Pollution control involves ambitious goals based on tenuous information about the benefits and costs of abatement. The vast economic literature on the subject, which includes this new book by economics professors Nicholas Muller (Middlebury College) and Robert Mendelsohn (Yale University), nearly universally recognizes that problem, but equivocates on its implications. The dominant view is that environmental threats are so clear that, even with the uncertainties, it is desirable to reduce emissions and improve enforcement. (Perhaps surprisingly, many adherents of this view are comfortable suggesting policies that explicitly consider the benefits and costs of abatement.) More skeptical critics suggest that pollution impacts are so complex that little basis exists for setting optimal policies and that existing policies extend control far beyond clear major threats.

Impediments to good analysis / How should we reconcile standard economic theories of optimal pollution control, the information available to implement control, and actual pollution regulation? A critical and widely recognized problem is that both the Clean Air Act and Clean Water Act deliberately employ dangerously loose significant-impact criteria that rule out explicit consideration of the costs and benefits of abatement, thus seemingly mandating pollution suppression regardless of cost. Literal application of the two laws would be both economically and literally fatal because eliminating all health-damaging pollution stops all economic activity and thus provision of basic sustenance. In practice, the EPA typically mandates the use of “best available” pollution suppression, which at least limits controls to what is supposedly technologically feasible and commercially available.

A second, also-well-noted problem with pollution policymaking is that purported pollutants are emitted from many sources, travel and transmute, and then settle someplace. This produces enormous problems of unraveling impacts. At best, impact measurement would require undertaking elaborate studies of the continuous actual exposure of representative samples of human populations to pollution and all other influences upon their health. At worst, ethically indefensible dose-exposure laboratory tests might be undertaken. Unfortunately, the looser standards actually used allow reliance on far less precise measures.

Given our lack of adequate knowledge, how does the EPA follow presidential and congressional orders to measure the benefits and costs of pollution control? While many supporting studies are considered, the benefits always are predominantly the reductions of premature mortality from decreased exposure to small particles. To estimate those benefits, the EPA leans heavily on just two one-time, non-random-sample surveys with highly atypical characteristics. (See my “The EPA’s War on Coal?” Spring 2013.) Updates are made only to examine the subsequent mortality experience of people originally sampled. The consequence is that, for whatever purpose a rule is proposed, the calculated side benefits from reduced mortality almost always vastly exceed the costs. To make matters worse, the EPA regularly warns that its impact studies have nothing to do with policy design.

In addition to the looseness of its damage criteria, pollution policy has political defects as well. The first is the enormous inertia inherent in lawmaking. The U.S. Congress characteristically badly sketches objectives, delegates the implementation to a regulatory agency, almost never returns to reevaluate its legislation, and when it does, reacts to failures from overreach by adding complications advertised as fixes that universally worsen the situation. Thus, the laws keep adding new goals and control requirements—all of dubious merit.

The laws and basic constitutional principles correctly make rule implementation an extremely complex, time-consuming process. The EPA must undertake extensive studies to justify actions and then solicit extensive comments about proposed regulations. After the agency issues rules that respond to the external comments, judicial review occurs. The complexities make for extremely slow decisionmaking.

Thus, any desire to introduce benefit-cost analysis into pollution control faces the double problem of legal prohibition of such quantification and lack of the knowledge needed to set a correct pollution-control level or charge. That ignorance also implies that any control rule, including all those now in use, has inadequate justification. This leads to difficulties that are universally finessed in suggesting pollution control improvements. (See “Uncertainty Can Go Both Ways,” Summer 2013.) Clearly, in the absence of satisfactory knowledge of the benefits and costs of emission control, no discipline can provide uncontroversial policy advice.

Implications for environmental policy analysis / Given the lack of health effects knowledge and the prohibition on the consideration of costs in environmental statutes, much more skepticism should be shown about what concerns are worth regulating. We are far from knowing how to set the levels of optimal charges or limits that textbook economic theory suggests should prevail, and the enforcement milieu precludes the steady adjustments to error that are essential.

Any study of pollution control should have some recognition of those limits. That is true for government reports, pure scholarly research, and especially for efforts such as Muller and Mendelsohn’s—presumably designed to influence public policy debates. Their book moves relentlessly and too rapidly. At every point where common sense and the literature suggest warning flags, Muller and Mendelsohn move on.

Their work illustrates a chronic problem in contemporary economics of encouraging elaborate formal analysis without concern for the feasibility of actual implementation. Critics of economic policy analysis argue that publication in prominent journals is overemphasized and the journals favor form over substance. While those criticisms are overdone, Muller and Mendelsohn do provide an example of empty formalism. The book’s origin as an expansion of an American Economic Review paper is a further indication of problems. (An irony is that Muller is now a colleague of David Colander, a leading critic of the stress on overly abstract journal articles.)

Sophisticated formalism and its defects / The authors argue that existing knowledge allows the calculation and imposition of specific-source optimal taxes on multiple pollutants. The bulk of the book is devoted to sketching and utilizing a model to measure optimal taxes. Scattered through the book are a few sensitivity analyses. All of those tests suggest strong reasons to doubt the validity of the proposed pollution control taxes. Yet, the authors slide past those warnings and argue their modeling serves as the practical basis for better pollution-control approaches.

The formal analysis starts with mathematical derivations of the standard rules of optimal pollution control, presents the “simplified” air pollution dispersion model used, rushes through listings without evaluating the equations used to quantify the impacts of the pollutants, and summarizes the results of about 60,000 calculations of the marginal damages of six pollutants at almost 10,000 sites. It shows the sensitivity analyses run on four power plants, moves on to argue that Muller and Mendelsohn’s model shows how to improve the sulfur dioxide quota-trading program under the 1990 Clean Air Act Amendments, and proceeds to provide purported indicators of priority areas for regulatory reform. The book then offers three chapters on the total damages from pollution.

Every step cries out for further discussion and qualification, which are not provided. The omissions include the overwhelming practical questions about the generation, transmission, and effects of pollution, the applicable laws, their implementation, the insuperable problems of satisfactory quantification, and the extensive criticism that governmental environmental policy and its purported quantified justifications have inspired. The most basic omission is recognition of the statutory reasons why the EPA does not stress economic efficiency.

The methodology uses a simpler emissions dispersion model than the EPA uses and tests whether the simplification is an adequate approximation of the EPA model. The EPA model is left unexplained; the approximation used is represented by two largely unexplained equations. Predictably, a good fit between the approximation and the more complex model is used to justify the simplification. The authors fail to consider the possibility that the larger model is defective.

The sensitivity analysis is a narrow mechanical exercise. After piling on a host of assumptions about the underlying data, Muller and Mendelsohn undertake multiple random samples from the data relating to four different actual power plants to simulate the distributions. Despite often finding that the standard deviations exceed the means by such large amounts that negative benefit values are highly probable, the authors express confidence the means estimated are useable.

A serious defect of the book is the superficial, fragmented treatment of mortality damages. On p. 4, we learn that mortality damages “comprise the bulk of air pollution damages.” On p. 44, in the impact-measurement description offered in Chapter 3, the book indicates that, for estimates of the numbers of mortalities, it relies on the same studies that the EPA employs in its cost-benefit analyses. Only on p. 111 are the calculated mortality effects shown. A sensitivity analysis on p. 117 discloses that Muller and Mendelsohn used mortality-evaluation methods radically different from the EPA’s, which means the authors used lower values for lost lives and they adjusted the values for the age of the decedents. The EPA estimates were 630 percent higher.

The only methodological concern addressed in Chapter 3 is rejecting objections to valuing lives. This is detrimental to the authors’ argument because so many of the sources they use to determine that value are defective. Muller and Mendelsohn rely on the same American Cancer Society (ACS) health effects study that provides the basis of the EPA’s work. The EPA also sponsored a Health Effects Institute (HEI) review of both the ACS study and an even less satisfactory one by the Harvard School of Public Health. The HEI study clearly indicates the fragility of the impact estimates. (Muller and Mendelsohn have cited the HEI study in the online appendices of two of the journal articles that are precursors to this book. My “EPA War on Coal?” article reviews the study’s defects.)

Given those knowledge gaps and the nature of the EPA, Muller and Mendelsohn should have exercised more caution about proposing practical use of their pollution-abatement methodology. Questions arise first about the wisdom of replacing the EPA’s model of pollution dispersion with an approximation. What suffices for an academic enterprise seems inadequate for policymaking. Even the EPA model may have limitations about what it does measure and necessarily ignores that people are sheltered from outside air most of the time.

The next concern is whether the EPA or any public agency should be trusted to select the best estimates of health impacts. Moreover, the Muller-Mendelsohn abatement-benefit calculation approach inspires skepticism. Their method is to add one ton of each pollutant at each site 60,000 times to calculate all the marginal benefits. It strains credulity that any model is precise enough to produce reliable estimates of the change in value. Moreover, no consideration was given to the costs of enforcement.

Their discussion describes the emission-trading element for sulfur dioxide in the 1990 Clean Air Act Amendments and then introduces a problem with the trading scheme that they claim to remedy: that the location and not just the magnitude of damages affects the benefits. Calculations are made first of the benefits of attaining the emissions target set by the amendments through shifting to quotas reflecting locational differences in damages. Then an estimate is made of how benefits are further increased by shifting to the much stricter emission limits that their model alleges is efficient. This is simply repetition of the EPA’s standard contention that any reduction in particulate emissions produces large benefits.

The book then offers four chapters summarizing aspects of the calculations. The first stresses pointless calculations of the largest gaps between marginal abatement benefits and marginal abatement costs. In this chapter, they note but ignore that both the size of the gap and the size of emissions determine the importance of different emissions. The calculations supposedly are to assist some unnamed regulator free to set priorities without recognition that an act of Congress is needed.

The next three chapters present calculations of the total cost of pollution disaggregated in several ways. The estimates rely on underlying valuations of the cost by pollutant and source. Each marginal damage figure is multiplied by the quantity of pollutant emitted to produce the value of damages by that pollutant at that source. Many aggregations are possible, and several are shown. Below, only those seeming particularly important are noted.

The first aggregation is by pollutant and type of damage. It shows that premature mortality is the dominant harm: $90 billion out of a $109 billion total. The authors add estimates, derived from another author’s figures, of damages from carbon dioxide. A dubious set of sensitivity analyses is provided. Four of those analyses deliberately produce much higher damage estimates—three involve assuming higher values to lives saved and the fourth employs another epidemiological study beloved by the EPA because the study reports much higher mortality effects than the other available study. (The HEI review found that the second study was unacceptably unreliable because it only covered six “cities.”) A clear explanation of the data sources used is lacking here and in the next two chapters.

The remaining calculations are of the damages by sectors of the economy. The first covers all of the 20 broad sectors into which the EPA divides the total economy. Carbon dioxide figures are given for the coarser five-sector subdivision used by the emissions data source, the U.S. Energy Information Administration. Since the EPA cannot assign transportation uses to user sectors, Muller and Mendelsohn provide an alternative sectorial breakdown that disaggregates transportation into several sectors, aggregates industry into two, and adds agriculture and residential. This discussion, unlike the initial breakdown, covers all emissions measured. Chapter 10 then presents tabulations of the damages caused by subsectors that produce high damages, such as coal-fired electric power plants. The tables show both their valuations and those arising from adopting the EPA’s assumptions about mortality.

Despite the wild differences in results arising from changing the cost-of-mortality estimate, the authors claim “we demonstrate that it is possible to provide reliable measures of pollution in national accounts.” (The pollution cost of coal generation of electricity is estimated to be between $18.7 billion and $142 billion.)

The book concludes with a reiteration of the authors’ basic premise that their efforts produce a useable method to move closer to the “textbook” rules of efficient pollution control.

Conclusion / Change in air pollution regulation is desperately needed. The cure is better recognition in the underlying law of the vast uncertainties involved and the great difficulty, if not impossibility, of resolving them. Those supposedly worried about excessive intervention should not support the pretense that some giant computer model can overcome these defects easily. Everything in the book reinforces preexisting skepticism about such claims. Thus, this effort at practical reform serves as an illustration of the barriers to improvement.