Microeconomic Systems as a Computational Science: A Computational Platform for Microeconomic System Simulation and Human-Subject Experimentation

Stephen A. Kunath

Advisor: Kevin A McCabe, PhD, Department of Economics

Committee Members: Daniel E. Houser, Thomas Stratmann

Vernon Smith Hall (formerly Metropolitan Building), #5075
April 24, 2023, 09:00 AM to 10:00 AM

Abstract:

This dissertation investigates how economics experiments are conducted and details a new computational framework for the description and provisioning of experiments. In addition to the text of this dissertation a novel computational economics platform called mTree has been developed. mTree provides economists with the ability to describe microeconomic systems using a simple set of constructs in the Python programming language. An mTree microeconomic system can then be used to run computational simulations of a microeconomic system as well as host an conduct human-subject experiments using the same underlying codebase, a capability other frameworks have never provided. 

The first chapter provides background on various issues surrounding economics experiments. One criticism of experimental techniques is that they lack generalizability to non-laboratory settings. The chapter describes the source of this criticism and suggests that an underlying and unaddressed problem relates to the manner of description of microeconomic systems. A further suggestion concerns the additional challenge of ensuring replicability of experiments and that a richer description of underlying microeconomic systems can provide more than a procedural and protocol-oriented description of an experiment. It concludes by suggesting that the tools available from computer science can be employed to provide improved frameworks and description languages for economic systems. 

The second chapter suggests that microeconomic systems align well with work in the field of computer science concerning distributed systems. Arguably, a microeconomic system is in fact a distributed computation problem that contains many of the same challenges faced by computer scientists. From this a new computational framework for describing microeconomic systems is offered utilizing capabilities of a computing technique called an actor system. This new framework is called mTree and can be used for describing microeconomic systems as well as conducting simulations of the microeconomic system as well as human-subject experiments utilizing the same underlying codebase. 

The final chapter provides several examples of microeconomic systems implemented in the mTree framework. These examples include a variety of microeconomic systems for use with human-subject experiments. The examples include both computer simulation code of the microeconomic system as well as human subject experiment implementations of the examples. The various microeconomic systems provided in this chapter are meant to serve as both a tutorial on the use of the mTree framework as well to serve as reusable components for future experiments and simulations.