raihanpka/indonesia-fiscal-pressure

Indonesia Fiscal Pressure Tracker

A curated, provenance-tracked time series of fiscal pressure indicators for the Republic of Indonesia. Built as the input data layer for the sim-id-fiskal service in Project Santara: An open-source counterfactual microservices platform for simulating Indonesia's economic, political, and climate systems.

Dataset Summary

Format: Long format. One row is one observation of one indicator at one date in one region. Designed for time series analysis, joins, and the gRPC boundary the sim-engine expects.

• —Total rows: 84,203.
• —Date range: 2013-05-20 to 2026-06-15.
• —Unique indicators: 70 across 5 sources.

Source mix:

Raw per-source CSVs are available in `data/` for users who want the original wide-format extracts.

Supported Tasks

Time series forecasting, macro shock simulation, causal analysis of fiscal and price shocks, inflation impact modeling.

Data Fields

Splits: Single train split. Use standard time-based splits (train on first 80%, test on last 20%).

Source Data

Five primary sources, all Indonesian government open data or public news:

1. 1.BI 7-day reverse repo rate (BI7DRR). User-provided CSV from bi.go.id. 124 monthly observations.
2. 2.JISDOR (USD/IDR). Converted from BI SOAP web service XLSX. 3,160 daily reference rates.
3. 3.PIHPS (Indonesian food prices). 2018-2026, combining azzandwi1 daily data (2018-2021) and BI GetDetailGridData2 backfill (2022-2026). 52 commodity varieties across both periods.
4. 4.Bapanas. Monthly national consumer prices from data.badanpangan.go.id. 25 commodities, 2021-01 to 2025-12.
5. 5.Curated BBM retail prices. From news articles (CNBC Indonesia, Suara, JawaPos, ICCT, Katadata, IDN Times, Kabarbaik, Kompas).

No manual annotations. The `indicator` column is a canonical name derived from source column names. No personal or sensitive data. Only public economic indicators.

Known Limitations

• —PIHPS 2018-2021 is daily continuous; 2022-2026 is business-day sampling. The 2022-2026 backfill uses BI's GetDetailGridData2 API which returns ~5 dates per query.
• —Bapanas ends December 2025. Bapanas has not published 2026 data.
• —Fuel prices are event-driven, not continuous. Indonesia announces BBM price changes on specific dates. Granularity is event-level.
• —No imputed values. Missing data is simply absent.
• —Provenance is two-level. Macro level (one entry per source_id pattern in sources.csv) and per-row level (one entry per data row in provenance.csv).

Social Impact

Part of Project Santara. The platform lets policy makers, journalists, and citizens ask "what if" questions about proposed economic policies grounded in real data.

Dataset Curators

• —Raihan Putra Kirana (Creator of Project Santara, me@raihanpk.com).
• —AI assistants (used as curators for the fuel-price source only, compiling news articles. AI was not used to impute any values).

Licensing

Apache 2.0. Source data is public domain per UU No. 14/2008 on Indonesian Open Data unless noted in the source registry.

Citation

@misc{indonesia-fiscal-pressure-2026,
  author = {Raihan Putra Kirana},
  title  = {Indonesia Fiscal Pressure Tracker},
  year   = {2026},
  howpublished = {Hugging Face Dataset},
  url   = {https://huggingface.co/datasets/raihanpka/indonesia-fiscal-pressure}
}

@misc{indonesia-fiscal-pressure-sources-2026,
  author = {Raihan Putra Kirana and {Bank Indonesia} and {Badan Pangan Nasional} and azzandwi1 and {CNBC Indonesia} and Suara and JawaPos and {ICCT} and Katadata and {IDN Times} and Kabarbaik and Kompas},
  title  = {Indonesia Fiscal Pressure Tracker --- Source Attribution},
  year   = {2026},
  howpublished = {Hugging Face Dataset},
  url   = {https://huggingface.co/datasets/raihanpka/indonesia-fiscal-pressure}
}

For per-source citations, see provenance.csv for individual article URLs and publisher metadata.

How to Load

Combined dataset (all 84,203 rows):

from datasets import load_dataset
ds = load_dataset("raihanpka/indonesia-fiscal-pressure", split="train")
print(ds[0])

Single source (per-source Parquet at repository root):

import pandas as pd
bi_rate = pd.read_parquet("hf://datasets/raihanpka/indonesia-fiscal-pressure/bi_rate.parquet")
pihps = pd.read_parquet("hf://datasets/raihanpka/indonesia-fiscal-pressure/pihps_daily.parquet")
bapanas = pd.read_parquet("hf://datasets/raihanpka/indonesia-fiscal-pressure/bapanas.parquet")

Available: bi_rate.parquet, jisdor.parquet, pihps_daily.parquet, pihps_api.parquet, bapanas.parquet, fuel_curated.parquet. All use the same 6-column schema.

Provenance (per-row source links):

import pandas as pd
data = pd.read_parquet("hf://datasets/raihanpka/indonesia-fiscal-pressure/train-00000-of-00001.parquet")
provenance = pd.read_csv("hf://datasets/raihanpka/indonesia-fiscal-pressure/resolve/main/provenance.csv")
joined = data.merge(provenance, on=["source_id", "date", "indicator", "region"], how="left")

Contributions

Open an issue or PR at `github.com/raihanpka/project-santara`.

Thank you for using this dataset. Hope it serves your research and public policy analysis well.

Copyright © 2026 Raihan Putra Kirana. All rights reserved.