Researchers Build Deregulation Index Using LLMs

Danilo Cascaldi-Garcia and Matteo Iacoviello (March 2026) construct a US news-based deregulation index spanning 1960–2025 using large language models to semantically classify newspaper articles. The index, validated against human labels and Federal Register measures, leads formal regulatory records by nearly one year and decomposes by sector and policy stage. The authors find positive deregulation shocks raise investment, productivity, stock prices, profits, and GDP.