Abstract
This replication package contains the SAS and Stata programs, pseudonymized data, and documentation needed to reproduce the empirical results in the paper "Star Firms, Information Spillovers, and Predictable Industry-Level Outcomes." The package includes code for constructing firm-level and industry-level datasets, estimating earnings growth and earnings surprise regressions, macro-outcome regressions (GDP and employment), dynamic panel GMM analyses, and portfolio construction for asset-pricing tests. Data sources used include CRSP, Compustat, I/B/E/S, LinkUp, BEA, BLS, patent data (Kogan et al. 2017), markup data (De Loecker et al. 2020), technology spillover scores (Bloom et al. 2013), vertical integration scores (Fresard et al. 2020), and Fama-French factors. All supplied datasets are pseudonymized (capped at 1,000 observations with pseudo variable values), except for industry code and linking tables (NAICS/BEA mappings). The analysis is conducted using SAS 9.4 and Stata/SE 18.0 on Windows 64-bit.