Social networks and the analysis of peer effects are a current hot topic across academic disciplines. The idea that a structure beyond our own direct scope and understanding, i.e. the social network we are embedded in, significantly influences how our lives evolve has become increasingly popular in many academic fields, such as medicine, sociology politics and economics. This week’s CSAE seminar lunch series was honoured by a lecture given by Margherita Comola on how an increase in the access to formal funds through the government changes the network of informal financial transactions in Nepal. In particular, the data come from a field experiment that randomized access to savings accounts among all households living in 19 villages in rural Nepal (in the surroundings of Pokhara). The study makes use of a very elegant identification strategy developed by Bramoullé et al. (2009) that exploits the intransitive nature of the network data.
The paper is thus set at the intersection of several exciting research areas: For one, it is vital to illuminate whether an intervention that randomly offers some people access to formal finance has an impact on their savings behaviour. This is related to the fundamental question of how potential poverty traps can be overcome by offering individuals commitment devices. Secondly, it is interesting to investigate whether the households’ increased access to funds will have positive spillovers for their peers. Will they share part of their surplus? Will they use this opportunity to engage in more informal financial transactions? Lastly, it is critical to think about changes in the network of informal financial flows. Will I start lending money to other people with whom I have no current financial transactions?
One of the main results of the paper is that people receiving access to savings accounts engage in more informal financial transactions with their peers. In other words, as a result of the intervention, the financial flows in the network are increased and substantially altered. This is an interesting finding as the access to formal financial saving could have resulted in lower need for informal financial access. Yet, an intuitive explanation on this positive effect of the intervention is as follows: access to savings can foster asset accumulation. Hence, households with greater resources might increase transfers to others, for example as a result of altruism, or in anticipation of social constraints. Comola subsequently shows how the availability of several rounds of network data can be used to eliminate biases arising from the assumption that these networks are stable over time – which is prevalent in the economics literature, mostly due to data availability constraints. Clearly, taking into account that the intervention changes the network is critical in disentangling the true effect of peers
Many challenges in the literature remain: Clear-cut evidence based on clean social network data and economically meaningful outcomes remains scarce. If network data is incomplete or error-ridden, then the validity of this empirical approach might not be given. It would be interesting to examine in future research whether the actual social network of individuals (e.g. one’s choice of friends or acquaintances) could be affected via government interventions.
Sharing norms and the resulting social constraints and other-regarding preferences are very appealing and simple explanations of the empirical findings that are not necessarily reliant on the changes in the network structure. It would be interesting to explore whether variations of such sharing norms at the village level play an important role for this paper’s results. Furthermore, it is interesting to explore whether this increase in informal financial transactions affects levels of trust or even other economic outcomes in the treated villages. For example, do these increased financial transactions affect peer effects in other domains such as consumption? Margherita Comola told me about an interesting and related study with data from Africa, where many of these outlined challenges in the literature shall be addressed in the future.
 This an important contribution to the difficult empirical task to overcome the reflection problem Empirical researchers face the problem that without intransitive network data they cannot disentangle whether people’s behaviour changes because of other people’s behaviour (genuine peer effect), their peers’ (exogenous) characteristics or a common shock (correlated effect) they are facing. Yet, in this methodology the effect arising from changes in the network was neglected. This smart idea by Marhgerita Comola is incorporated in her dynamic peer effect model.