. I test this hypothesis with new data on religious institutions and religious participation, employing an instrument variable approach to establish the link between migration and conversion. I then use information on migration outcomes, transfers, consumption smoothing, and religiosity to establish the mechanism. I find evidence supporting the network story for migrants. Households at origin convert as a form of investment in the migrant. 

We propose an econometric framework that accounts for the effects of both observed and unobserved

characteristics of social networks on individuals’ decision to undergo treatment in counterfactual

models based on potential outcomes. We show that network peer effects provide exogenous variation

to identify the marginal treatment effect (MTE) and the marginal policy-relevant treatment effect

(MPRTE). Monte Carlo experiments demonstrate that ignoring the influence of social interactions

on an individual’s decision in counterfactual analysis can lead to misspecification, substantially

biasing both the MTE and MPRTE estimates. We apply the proposed methodology to the college

attainment model in the US, using Add-Health data that contain high school friendship networks.

The results indicate that not controlling for high school friendship network peer effects misidentifies

the marginal return to attending college, as well as the marginal effect of policies aimed at improving

college attendance.