Description
- Abstract:
- In the wake of the Second World War, as they presided over major public expenditure reforms, European and American governments supported the development of rigorous mathematical models of economies to guide economic policy. Over the next two decades, general equilibrium models (or Walrasian economies) emerged as the dominant framework. However, as these models were often analytically intractable, as early as the 1960s, a group of researchers led by Herbert Scarf turned their attention to finding “a general method for the explicit numerical solution of [general equilibrium models].” While some methods had limited success in solving simple models, 50 years later, a general method for computing solutions to more complex models remains elusive. Nevertheless, these models---and their often inaccurate solution methods---continue to be widely used in applications such as resource allocation and public policy analysis, raising concerns about the impact of inaccurate solutions on the public good. This thesis addresses this issue by leveraging tools from computer science and game theory to analyze algorithms for general equilibrium models. The first part of this thesis focuses on variational inequalities (VIs), a mathematical modeling paradigm, and their application to Walrasian economies (i.e., models driven by demand and supply). The second part of this thesis concerns pseudo-games, a multiagent optimization framework, and their application to Arrow-Debreu economies (i.e., Walrasian economies in which demand and supply are generated by consumers and firms). The final part of this thesis concerns Markov pseudo-games, a multiagent \emph{stochastic} optimization framework, and their application to Radner economies (i.e., a generalization of Walrasian and Arrow-Debreu economies that explicitly model time and uncertainty). While Parts 1 and 2 of this thesis resolve Scarf's challenge by solving general equilibrium models developed during his lifetime, Part 3 merely scratches the surface of a new research direction. Above all, it raises an exciting and contemporary analog to Scarf’s challenge at the intersection of deep learning, reinforcement learning, and mathematical economics---namely, finding a general method for the explicit numerical solution of Radner economies.
- Notes:
- Thesis (Ph. D.)--Brown University, 2025
Citation
Göktaş, Denizalp,
"An Algorithmic Theory of General Equilibrium"
(2025).
Computer Science Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:qekb9qv7/
Relations
Collection:
-
Computer Science Theses and Dissertations
Theses and Dissertations for the Computer Science department....