Characterising Informal Risk Sharing in Family Networks

Risk is extremely prevalent in rural areas of developing countries, but markets for credit and insurance are undeveloped and governments rarely provide the type of social insurance common in developed economies, with the consequence that risk has considerable repercussions for household wellbeing both in the short and long run. Households in these contexts resort to informal tools, such as transfers and loans between relatives and friends, to share idiosyncratic risk. Such informal risk sharing likely occurs within social networks (friends and family), since interpersonal relationships between network members can alleviate frictions such as asymmetric information, as individuals who know each other well and interact frequently can share information and effectively enforce informal arrangements. Family networks, in particular, fulfil these features and form a "natural" risk sharing institution. This research aims to deepen our understanding of risk sharing in family networks theoretically and empirically, focusing explicitly on the role of network structure (representation of who is related to who). This is likely important because many informal tools employed for risk sharing rely on bilateral relationships. Therefore, who one is linked with and who their links are further linked with is will shape both how risk is shared and the amount shared. Very few empirical studies in economics have considered the role of network structure in shaping economic outcomes, though this has been shown to be important in fields such as sociology. A reason for this is a lack of suitable datasets containing both information on network links and extensive socio-economic information. A key innovation of this study is to circumvent this problem using a rich dataset from Mexico with both of these. Family network links can be identified using the Mexican naming convention, whereby each individual has 2 surnames - one from each parent. This dataset has detailed socio-economic information (at multiple points over 10 years) on all households living in over 500 villages, allowing me to identify the entire village based extended family network and construct valid tests for risk sharing.

The proposed research will:
1. Provide empirical evidence on the importance of network structure for risk sharing. Recent theoretical work confirms this and predicts that network structure matters for whether or not risk sharing occurs and the amount that can be achieved. Yet, there is almost no empirical evidence on this, due to data constraints mentioned above. This research will construct and implement empirical tests for these models, exploiting the rich information in the data to account for concerns that results may be driven by the fact that family networks (and network structure) may have formed specifically for risk sharing purposes.
2. Investigate empirically predictions from theory that networks of higher value provide better insurance. An original contribution will be to approximate the value of a link (and network) using information on consumption and wealth.
3. Investigate whether altruism, or mutual caring, in family networks allows for more risk sharing under limited commitment, by allowing for a wider set of risk sharing arrangements. I will investigate whether this is the case theoretically and empirically.
Understanding the drivers of informal risk sharing is important for policy: government policies aimed at improving insurance could do the converse by crowding out informal risk sharing. This research will inform the design of government policies and insurance products aimed at improving insurance for the poor, a subject of on-going interest and debate in Latin America and other low and middle-income countries, and among insurance companies.

The results of this research will be of interest to both academic and non-academic audiences. On the academic side, it will be of interest to researchers in a number of disciplines, including those studying risk sharing and insurance, the role of networks and social protection issues in low and middle income countries in particular. This will include primarily academics working in the fields of economics, public policy and sociology. The non-academic beneficiaries of this research will be policymakers in low and middle income countries, International Financial Institutions and NGOs and insurance companies working on developing micro-insurance products for the poor.
On the academic front, this research will generate benefits by providing evidence on the importance of explicitly considering network structure (an exact representation of who is related/connected to who) in characterising informal risk sharing in developing countries. A small recent theoretical literature suggests that this is likely an important determinant of whether and the amount of risk sharing that can be supported in a network. Yet, almost no empirical evidence exists on this, due to a scarcity of suitable data. This research overcomes this using a rich data source from Mexico. Further, it will provide evidence of the importance of network value in shaping informal risk sharing, with a key innovation being to approximate the value of a network from available information on wealth. The research will also broaden understanding of risk sharing by developing a theoretical model, which differs from existing models by allowing for altruism between links, and testing its predictions empirically. More generally, it will provide evidence on the importance of considering network structure in understanding the role of social networks in developing countries. Though the latter has been investigated by recent research, very few studies, particularly empirical studies in economics consider the former, though this has been shown to be important in fields such as sociology. Finally, it will make methodological contributions to the empirics of networks in economics, currently a small but growing research area.
On the non-academic side, the results of this research will have important implications for the design of development policies, including social insurance policies, and for the creation of insurance products for the poor (also called micro-insurance). There is a lot of interest and on-going debate among policymakers in low and middle income countries regarding the introduction and expansion of social insurance services. There has also been much interest in providing micro-insurance from insurance companies, policymakers and NGOs, and designs are currently being piloted in a number of countries around the world, with poor take-up of these proving to be a challenge. Careful design of these policies and products is important in determining take-up and also in improving welfare. Better knowledge of existing arrangements and the role of network structure in particular, can help with this and also provide information on the likely demand for such services.
To maximise impacts, I will disseminate results of the research through:
1. Academic Seminars and conferences, such as the Latin American and Carribean Economic Association meetings, and seminars at UCL and IFS and other institutions in Latin America.
2. Academic papers to be submitted for publication in top peer-reviewed journals, and to the IFS working paper series.
3. Non-technical summaries will be published in the form of IFS Briefing Notes and IFS observations. These are freely available from the IFS webpage. Short summaries will also be submitted for publication on the Vox Lacea blog.
4. Organisation of a non-technical workshop at the IFS, inviting stakeholders from institutions including DFID, representatives from NGOs such as Oxfam among others.