NETWORKS: Economic Allocations in Social Networks: Evidence and Theory

November, 2011 to October, 2016
EU FP7 - ERC Starting Grant

Social networks affect many economic interactions, and the social capital embedded in them may help explain broad, macro level outcomes. A recent theory literature develops models of economic allocations in networks. But at this point, we have little evidence on which network mechanisms are most important in practice, how networks interact with markets, and how these micro forces translate into aggregate outcomes.In this proposal I combine micro level measurement with theory to evaluate various mechanisms through which networks affect economic allocations. I explore this theme in four domains. (1) Borrowing and informal insurance in development. I measure the value of connections for borrowing, and study how transfers propagate through the network using field experiments in Peru. (2) Peer effects and the social multiplier. In field experiments I measure financial and peer-based incentives, and how they reinforce each other. I also measure knowledge diffusion about exporting in corporate networks, and the resulting multiplier effect of reducing trade barriers. (3) Information aggregation. I measure how different pieces of information are filtered and aggregated in the social network. (4) Favouritism. I study the economic causes and consequences of favouring friends in a field experiment. I also measure economic misallocation resulting from politicians favouring connected firms in Hungarian data, and the cost to aggregate productivity.In all projects, my measurement emphasizes causality through field experiments and a unique firm level dataset with many sources of variation. Estimating models allows me to contrast theories and generalize the empirical findings. The results will help evaluate the importance of social networks for microeconomic and aggregate allocations, yield lessons on how organizations and policies leverage social mechanisms, and may open a new research area on mechanism design with network effects.