How we did it
During the summer of 2006, researchers for the Center for Public Integrity obtained a draft document marked “Enforcement Confidential” from the Environmental Protection Agency, titled “Top 100 PRPs Based on Number of Sites At Which They Are Involved.”
The document, dated December 2001, listed many of America’s Global 500 companies and their connections to hundreds of sites identified by the EPA’s Superfund program — a law passed in 1980 requiring the government to find and supervise the clean up of America’s most toxic areas.
In the five years since the list was updated, the number of Superfund sites has changed by less than 5 percent, according to the Center’s analysis — so the companies associated with the majority of them have remained largely the same. Asked why the list had not been updated since 2001, the Center’s source said it was too politically hot to continue work on the project.
The Center picked up where the EPA left off. Because the list contained only the number of sites connected to each company, the Center attempted to identify the specific sites by running thousands of queries through the EPA’s databases and by contacting each of the companies involved. Center researchers were able to draw a more complete picture of the companies that the EPA connects to the most Superfund sites.
The Center also obtained an internal EPA memorandum that discussed improvements needed for the data behind it. One such issue was confusion over parent companies and subsidiaries for the “potentially responsible parties,” or PRPs — the EPA’s term for entities that might be responsible for waste at certain Superfund sites. Another was the challenge posed by the arcane maze of mergers and acquisitions of corporate America — who owns what now and did they own their Superfund liabilities as well.
First, the Center established the parent companies of the identified entities on EPA’s list by using Hoovers, a business Web site. It also researched companies that had received more than $25 million in EPA contracts between fiscal years 1998 and 2005. Center researchers sifted through company filings with the Securities and Exchange Commission to help verify the results as well.
Downloading the family trees of the companies on the EPA list and top EPA contractors, the Center then coded each of the roughly 10,000 firms through EPA’s Comprehensive Environmental Response, Compensation and Liability Information System, or CERCLIS, to find their connections to Superfund sites. The companies were also coded in databases of federal lobbying, EPA and congressional travel paid for outside interests, federal grants and EPA fines and penalties related to Superfund.
When the Center began poring through more than 100,000 records of the various organizations linked by EPA to Superfund sites, there were problems. More than 60,000 different entities were listed in the data report that contains PRPs, even if an official “notice letter,” typically the first formal contact from EPA, has not been sent.
Much of the EPA information was inaccurate. Thousands of names were misspelled and records incomplete or out of date. In some cases, a company was listed many different ways. General Electric Co., for example, which is connected by the EPA to 116 Superfund sites, is spelled more than 30 different ways throughout the database.
Every company was then contacted by the Center about data that was linked to them. Each were sent two certified letters and then contacted several times by Center staff. In total, more than 200 letters were mailed and close to a thousand e-mails and phone calls were made.
Some companies quickly replied to the Center’s letters. One company hired an outside consultant to review the data. Several offered “no comment” responses and many abruptly ended phone calls and shot back curt e-mails, complaining about EPA data.
Thirty companies challenged connections to 188 Superfund sites, saying either they had sold their Superfund liabilities to other companies or had informed the EPA that they were not responsible for particular sites. About 20 other companies stated that they had no records involving 130 Superfund sites connected to them by the EPA data.
To estimate the number of people living near Superfund sites, the Center conducted an analysis using methods common among academics and government administrators. An artificial 10-mile buffer was created around site locations provided by the EPA. All population blocks from the 2000 census placed within those buffers by the U.S. Census Bureau were judged to be within 10 miles of a Superfund site.
The Center filed two key Freedom of Information Act (FOIA) requests to the EPA for the data used in the Superfund Today story — one for a database that tracks fines and penalties assessed against parties who are potentially responsible for Superfund sites, as well as the amount actually recovered on those fines and penalties. The second requested the amount of money the EPA itself spent on cleanups where there was an identifiable PRP, as well as the amount of money actually recovered for work done at those sites. When the EPA cleans up a site where a PRP exists, the agency is supposed to try to recover that money from the PRP.
As a result of the FOIA requests, the Center obtained a database that contained records drawn from the EPA’s Internal Financial Management System (IFMS) and the Comprehensive Environmental Response, Compensation and Liability Information System (CERCLIS).
A full version of the CERCLIS database is also available for download from EPA’s web site. The EPA updates this dataset monthly. The Center downloaded the updates as they became available. The latest was retrieved on March 21, 2007; that report was created on March 20.
The Center used the financial management data to develop statistics that showed a decline in the number of fines and penalties assessed each year and the amount recovered on those fines and penalties.
The same database contained records on cost recovery from PRPs after the EPA cleaned up their sites. The Center’s analysis of the financial data was checked against EPA and other government reports. The overall trends appeared the same and yearly comparisons were often close. Frequently, however, the comparisons were too far off. So for cost recovery numbers, the Center used the EPA’s annual financial statements.
For statistics on the number of cleanups that actually began each year the Center used the part of the CERCLIS database that records specific actions at a site, including, assessments, inspections, and cleanup. By comparing the results of analysis on the CERCLIS dataset with EPA-generated reports, and studying EPA reports and documentation, the Center was able to double-check its methods and results.
For statistics on the number of cleanup actions that took place at each site per year, the Center used the same CERCLIS database listing site actions.
For the number of cleanups that were completed, the Center used the part of CERCLIS data that record details about sites, including location, National Priorities List status and “construction completions.” Results were checked against the simplified data lists available online. “Construction completion” is a term used by the agency when “any necessary physical construction at the site is complete, whether or not final cleanup levels or other requirements have been achieved; or EPA has determined that the response action should be limited to measures that do not involve construction,” and therefore the site no longer poses a threat to the surrounding community. For information and listing of contaminants at sites, the Center used an EPA data report known as “List 10.”
For the number of sites proposed, finalized and deleted to the National Priorities List (NPL), the Center used CERCLIS database and compared results with the EPA’s printed and online reports.
For the story detailing privately sponsored travel by EPA officials, the Center sent FOIA requests for a database maintained by the Office of Government Ethics that collected documents detailing requests by EPA officials and employees to travel. This database contained more than 10,000 trips costing more than $12 million and disclosed to the EPA between October 1997 and March 2006. As with any database, the information listed was not perfect and Center found at least one case where the EPA’s database of trips included travel that was never taken. An EPA ethics official vouched for its accuracy, however.
Center researchers cross-listed this database with other information sets dealing with lobbying, CERCLIS, EPA contracting and federal grants. The Center also requested and received more than 700 pages of forms in which travelers and EPA officials discuss the merits and ethical issues surrounding these trips.
For the lobbying story, the Center acquired its lobbying data by downloading the filings for January 1, 1998 and October 19, 2006, posted at the Senate Office of Public Records [log-in required]. Organizations are obligated to report any money spent on lobbying in filings due twice each year. If the amount does not exceed $10,000, the organization can report "less than $10,000." The Center treated those filings as "$0" when calculating totals.
There are complex and varied inconsistencies in how lobbying money is reported. Specifically, some organizations appear to lump the costs of in-house lobbying activities with money they spend on outside firms into an overall total, as they are directed to do by the Senate's reporting guidelines. But, in hundreds of other cases, organizations appear to be ignoring the guidelines by reporting just their internal lobbying costs.
Since there is no way to tell based on an individual filing whether an organization is reporting its spending properly, the Center has developed a method to account for these apparent differences that results in a best estimate of the total each organization is spending on federal lobbying.
The Center examined each organization's in-house and outside lobby spending totals. Whenever the in-house amount spent by an organization exceeded the amount reported by outside lobbyists it hired, the Center designated the in-house figure as the lobbying total for that six-month period and ignored the outside amount. This was done to avoid the potential for any double counting.
Likewise, if the amount reported by outside lobbying firms hired by an organization exceeded the in-house amount, the Center chose the outside amount as the total for that period, and ignored the in-house money.
In many cases, the figures significantly understate the actual amount of money certain organizations are spending.
For the bankruptcy story, the Center used FOIA requests as well as internal documents obtained from the EPA. The Center staff also retrieved claims filed by the EPA in bankruptcy courts and settlement agreements between the EPA and bankrupt companies.
The Center ran those companies through the EPA’s CERCLIS database to find what Superfund sites they were connected to. In addition, the Center also secured EPA Region 3’s database that details bankruptcy notices the regional office has received since November 1998 as well as the Department of Justice’s data extract of cases they pursued on behalf of EPA and other federal agencies through FOIA requests.
Some records, however, that included law enforcement information were withheld from the Center. For at least EPA Region 3, the Center was able to determine how many notices the agency was aware of and how many resulted in court actions.
For the EPA contracts story, the Center had the National Institute for Computer Assisted Reporting (NICAR) do the data analysis for the section on contracts. to examine the EPA’s contracting trends. The Center acquired from NICAR a list of the EPA’s 250 largest financial recipients of contracts between 1998 and 2005. Using this, the Center focused on contractors that received at least $250,000 during the period of our study.
NICAR mined the data for various kinds of contracts (including no-bid and cost-reimbursement contracts). They also analyzed the data for small business and minority contracts.
The Center monitored fluctuations of the list of Superfund sites that the EPA designates as "not under control," by downloading the data from the EPA Web site monthly and comparing more recent versions to previous versions for the human exposure story.
For the number of contaminants found at these sites, the Center obtained EPA data through a formal request. The Center also submitted Freedom of Information Act requests for correspondence between the EPA and Congress.
Project Manager: Alex Knott
Writers: Kevin Bogardus, Anupama Narayanswamy, and Joaquin Sapien
Database Editor: Helena Bengtsson
Project Database Editor: Richard Mullins
GIS (Map) Analysis: Ben Welsh
Researchers: Tom DeCesar, Sarah Laskow and Devin Varsalona
Editor: Diane Brozek Fancher
Editorial Projects Coordinator: Leah Rush
Copy Editor: Marcia Kramer
Research Editor: Peter Newbatt Smith
Senior Graphic and Web Designer: Jyoti Sauna
Lead Web Developer: Satya Yalamanchi
"Wasting Away" was made possible through generous grants from the Richard & Rhoda Goldman Fund.
Core support of the Center for Public Integrity was provided by:
Carnegie Corporation of New York
John S. and James L. Knight Foundation
John D. & Catherine T. MacArthur Foundation
John & Florence Newman Foundation
Park Foundation, Inc.
Rockefeller Brothers Fund
Supporters and members of the Center for Public Integrity