The EPA's Irrational, Uncompliant Benefit-Cost Analysis Requires Stronger Enforcement Mechanisms

The Environmental Protection Agency (EPA) is seeking comments on its interim final rule titled, “Rescinding the Rule on Increasing Consistency and Transparency in Considering Benefits and Costs in the Clean Air Act Rulemaking Process” (hereafter referred to as the interim final rule). The interim final rule will rescind a different final rule (hereafter referred to as the benefit-cost rule) enacted under the previous administration, which imposes certain benefit-cost analysis requirements on EPA regulations implemented under the Clean Air Act (CAA).

I appreciate the opportunity to submit this public comment on the interim final rule. The Fourth Branch Project at the Mercatus Center at George Mason University is dedicated to advancing knowledge about the effects of regulation on society. As part of its mission, scholars conduct careful and independent analysis that employs contemporary economic scholarship to assess regulations and their effects on economic opportunities and societal well-being.

This comment provides information to the EPA about problematic analytical practices at the agency, practices that the benefit-cost rule—which the EPA is currently seeking to rescind—could start to address. The move to rescind therefore raises an alarm about the EPA’s commitment to rational, evidence-based policy. The following are three aspects in which the EPA’s benefit-cost analysis is currently deficient:

1. EPA benefit-cost analysis appears to be irrational.
2. EPA benefit-cost analysis is not in compliance with aspects of existing executive orders and the EPA’s own guidance on benefit-cost analysis.
3. The existing process appears insufficient to address these problems. Hence, the EPA strongly needs further enforcement mechanisms, such as the benefit-cost rule that the EPA is repealing.

EPA Benefit-Cost Analysis Appears to be Irrational

A market failure arises when transaction costs or a lack of property rights prevent beneficial trades from occurring that otherwise would have made society better off. A common list of market failures includes externalities, asymmetric information, underprovision of public goods, and monopoly power. However, in recent years economists have identified another source of market failure, known as a “behavioral market failure.” A behavioral market failure results from any one of an array of cognitive biases that lead individuals to act in a manner that is not perfectly in alignment with their own interests.

A market failure occurs “when the outcome of an economic transaction is not completely efficient” (emphasis added). The word “completely” is important here, because any market that has not totally exhausted all beneficial gains from trade, thereby resulting in a Pareto efficient competitive outcome, is said to fail according to the standard, neoclassical theory. This logic can be extended to the realm of behavioral economics. An action can be judged to be irrational if it is not perfectly rational. Such an action therefore constitutes a failure of rationality and can be labeled a behavioral market failure.

In recent years, scholars have also noted that the same cognitive biases that lead to irrational behavior in the marketplace also affect regulators in the government, thereby leading to “behavioral government failures.” This development should not be surprising, given that regulators are human too. The EPA’s current regulatory analysis practices exhibit a plethora of cognitive biases in action, amounting to a case of behavioral government failure. This fact has relevance to the benefit-cost rule, which the EPA is repealing, because it was presumably intended to address the EPA’s less-than-adequate CAA economic analyses.

Neglecting opportunity costs, anthropomorphizing society, and excessive short-term thinking are three examples of cognitive bias that repeatedly show up in EPA benefit-cost analyses.

Opportunity Cost Neglect

Opportunity cost neglect occurs when individuals do not “explicitly consider the outside goods that an expenditure displaces” or, more generally, when there is an “underweighting of opportunity costs” in decision-making. EPA regulatory analysis systematically fails to account for opportunity cost adequately because the EPA confuses two analytical concepts: the social discount rate and the opportunity cost of capital.

It is incorrect to use a social discount rate for the purposes of accounting for the opportunity cost of capital for “projects with benefits that are nonpecuniary, such as investments that reduce health or environmental risk.” In such cases, the appropriate tool is a shadow price (or, similarly, a marginal cost of funds factor). Virtually all EPA regulations have benefits that are nonpecuniary because EPA rules address health or environmental risk. Yet EPA almost never uses the shadow price method, which is the correct way to account for the opportunity cost of capital for environmental regulations.

Consequently, EPA analysis is systematically biased in favor of regulations, because capital investments that are displaced through compliance with EPA regulations receive too little weight in analysis relative to the nonpecuniary health or environmental benefits that the EPA strives to achieve with its rules. Consider the following simple example: a regulation displaces $1 million worth of capital investment next year in order to achieve $1 million of nonpecuniary environmental benefits the same year. If one assumes that no other benefits or costs accrue from this regulation, then according to current EPA practices, these benefits and costs would be discounted at the same rate and thus would have the same present value. Therefore, regardless of what social discount rate is used, the net present value of this regulation would be zero, according to the current EPA practice to discount all benefits and costs using a single discount rate, without adjusting the value of capital investment in any way.

This line of reasoning suggests that the EPA would presumably be indifferent between the hypothetical regulation going into effect and it not going into effect (since the policy has a net present value of zero). But this clearly cannot be right. Capital investment has a higher opportunity cost than an equivalent dollar value of consumption or nonmarket investment. By the EPA’s logic, society should be indifferent between investing 70 percent of national income and consuming it, yet these alternatives have radically different implications. By most reasonable standards, the hypothetical regulation should fail a benefit-cost test. But by the EPA’s standard, it would not fail such a test. The EPA’s analysis gives too much weight to nonpecuniary outcomes relative to capital investment; therefore, displaced investments are underweighted (resulting in opportunity cost neglect).

In the interim final rule preamble, the EPA notes that the benefit-cost rule would have “expanded the universe of CAA rulemakings for which the EPA would be required to conduct BCAs [benefit-cost analysis] without justifying why such expansion was necessary or appropriate.” The benefit-cost rule would have expanded the use of benefit-cost analysis beyond “economically significant” rules to “significant” CAA rules as well. The EPA claims the previous administration failed to justify this expansion. However, as a result of opportunity cost neglect, many rules labeled as significant are likely to be economically significant for the simple reason that the EPA as a general practice ignores or underweights the compounding returns of capital investments. The requirement for increased scrutiny of significant rules therefore seems necessary, reasonable, and appropriate.

Anthropomorphic Bias

Anthropomorphism occurs when people “ascribe human properties to non-human things.” Anthropomorphic bias “arises as a consequence of our consciousness of selves and others as embodied beings.” Thus, people have a tendency to attribute human characteristics such as purpose and agency to nonhuman objects, entities, and systems, which have no such characteristics. Anthropomorphic bias “can be understood as an innate existential tendency of human embodied thought, thereby presenting a potential problem to the fields of the philosophy of science and embodied cognition, and to social scientific experimental design and interpretation.”

This type of bias has relevance to EPA benefit-cost analysis because the EPA ascribes characteristics of individuals to society in its regulatory analysis. For example, Office of Management and Budget (OMB) Circular A-4 notes that the 3 percent social discount rate employed in virtually all EPA regulatory analyses represents “the rate at which ‘society’ discounts future consumption flows to their present value.” Of course, society does not discount anything because society is not an individual with agency or preferences, including time preference. Yet the EPA routinely makes this mistake in its regulatory analyses. (The EPA’s mistake is also a fallacy of composition, which occurs when someone assumes that something true at one level of analysis is also true at a more aggregated level of analysis.)

The EPA’s position that society possesses characteristics of individuals likely stems from the Ramsey growth model, which is often referred to in the literature on social discounting (and in EPA guidelines). In the Ramsey framework, analysts adopt the perspective “of a social planner who weighs the utilities of present and future generations or those preferences of the current generations regarding future generations.” This social planner exists outside of society, as if choosing from a place of impartiality, and this abstraction can be thought of as a proxy for the current generation’s welfare. However, evidence and science should be at the center of the EPA’s decision-making, not a desire to please an abstract social planner or an embodied notion of society.

Present Bias

Economists Ted O’Donoghue and Matthew Rabin define present-biased preferences as follows: “When considering trade-offs between two future moments, present-biased preferences give stronger relative weight to the earlier moment as it gets closer.” The EPA’s benefit-cost analysis is an example of present bias in action, because the analysis gives complete weight to the present moment. Thus, as two moments get closer, when the closer moment becomes the present, it gets complete weight in the EPA’s analysis, whereas the future moment gets no weight (or, alternatively, the future moment gets whatever weight people in the present moment decide to give to the future moment).

Benefit-cost analysis has a grounding in general equilibrium theory. However, the notion of a competitive general equilibrium is itself biased toward the present. As Nobel laureate Gerard Debreu notes, “the economy is considered as of a given instant called the present moment.” Thus, the economy can be understood “as a stationary state, in which all choices are made once and for all” in the present moment. Time enters the analysis to the extent that goods are distinguished from one another in terms of their location and delivery date, but the choice of how to allocate resources is based on preferences and production sets in a single instantaneous moment: the present. General equilibrium theory, and benefit-cost analysis that is based on it, is a static, present-biased framework.

The EPA, in the preamble to the interim final rule, notes that the benefit-cost rule “was not necessary,” that it “lacked a rational basis,” and furthermore that the previous administration “failed to articulate a rational basis for” it. What the earlier discussion highlights is that it is the EPA’s own benefit-cost analysis that often lacks a rational basis. Stronger enforcement mechanisms are almost certainly needed to correct for the cognitive defects found in EPA analysis, including analysis for CAA rules. Notably, all of the deviations from rationality discussed here relate in some form to discounting. Perhaps this is not surprising, given that “many of the biases identified by behavioral economists, who study the intersection of economics and psychology, relate to situations involving tradeoffs that occur over time.”

EPA Benefit-Cost Analysis Fails to Comply with Aspects of Its Own Guidance and Existing Executive Orders on Regulatory Analysis

The EPA claims that the benefit-cost rule could have resulted in misleading net benefit calculations because, for example, of the way the that the benefit-cost rule would have required the EPA to report benefits information by transparently breaking it down into its different subcomponents. However, the EPA’s analysis already produces misleading net benefit calculations that are not consistent with what its own interim final rule claims EPA analysis measures.

The EPA states in the interim final rule that its “BCAs assess economic efficiency by asking whether it is theoretically possible for those who gain from the policy to fully compensate those who lose and remain better off. When the answer to this question is ‘yes,’ then net benefits are positive.” Intratemporally, it is true that the costs and benefits of EPA air rules are measured in terms of consumer willingness to pay or something similar and are expressed in monetary units. However, intertemporally, benefits and costs are converted into units on a common social welfare scale through the practice of discounting. These social welfare units are not equivalent to money and therefore do not answer the question of “whether it is theoretically possible for those who gain from the policy to fully compensate those who lose and remain better off.” Thus, the EPA’s benefit-cost analysis does not measure economic efficiency as defined by the EPA itself.

This lack of a coherent economic measure also means that the EPA violates at least two executive orders related to regulatory analysis. Executive Order 12866 requires “an assessment of the potential costs and benefits of the regulatory action,” for all significant rules. Executive Order 13563, issued by President Barack Obama in 2011, reaffirms the principles of Executive Order 12866, as did a recent memorandum from President Joe Biden. As Circular A-4 notes, “In presenting the stream of benefits and costs, it is important to measure them in constant dollars” (emphasis added). According to EPA guidelines, the benefit-cost “ratio indicates the benefits expected for each dollar of costs.” The EPA violates its own guidelines and the requirements of Executive Order 12866 because “an assessment of potential costs and benefits” should be measured in dollars, not the arbitrary social welfare scale that the EPA is using. The EPA chastised the previous administration for not demonstrating “that an actual or theoretical problem existed.” If the fact that the EPA’s so-called benefit-cost analysis does not measure either costs or benefits isn’t evidence that a problem exists, it is hard to know what would constitute evidence.

The Existing Process Is Insufficient

The EPA has concluded that “existing directives under E.O. 12866 and guidance to conduct BCAs for economically significant rules, while retaining flexibility in analyzing costs, benefits, and other factors for non-economically significant rules, strike the better balance between agency resources and the information provided by economic analysis for such rules.” Furthermore, the EPA notes that “existing guidance affords flexibility for the EPA to conduct the type of analysis warranted by a particular rulemaking.” The EPA concludes, “Therefore, the EPA has determined that the existing process already provides sufficient consistency and transparency.”

Yet, as noted earlier, the EPA’s benefit-cost analysis routinely violates both its own guidance and existing executive orders. In fact, it is not even clear what EPA analysis is measuring. Moreover, this situation creates questions about the effectiveness of OMB review of EPA analysis, which serves the role of enforcing regulatory analysis guidelines and ensuring minimal quality control. More specifically, despite the Administrative Procedure Act, OMB review, provisions specific to the CAA, EPA economic analysis guidelines, and the fact that “presidents since the 1970s have issued E.O.s directing agencies to conduct analyses of the economic consequences of regulations,” the problems highlighted earlier have been allowed to continue and fester for years with seemingly no course correction. It is difficult to lend credence to the belief that the existing process strikes a better balance or that EPA should be afforded flexibility to deviate even more from analytical best practices as it sees fit for particular rulemakings. Stronger enforcement mechanisms are surely needed to get the EPA’s analytical ship back on course.

Conclusion

In the preamble to the interim final rule, the EPA criticizes the previous administration, stating that it “failed to point to a single example of a rule promulgated under the CAA where problems emerged.” In fact, much more difficult than identifying instances of problematic regulatory analysis practices would be identifying instances of rational and coherent regulatory analysis conducted by the EPA. To EPA policymakers who seek examples of regulations that suffer from the problems described earlier, I refer to those rules from the EPA that appear in the annual OMB report to Congress on the benefits and costs of federal regulations. With perhaps only a few exceptions, virtually all of the EPA CAA analyses suffer from versions of the issues discussed here.

The EPA is developing a dangerous attitude and hostility to science, which does not seem consistent with a recent memo from President Biden on “Restoring Trust in Government through Scientific Integrity.” The benefit-cost rule was intended to shore up some of the analytical failings at the EPA’s Office of Air and Radiation. By any reasonable standard, the EPA’s benefit-cost analysis practices have gone awry.