Hypothesis

Self-reported emissions data submitted to EPA's Toxics Release Inventory (TRI) by industrial facilities contain detectable digit-frequency anomalies (Benford's Law violations) that systematically precede EPA enforcement actions by 6-18 months.

Background

Benford's Law predicts that leading digits in naturally occurring numerical datasets follow a logarithmic distribution (digit 1 appears ~30.1%, digit 9 appears ~4.6%). Manipulated or fabricated data often violates this distribution. Financial auditors have used this technique for decades, but it has never been systematically applied to environmental compliance data.

Testable Predictions

1. Facilities that receive EPA enforcement actions (consent decrees, penalties > $25,000) should show statistically significant Benford's Law violations in their TRI submissions for the 1-3 reporting years preceding the enforcement action (chi-squared test, p < 0.01)
2. The magnitude of digit-frequency deviation (measured by Mean Absolute Deviation from Benford's expected frequencies) should correlate positively with penalty severity (Spearman ρ > 0.3)
3. Compliant facilities (no enforcement actions in 10+ years) should conform to Benford's distribution within expected statistical bounds
4. The pattern should be strongest in volatile organic compound (VOC) and criteria pollutant categories where self-reporting incentives to understate are highest

Falsification Criteria

If a sample of 1,000+ facilities with enforcement actions shows no significant difference in Benford's Law conformity compared to 1,000+ matched control facilities (same industry NAICS code, similar production volume), the hypothesis is falsified.

Data Sources

All freely available:

• EPA TRI: Toxics Release Inventory (envirofacts.epa.gov) — 30+ years of self-reported emissions by facility
• EPA ECHO: Enforcement and Compliance History Online (echo.epa.gov) — enforcement actions, penalties, consent decrees
• EPA ICIS-Air: Integrated Compliance Information System for air permits

Implications

If confirmed, Benford's Law analysis could serve as a low-cost, automated pre-screening tool for EPA enforcement targeting — identifying likely data fabricators before physical inspection. This converts a reactive enforcement model into a predictive one using data facilities already submit.