Overview

This workshop will focus on computational aspects of economic dynamics. Economic dynamics can be broadly characterized as analysis of dynamical systems as they pertain to economic behavior: i.e., sequential economic models. Earlier studies of economic dynamics, dating back to the early 20th century, mostly concerned themselves with the analysis of market dynamics in repeated trade settings (e.g., the Cobweb model, tâtonnement). A more recent line of study, dating back to 1960, has sought to more explicitly model both time and uncertainty (e.g., Arrow securities, Radner’s exchange economy). While a great deal of progress has been made in understanding the computational properties of the former, much less is known about the computational properties of the latter (i.e., time and sample complexity). 

Economic dynamics are studied by multiple communities—including macroeconomists, financial economists, and computer scientists—who analyze these models using myriad techniques, including dynamic programming, convex optimization, multiagent reinforcement learning, and, more recently, deep learning. Most research on the computational aspects of these models are based on empirical and/or simulation-based analyses; as such, there exists a wide range of open (theoretical and other) research questions at the intersection of economics, computer science, and machine learning. 

The goal of this workshop is to facilitate interdisciplinary collaboration among researchers in algorithmic game theory, multiagent reinforcement learning, deep learning, microeconomics, and macroeconomics with overlapping interests in this area.

Intended Audience

• Graduate Students in economics and computer science.

• Economists who solve or analyze dynamic models as part of their research.

• Computer scientists who work on market equilibrium computation or learning in games.

• AI Researchers who work on economic applications of multiagent reinforcement learning and deep learning.

• Policy analysts using dynamic economic models in their work.