ORC Seminar: Points-Based Allocation with Costly and Strategic Reporting

Elizabeth Paulson

Assistant Professor
Harvard Business School

Abstract: Agencies often allocate scarce resources with transparent point systems in which individuals choose whether to apply and which score-relevant features to report. Because announced point values can influence these choices, we study two behavioral responses. First, application and documentation are burdensome, so costly features can discourage application and reduce program access. Second, applicants may strategically activate features absent at baseline to improve their scores. These responses put opposing pressure on point values and shape both optimal feature selection and point allocation. We show that access and integrity requirements can be incompatible, and that adding correlated—even statistically redundant—features can improve design. Finally, we address a practical hurdle: how to redesign high-stakes point systems with limited information. We propose a DRO-based approach to identify promising one-shot redesigns, and demonstrate the approach on the existing SNAP benefit formula using public quality control data.

Bio: Elisabeth develops analytical methods to improve decision-making in public and social-sector operations. She studies how algorithmic systems can personalize interventions, allocate scarce resources, and incorporate fairness considerations in ways that are useful to the people and organizations that deploy them. One application setting of focus is improving access to healthy food. Another stream of work develops algorithmic decision-support tools to improve refugee resettlement. Elisabeth holds a PhD in Operations Research from the ORC.