Peru GDP Real-Time Dataset (1994-2025)

This dataset provides a comprehensive real-time database (RTD) of Peru's Gross Domestic Product (GDP) growth rates, tracking revisions from 1994 to 2025. To the best of our knowledge, this is the first fully reproducible real-time dataset for an emerging economy. The accompanying open-source code pipeline allows for automatic updates to extend coverage to the present and beyond. Data are sourced from the Central Reserve Bank of Peru (Banco Central de Reserva del Perú, BCRP) Weekly Reports and systematically transformed into structured formats suitable for revision analysis, nowcasting, and forecasting research. The deposit is organized in a three-tier structure reflecting the complete data processing pipeline: Tier 1: Raw data (raw/old_weekly_reports/): 420 manually-curated CSV files from pre-2013 scanned Weekly Reports, organized by table type (table_1 for monthly, table_2 for quarterly/annual) and year folders (1994–2012). These files are REQUIRED to execute the code pipeline. Tier 2: Input vintages (input/): 696 cleaned and standardized CSV files representing the intermediate processing stage, organized by table type and year folders (1994–2025). These files enable partial pipeline reproduction and intermediate validation. Tier 3: Final outputs (output/): 16 analysis-ready CSV files organized in two complementary formats: Vintage format (output/vintages/): columns represent release dates, enabling analysis of information available at specific points in time. Releases format (output/releases/): columns represent revision sequences (1st, 2nd, 3rd+ releases), facilitating revision pattern analysis. Each output format is available in four variants: base (standard growth rates), adjusted (base-year changes flagged with sentinel values), benchmark (binary indicators for methodological revisions), and combined (adjusted + benchmark). Key features: Three-tier structure: Raw data → Input vintages → Final outputs, enabling complete data lineage and reproducibility. Coverage: Monthly, quarterly, and annual GDP growth rates (1994–2025). Vintages tracked: 1,000+ data releases across 30+ years. Industries: Aggregate GDP + 8 economic sectors (Agriculture, Fishing, Mining, Manufacturing, Electricity, Construction, Commerce, Services). Base-year adjustments: Accounts for methodological changes (1990, 1994, 2007 base years) Benchmark indicators: Files with "benchmark" in their names contain binary indicators (1.0 = benchmark revision with simultaneous update of monthly and quarterly/annual tables, 0.0 = otherwise) instead of growth rates. Dual Formats: Vintage format (columns = release dates) and Releases format (columns = revision sequences) Fully reproducible: Complete Python pipeline available on GitHub. Automatically updatable: Pipeline downloads post-2013 PDFs from BCRP automatically — no manual download required. Usage options: Direct analysis: Download Tier 3 output files for immediate use. Partial reproduction: Start from Tier 2 input vintages. Complete reproduction: Execute full pipeline from Tier 1 raw data using the code repository The data enable research on GDP revision patterns, real-time forecasting, nowcasting, forecast evaluation, and cross-country comparisons of statistical practices in emerging economies. Related code repository: https://github.com/JasonCruz18/peru_gdp_revisions