What Combined Federal Enforcement Data Reveals About Worker Fatality Risk

I analyzed enforcement records on half a million US employers from 17 federal data sources (source: fastdol.com). The interesting finding isn't one number; rather, it's what the OSHA data reveals in combination with data from other agencies such as EPA and NLRB. A few of the patterns below are obvious in hindsight, but a few are not. All of them, however, are invisible if you're only pulling OSHA, or only pulling WHD, or only pulling one source at a time.

Cross-agency enforcement count predicts fatality rates

How many federal agencies have enforcement records on an employer predicts worker fatality rates almost linearly.

An employer cited by five federal agencies is four times more likely to have a worker fatality on record than one cited by a single agency. The base rate across all 508K employers is 3%.

The reason this works: most of those agencies — WHD, EPA, MSHA, FMCSA, NLRB, OFLC, CPSC, NHTSA — don't investigate worker fatalities. A death at a facility doesn't trigger EPA inspections or NLRB charges. When a company accumulates records across multiple agencies, each agency had independent reasons to act. The cross-agency count is a real signal, not an artifact of one agency's response to a bad event.

Specific agency combinations carry more weight than the count alone

The record count alone is telling, but the combination tells you more. A few specific stacks stand out.

Employers with records at both OSHA and EPA have a 9.0% fatality rate — roughly 3x the base. Stack NLRB on top of that and you get to 13.4%. An OSHA+EPA+NLRB employer is more than four times more likely to have a worker fatality on record than the average employer in the sample, and the signal is statistically overwhelming (p < 10⁻¹²⁵).

What stands out is that the combination outperforms any of its individual components in revealing risk of workplace death. OSHA alone identifies 130K employers with some serious violation history. EPA alone identifies 8K with real non-compliance patterns. NLRB alone identifies 115K with at least one labor practice filing. The intersection of all three is just 1,526 employers — and they carry the highest fatality concentration of any named cohort I tested.

This is why I built fastdol.com. Single-agency searches treat OSHA and EPA and NLRB as three independent questions. The right question is: which employers show up in all three at once.

EPA non-compliance is a workplace-safety signal

EPA doesn't investigate worker deaths. But employers out of compliance with the EPA for four or more quarters have a 7.8% fatality rate vs 3.0% for everyone else. That's a 2.6x lift, and it's one of the cleanest findings in the data.

The interpretation is operational. A company that runs its environmental program badly for years tends to run its safety program badly too. Two agencies, one underlying problem. If you're reviewing a commercial target and their EPA ECHO record shows multi-year non-compliance, the workplace safety record deserves a closer look than the OSHA record alone suggests.

Parent-company enforcement rollup reveals hidden risk

This one only exists if you do entity resolution across locations, which is most of what's hard about aggregating federal data to begin with.

96 parent companies in the sample have worker fatalities recorded at five or more of their locations. Those aren't small companies — they're names like Walmart (98 locations with fatalities out of 2,295 tracked), Tyson Foods (57 of 256), USPS (74 of 3,268), UPS (52 of 1,449), Waste Management (54 of 472), Asplundh Tree Expert (57 of 472).

Interestingly, when you look at all the locations operated by those 96 parents, the fatality rate across the full 25,039 locations is 5.0% — significantly above the 3.0% base. The pattern isn't just “bad parent has bad locations you already know about.” It's that their other locations are elevated too. An employer's parent-level history is predictive of location-level risk even for locations that haven't individually been flagged.

Underwriters can't see this pattern from any single agency search. Individual locations don't show the full parent footprint, and parent names are often inconsistent across agencies. The pattern only becomes visible after you join them.

What this is useful for

For underwriters working commercial risk, the cross-agency count and specific-combination patterns are signal you can use today. EPA non-compliance is free to check and meaningfully predictive. Parent-company rollup gives you visibility into location-level risk that the location's own record doesn't show.

Journalists and compliance researchers get something different out of the same data. The parent-level patterns are story leads. The same aggregation that makes underwriting better also makes accountability reporting possible.

And for trade associations or policy folks, the agency-count ladder itself is the argument for why workforce-safety policy needs to think across regulatory boundaries. A company that fails at OSHA, EPA, and NLRB all at once is not a one-agency problem.

Next week I'll dig into a related question: where federal contract dollars actually flow relative to enforcement records. Worth a whole post of its own.

Methodology

507,944 US employers with federal enforcement records across 15 sources. Stratified: all multi-source employers plus a 20% sample of single-source. Entity resolution uses normalized employer name, state, and ZIP. Parent rollup uses a curated seed table plus OSHA ITA company-name data. Findings reported with relative risk, 95% confidence intervals, and chi-squared p-values where applicable. All findings above clear p < 0.001.

Full methodology and source documentation: fastdol.com/methodology.

Aggregate analysis of public federal enforcement data. Not a statement about any specific employer's current operations or future risk. Not legal, financial, or underwriting advice.

FastDOL aggregates enforcement records from 17 federal sources into queryable employer profiles with entity resolution across agencies. Free tier: 50 lookups per month. fastdol.com/docs