Elliott Ash (2026) provides budget data and code used to predict corruption across Brazilian municipalities via gradient-boosted classifiers. The collection includes municipal-level predictors and corruption labels used to simulate policy interventions that double detection rates compared to random audits.
Use Cases
- Predicting corruption outcomes using municipal budget data as features
- Simulating the impact of targeted audit policies on corruption detection rates
- Replicating the gradient-boosted classifier model developed by Elliott Ash
Strengths
- Includes replication code for tree-based gradient-boosted classifiers
- Validated against previous corruption studies
- Smallest geographic unit is the municipality
Limitations
- Geographic bias restricted to Brazilian municipalities
- Specific budget column names and record counts are not provided in the metadata
Provenance
- Source
- Elliott Ash, ICPSR Harvested Dataverse
- Freshness
- Last updated February 2026
- Geography
- Brazil