Margherita Comola*, Silvia Prina
This article was originally published in the October 2020 edition of the 5 papers… in 5 minutes.
The importance of social interactions is currently an active field of study in economics. But in which circumstances we form and sever these links? In reality social connections are not set in stone, but they rewire following the flow of one’s life and the institutional landscape. This leaves us with an important question: when we evaluate policy measures, shouldn’t we also take into account their interaction with social networks?
In this article Margherita Comola and Silvia Prina explore the unexpected policy consequences of social network changes and propose a model to answer these questions. The starting point of this research was a randomized experiment they conducted in 2010 in rural Nepal. In this experiment, women in the treatment group were given the option of opening a formal saving account at the local bank office, and the researchers have noticed that the access to formal savings changed the pre-existing network of informal financial arrangements: women who accumulate via a formal saving account increase their informal financial exchanges with their friends.
Policies which target directly and/or primarily social interactions are rather rare: this is the case for instance of programs reshuffling students across schools to increase diversity. But also the large majority of seemingly-neutral economic policies which are primarily targeted at increasing individual well-being, like the one implemented in Nepal, may have consequences on social interactions. The authors focus first on the consumption patterns of the women involved in the experiment, which is an accurate marker of their well-being. In this context Comola and Prina imagine two simultaneous forces in act. On the one side, the intervention has increased one’s consumption directly, because women could save up to smooth consumption: this direct effect of the intervention can be easily measured with standard evaluation methods. On the other side, women may have also changed their social interactions as a consequence of the intervention, and now privilege links towards different partners than they did before, for example higher-consumption ones. When combining this network change with the natural tendency to align own behavior with the one of peers (1), women end up increasing their consumption also because their new friends consume more. This is an indirect effect of the intervention that flows through network changes, and is neglected by standard evaluation methods.
The main contribution of this paper is methodological: the authors formalize a model in which peer effects spread through a network which in turn depends on the policy, and from this they derive a novel measure of the treatment effect which also takes into account intervention-driven network changes. In the Nepalese illustration, they show that neglecting the network change results in a significant underestimation of the impact of the randomized intervention on consumption. However, their measure could be applied to many different scenarios where it is crucial to draw correct conclusions about the interplay of policy measures and social networks.
………………..
(1) This has been largely documented by economists in many different settings – e.g. educational achievements, consumption choices and subjective well-being
………………..
References
Original title of the article : Treatment Effect Accounting for Network Changes
Published in : The Review of Economics and Statistics, forthcoming
Available at : https://doi.org/10.1162/rest_a_00908
* PSE Member
Credits : Shutterstock – Andrii Yalanskyi