America's super polluters — methodology

How do industrial facilities across the United States contribute to bad air — not just the chemicals they pump out or the greenhouse gases they emit, but both? Which are the biggest offenders?

You’d think it would be simple to find out. It wasn’t.

The U.S. Environmental Protection Agency has two datasets, the Toxics Release Inventory and the Greenhouse Gas Reporting Program, that each track half of what we wanted to know. Unfortunately, the datasets weren’t designed to mesh.

More than 20,000 facilities report to the toxics inventory, which is known as TRI and requires annual figures from a fairly broad swath of larger industrial emitters (sites with at least 10 employees that meet chemical-use thresholds and are either federally owned or in key sectors — manufacturing, utilities, mining, hazardous waste, publishing and certain wholesalers).

More than 7,000 facilities send their emissions figures to the Greenhouse Gas Reporting Program, which requires details from sites releasing at least 25,000 metric tons of “carbon dioxide equivalents” per year. (Different types of greenhouse gases don’t all have the same planet-warming power, so the EPA wants an apples-to-apples figure.)

Putting the two datasets together and analyzing them took us months. First off, they use different identification numbers to denote specific facilities. In conversations with the EPA, we learned that the agency had compiled separate ID numbers through its Facility Registration Service that could be cross-walked to the two inventories’ IDs, allowing us to figure out which facilities had reported to both. (The EPA did not include the Facility Registration Service IDs with the downloadable TRI datasets at the time but has since added them.)

More work was required, though, because some complexes have more than one Facility Registration Service number. We looked up all sites ranked in the top 300 on each inventory that didn’t appear to report to both programs; many actually did.

In cases where a company split one complex into pieces for purposes of emissions reporting, we combined those reports — just those made by the company, not on-site contractors. Some companies reported a complex’s emissions jointly to the TRI, in pieces to the greenhouse-gas inventory and had more than one Facility Registration Service ID to boot. This part of the process involved much eyeballing of satellite photos and consultations with the EPA and individual companies.

The emissions are self-reported by companies. (We reached out to owners of top-100 facilities to give them a chance to double-check the figures.) Firms can at any time revise their TRI numbers, which are often estimated, so we downloaded them again on Sept. 1, 2016, to get the latest possible snapshot of 2010-2014 emissions. The 2014 figures are the most recent final numbers available for both programs, as of our publication date; the EPA recommended against relying on preliminary figures. You can see the results here in our combined dataset

The EPA has itself looked at integrating the TRI and greenhouse-gas figures and ran into the same challenges we did. The agency said our approach was reasonable.

To understand the demographics of people living near the facilities, we used census-block data, combining them and apportioning their information as necessary to stay within a three-mile radius. The method is similar to one that the EPA used for its environmental-justice mapping tool.

At the company level, the University of Massachusetts Amherst’s Political Economy Research Institute has produced separate rankings of the biggest emitters of air toxics and greenhouse gases. For a 2013 paper about health benefits as a positive side effect of tackling climate change, researchers connected U.S. facilities releasing substantial amounts of greenhouse gases to their unhealthy air pollution emissions; a 2016 research brief looked at a similar trend among California facilities.

But several academics said they were not aware of any attempt besides ours to pull these two types of air-pollution categories together for all reporting industrial sites and show the results.

“This is going to be really interesting to people,” said Sid Shapiro, a professor at the Wake Forest University School of Law in North Carolina with expertise in administrative procedure and regulatory policy. “It’s important to know.”