OIRA's Fuzzy Math on Coal Ash: A Billion Here, a Billion There

by Rena Steinzor

July 13, 2010

This post was written by CPR President Rena Steinzor and Michael Patoka, a student at the University of Maryland School of Law and research assistant to Steinzor.

Last October, the EPA proposed to regulate, for the first time, the toxic coal ash that sits in massive landfills and ponds next to coal-fired power plants across the nation. The 140 million tons of ash generated every year threaten to contaminate groundwater and cause catastrophic spills, like the 1-billion-gallon release that devastated Kingston, Tennessee in 2008. The EPA recommended that coal ash be listed as a subtitle C “hazardous waste,” making it subject to federally enforceable disposal requirements under the Resource Conservation and Recovery Act (RCRA). But by the time that the Office of Information and Regulatory Analysis (OIRA) was through “reviewing” the agency’s proposal, the rule had been watered down to suggest a choice of three alternatives: the original proposal as well as two much weaker options. In addition, the accompanying regulatory impact analysis (RIA) had ballooned into an impenetrable, strictly quantitative cost-benefit analysis—one that systematically underestimates the benefits and overestimates the costs of subtitle C regulation.

Among the many shortcomings and biases of the RIA, of particular concern is the industry-fueled suggestion that a subtitle C listing would “stigmatize” the beneficial use of coal ash (in products like concrete and wallboard, as well as in road beds and farmlands), even though beneficially used coal ash would remain officially exempt from the rule. Because beneficial use is associated with valuable economic and environmental benefits, this highly speculative “stigma effect” is estimated to cost $233.5 billion in lost benefits—a cost that dwarfs any expected benefits of regulation by several orders of magnitude.

Combing through the RIA, one is frequently struck by a sense of futility. Even if the rule, for instance, were to prevent thousands more cancer cases than estimated, the benefits would still not be enough to outweigh the insurmountable “stigma” cost. Indeed, one could be forgiven for skimming through the analysis of benefits, in light of its inconsequentiality.

But that would be a mistake. A close examination of the benefits analysis reveals an assortment of technical errors that substantially understate the risks posed by coal ash. Even if the ultimate balance of costs and benefits would remain relatively unchanged, such errors should not go unnoticed. At the very least, they point to the roughness and unreliability of an analysis that transforms real-world threats into a tangled maze of abstract number-crunching.

The RIA’s prediction of future coal-ash spills provides a number of illustrative errors. Based on a timeline of recent spills, the RIA builds two separate statistical models of future spills, one for “significant releases” (between 1 million and 1 billion gallons) and one for “catastrophic releases” (1 billion gallons or more). From a survey of utility companies, the RIA identifies 42 releases of coal ash from wet ponds (or surface impoundments) that occurred in the past 15 years. The relevant releases (one catastrophic and five significant) are averaged over the 15-year period to obtain projected spill rates.

There’s only one problem with all this: the survey question itself asked utility companies to report spills that occurred within the last 10 years, not 15. Presumably, the RIA assumed a 15-year period because one of the 42 releases is reported as occurring in 1995, so that the list of releases appears to span the period of 1995-2009. Of course, the fact that one utility company might have disclosed a spill outside of the scope of the question is no justification for widening the time period of the other reported releases by five years, which artificially lowers the predicted spill rate.

Even more alarming, though, is that the “1995” spill at the root of this error—the 100-gallon release at the Colstrip Steam Electric Station owned by PPL Montana LLC—actually occurred in 1999, according to PPL’s survey response. Apparently, the year was changed to 1995 due to a typographical error in transcribing that survey response into the database of results.

Using a 10-year period instead, the number of Kingston-like catastrophes predicted over the next 50 years would rise from 3 to 5, and the number of significant releases predicted would rise from 17 to 25. The monetized benefits of avoiding spills under subtitle C regulation would increase by $881 million (at a 7-percent discount rate).

Furthermore, for 27 of the listed spills, the amount of the release is designated as “unknown” because the utility companies failed to specify how many gallons were spilled in their survey responses. But by excluding all the “unknowns” from the predictive model, the analysis implies that none of them were significant—an affirmative assumption that is no more justified than assuming that all of them were significant.

Moreover, the analysis is woefully incomplete in relying only on the companies’ vague survey responses, without demanding more detailed information from the companies or conducting even the slightest independent investigation of recent spills. For instance, a simple Internet search reveals that the 2002 spill at Bowen Power Station in Georgia, whose magnitude is listed as “unknown” in the RIA, released more than 2 million gallons of coal ash. A spill of this magnitude would qualify as “significant,” and thus, with this one additional detail, the average number of predicted significant spills would rise from 25 to 30, and the average benefits of avoiding spills would increase by $19.6 million. How many other spills of “unknown” magnitude could turn out to have been “significant”?

These are but a handful of the errors that tend to get obscured as surreal quantities of Monopoly money are shuffled about in the cost-benefit analysis. A billion dollars is bound to get lost here or there, due to sloppy calculations. The fact that correcting these errors barely makes a dent in the final balance is a reflection of the inherently distorted values of a method of analysis (cost-benefit) that seeks to measure all things in terms of dollars and cents. Here we have a regulation intended to prevent the obliteration of lives, homes, communities, and the environment by massive spills of toxic sludge, a hazard that has been made all too clear by the Kingston disaster. And yet, out of concern for the industry’s interest in maintaining a multi-billion-dollar market for recycled coal ash, and couching such concern in optimistic predictions of huge environmental benefits from coal-ash recycling (never mind its potential dangers), the cost-benefit analysis completely overrides the imminent environmental objective. It does, however, cast the problem in stark relief: this is simply a case of industry profit pitted against communities and lives.

In any case, if OIRA is bent on using cost-benefit analysis to highlight the “burdensome” nature of even modest proposals for much-needed health and environmental regulations, and if it insists on assigning such a central and determinative role to quantitative estimates—then the least it can do is double-check its own numbers.

Untitled Document

An Error Has Occured

CPR Developers have been alerted. Apologies for the inconvenience.