MIT Media Lab, E15, Bartos Theater
How different are the characteristics of societies that are constrained to local interactions in networks as compared to societies where all interactions happen in organized markets? Among most species and even in modern human societies, exchange, whether of food, information, or labor, naturally tends to occur locally, as encounters happen between nearby individuals in networks. In the absence of coordination across multiple individuals and arbitrage, such exchanges are also often bilateral. Neoclassical economic theory has provided us with a deep understanding of organized markets. We understand equilibrium in these markets as well as the welfare properties and dynamics. This understanding has guided the design of public institutions and the control of externalities for centuries. Using the model of organized markets, public policies are designed for an independent population and completely ignore the interactions among individuals in the population. Network exchange theory in sociology has provided us with very important insights into the properties of local exchanges, the origin of social capital, power in social relationships, fairness in local exchanges and peer-pressure.
We still do not understand however, how these local exchanges govern the large scale properties of networked societiesstability, welfare, dynamics, and fairnessand how can we use peer-pressure to improve social welfare. This lack of understanding has prohibited us from designing effective public institutions and policies for networked societies. Today we have easy availability of `big data' about social and economic interactions. To use this new resource for the betterment of society, we need to understand exchanges in networked societies, build tools for computing the structure of stable and fair networked societies, and predict how they may respond to policy changes. Ankur Mani's thesis is an exploration of questions that fill the gaps in our understanding of networked societies. It will also provide the computational tools for computing the structure of stable and fair networked societies and proposals for the design of effective public institutions. Although his particular examples concern the design of public policies, these recommendations can also be used for revenue management.
The contributions of this thesis include:
A theoretical framework for understanding bilateral exchanges in networked societies and designing public policy and revenue management policies.
Experimental evidence that the proposed design of public policies using peer-pressure is more efficient than existing policies such as individual taxation and subsidies and its applications to health and environmental policies.
Computational tools and algorithms with proofs of correctness and convergence for computing the structure of stable and fair societies.
Host/Chair: Alex 'Sandy' Pentland
Asuman Ozdaglar, Parag Pathak