Replication package for "Place-Based Development Credit and Productive Transformation in Lagging Regions: Evidence from Brazil's Northeast Development Fund"

This replication package contains the cleaned data, code, documentation, and metadata needed to reproduce the main empirical results and robustness checks for the paper “Place-Based Development Credit and Productive Transformation in Lagging Regions: Evidence from Brazil’s Northeast Development Fund.” The paper studies Brazil’s Constitutional Financing Fund for the Northeast (FNE), a large-scale place-based development-credit program targeted to the legal SUDENE area. The empirical design uses a municipality-year panel for 2010–2024 and a shift-share exposure measure that combines initial municipal sectoral value-added shares with external sector-year growth shocks in real FNE credit, excluding the municipality’s own state. The main results show that predicted FNE exposure increases formal private employment, private payroll, private establishments, private and industrial value added, and GDP per capita, especially in initially low-capacity municipalities. However, the conversion into value added per worker is capacity-dependent: exposure raises private value added per worker in higher-capacity municipalities, while this productivity-conversion effect disappears in low-capacity municipalities. The package includes: processed municipality-year and municipality-sector-year datasets; the initial productive-capacity dataset; the SUDENE municipal list used for the estimation sample; municipality crosswalks; an IPCA deflator; variable dictionaries; diagnostic files; and R scripts to reproduce the main tables and robustness checks. All data are derived from public Brazilian administrative and statistical sources, including Banco do Nordeste/FNE, IBGE municipal accounts and population data, and RAIS formal labor-market records. This version is prepared for replication and review purposes before journal submission. The package may be updated after peer review if the paper, code, or robustness checks are revised.