Observational Matching & Multi-Jurisdiction Attribution
Matching design on observational data (174 matched clusters, null models) and multi-jurisdiction burden attribution (HHI, Pielou indices across 13 states / 270 units).
Project Overview
Measured policy impact from observational data with a quasi-experimental matching design: 174 matched clusters across two reporting vintages, benchmarked against null-model counterfactuals, with variance decomposition isolating where the national reporting bias originated. Built a multi-jurisdiction burden-attribution model from scenario-based projection across 13 state jurisdictions and 270 tracked units, combining simulation and network analysis with concentration (HHI) and evenness (Pielou) indices to quantify where policy impacts cross regulatory boundaries no single authority spans.
My Role
Lead Quantitative Modeler. Designed observational matching estimators, null-model counterfactuals, and multi-jurisdiction attribution indices.
Tools / Stack
Project Details
Implementation Roadmap
Development Process.
Construct matching design across 174 matched observational clusters spanning two reporting vintages.
Execute null-model counterfactual simulations and perform variance decomposition to isolate reporting bias.
Model trans-boundary spatial projections across 13 state jurisdictions and 270 tracked units.
Compute Herfindahl-Hirschman Indices (HHI) and Pielou evenness metrics to quantify governance concentration.
Present findings at IALE North America (Keynote, 2026) and Southeast Climate Adaptation Science Center.
Case Study Analysis
Challenges & Outcomes.
Policy impacts and environmental assets cross state and administrative boundaries that no single governance authority spans, confounding traditional impact attribution.
Developed multi-jurisdiction burden-attribution models pairing matching estimators with HHI concentration and Pielou evenness indices.
Isolated national reporting bias mechanisms and provided agency leaders with defensible trans-boundary attribution frameworks.