Replication package for the paper "Renewable Energy Curtailment in Brazil: High-Frequency Evidence on Operational Constraints and Grid Saturation"

Replication package for the paper "Renewable Energy Curtailment in Brazil: High-Frequency Evidence on Operational Constraints and Grid Saturation". The study investigates renewable energy curtailment (constrained-off) of wind and solar plants in the Brazilian National Interconnected System (NIS) between 2021 and May 2026, using operational records from the Brazilian National System Operator (ONS) and registry data from the Brazilian Electricity Regulatory Agency (ANEEL). The empirical strategy combines descriptive analysis, temporal decompositions, operational heatmaps, and panel econometric models estimated on a daily submarket-level dataset (two-way fixed effects with HC3 heteroskedasticity-robust standard errors). This package provides the processed and aggregated analytical datasets, a variable dictionary/metadata, and the replication scripts needed to reproduce the main tables, figures, and econometric results reported in the article. Because of file-size and redistribution constraints, the original high-frequency plant-hour database (~84 million observations) is not deposited in full; the deposited daily submarket-level panels are sufficient to reproduce the reported descriptive and econometric evidence. Contents: processed daily submarket-level panels (wind and solar generation, short-run price PLD, load, renewable saturation); raw-source metadata; variable dictionary; replication scripts (data preparation, descriptive analysis, TWFE econometrics and robustness, reproducibility checks); and the eight manuscript figures. Software requirements are listed in requirements.txt and environment.yml. Data sources: ONS constrained-off records for wind and photovoltaic plants; ONS capacity-factor and submarket load data; ANEEL SIGA plant registry; CCEE short-run price (PLD); EPE documents for policy context.