The Wharton Geospatial Initiative, part of the Wharton Real Estate Center, is advancing research on the impact of zoning and land-use regulations on housing supply through the National Land-Use Restrictiveness Index (NLURI). This project seeks to measure how restrictive zoning regulations are across the United States by estimating the maximum number of legally allowable housing units for every census block group. By systematically interpreting and quantifying zoning codes, the NLURI provides a consistent, data-driven measure of land-use regulations to support research in real estate economics, public policy, and urban planning.
The project involves analyzing zoning ordinances from thousands of jurisdictions nationwide. Each ordinance defines development standards—such as maximum dwelling units per acre, minimum lot size, and building dimensional requirements—that shape how land can be used. The NLURI team has developed a methodology to translate these complex, locally specific regulations into comparable metrics across jurisdictions. This work enables researchers to better understand how land-use regulations contribute to differences in housing supply, costs, and accessibility across regions.
A significant portion of the project focuses on automating the extraction of zoning standards to speed up data collection and improve scalability. Historically, interpreting zoning codes has been a manual and time-consuming process, requiring researchers to read through ordinance text and derive maximum allowable densities based on different residential building scenarios. To address this, the team is building a workflow that leverages large language models (LLMs) and retrieval-augmented generation (RAG) techniques. This workflow will allow zoning ordinance text to be indexed, queried, and automatically interpreted into the key variables needed for the NLURI.
The end result will be a national dataset linking zoning restrictiveness to local housing outcomes, providing an important tool for researchers, policymakers, and practitioners. The combination of geospatial analysis, machine learning, and natural language processing offers a new model for studying regulatory impacts on housing supply at scale. This dataset and methodology will support future studies beyond NLURI, enabling similar analyses of land-use policies and their economic and social implications nationwide.