An Agent-Based Modeling Approach for Understanding General Education Course Allocation Mechanisms in Practice

Abstract

The theoretical properties of general education course allocation mechanisms are interesting and important, but it is also important to think about how these mechanisms will perform in practice. Even though mechanisms can be strategy-proof, Pareto efficient, or have several other theoretically provable properties, the positive welfare gains of these properties may not be realized in practice. The implementation of course allocation mechanisms in practice is highly dependent on the heterogeneous, imperfectly rational network of students who are requesting course allocations. Agent-based modeling (ABM) is a bottom-up approach that captures such heterogeneous individual behaviors to better understand the macro policy outcomes. This thesis provides a modular and flexible agent-based modeling framework to help Harvard policymakers make more well-informed decisions about Harvard's general education course allocation mechanism and ultimately other allocation mechanisms more broadly. This thesis defines Harvard's general education course allocation problem and applies this course allocation ABM framework to realistically simulate and evaluate a variety of general education course allocation mechanisms, including one that outperforms Harvard's new general education lottery piloted in Spring 2020.