Is Cost-Benefit Analysis Needed?
Cost-benefit analysis (CBA) is often touted by the administration and conservative think tanks as a neutral tool in policymaking, but recent studies by legal scholars show that CBA is inherently political and may even advise against what we consider our most immutable public protections.
CBA is a policymaking tool by which the costs of imposing a regulation are weighed against the potential benefits of reducing the harm. For example, in the case of pollution regulation, cost is generally construed as the cost of implementing technology to comply with regulation. These costs are more easily quantifiable than other factors, although some evidence exists that costs are often inflated.
The benefits of a regulation require two separate analyses: an assessment of the risk posed by the harm in question as well as a monetization of the potential benefits. Both factors prove to be difficult to calculate; many benefits resist monetization, and risk assessments can be hindered either through incomplete datasets or a large degree of indeterminable factors. In order to estimate the health effects of a regulation, for example, agencies generally must rely on laboratory data on other species or on human experience with much higher levels of exposure. To extrapolate from this data the potential benefits of a regulation requires a large degree of guesswork, and agencies often come up with wide ranging numbers on the potential health benefits.
Proponents of CBA believe that regulators are irrational in their policy decisions and that CBA is necessary in developing regulations because it acts as a neutral, rational tool, evenly weighing all considerations, but many public interest advocates have argued that cost-benefit analysis unfairly targets environmental, health, safety and other social regulation and will always favor lowering costs rather than creating more stringent protections. If cost-benefit analysis really is a neutral tool, then it must be equally capable of siding with more stringent regulation as it does weaker regulation. Three recent articles examine the neutrality of CBA both in theory and in practice and analyze the arguments of CBA's greatest proponents. Lisa Heinzerling, Frank Ackerman and Rachel Massey's "Applying Cost Benefit Analysis to Past Decisions: Was Environmental Protection Ever a Good Idea?," David Driesen's "Is Cost-Benefit Analysis Neutral?," and Richard Parker's "Is Government Regulation Irrational? A Reply to Morall and Hahn" all find that neither in theory nor in application is cost-benefit analysis a neutral tool. Moreover, this "simplistic scorecard" fails to embody our national regulatory priorities.
The "Simplistic Scorecard"
Advocates of cost-benefit analysis claim that regulation is irrational and that cost-benefit analysis is necessary to rein in costly, burdensome measures. Yet as Richard Parker points out in "Is Cost Benefit Analysis Irrational?," the arguments of CBA's biggest supporters are themselves irrational, relying on fuzzy numbers and misguided assumptions to prove the case for this weak policy tool.
Some of the most compelling cases for the necessity of cost-benefit analysis is necessary in policy-making have come from John Morrall, an economist at the Office of Management and Budget (OMB), and Robert Hahn, of AEI-Brookings Joint Center for Regulatory Studies. Both have argued, after applying cost-benefit analysis to the totality of government regulation, that the costs of federal regulation far outweigh the benefits, proving, they assert, that government regulation is fundamentally irrational and overzealous. They point to what appear to be egregious examples of overly-cautious regulations, citing cases in which regulation costs up to $72 billion for every life saved. Yet, as Parker easily points out, their arguments are based on shaky assumptions that fail to take into account the full range of considerations necessary for an agency to make a rational policy decision.
For all the lip service on hard numbers, both Morrall and Hahn's data contained many undisclosed assumptions, and calculations and methodologies were often not revealed. Parker observes that Hahn's gaps are particularly noteworthy: "Hahn's original studies do not so much as list the rules in his database." Hahn also offers no documentation of his calculations. After soliciting Hahn, Parker finally received Hahn's data, but his Excel spreadsheet only gave a tally of costs and benefits and his final calculation, without indicating any of the assumptions Hahn made to arrive at numbers he used.
Even without being able to reproduce the results, it is clear that Morrall and Hahn make assumptions that are biased against regulation. In fact, as Parker points out, the data set that Morrall chooses to work with focuses on some of the most extreme cases of costly regulation. Morrall chooses an arbitrary set of 16 highly costly regulations in order to make his claim that regulators make irrational choices. Most of these regulations have to do with just a handful of pollutants, which have, as Parker notes, "generated some of the most heated and heavily litigated controversies in all of environmental law." Clearly Morrall's case examples do not represent a neutral dataset.
Morrall and Hahn also choose to use a very high discount rate, which discounts the future costs and benefits of a regulation using a constant exponential rate. Though many scholars disagree on the practice of discounting benefits at all, there is little consensus on a discount rate higher than three percent. Yet Morrall chooses a discount rate of 10 percent and Hahn uses discount rates at three, five and seven percent. Such rates can dramatically reduce the potential benefit of a regulation. As Parker points out, "discounting a constant stream of benefits over 25 years will reduce its present value by 30 percent at a three percent discount rate, or 50 percent at a seven percent discount rate."
Morall and Hahn also assume a latency period in the calculation of certain benefits. For risks such as cancer which may not manifest themselves for a number of years, they add in a latency period to their calculation of benefits, which greatly reduces the present benefit of avoiding a future harm. As Parker points out, if one assumes "that the stream of cancer risk reduction benefits (which dominate many of the health and environmental rule benefit numbers) only begins to accrue after a latency period of some 15-35 years, the impact of discounting can become truly enormous. Discounting a constant dollar annual benefit accruing over 25 years -- beginning 35 years out -- will effectively shrink benefits by a factor of four at a three percent discount rate, and a factor of twenty at . . . [a] seven percent rate." This assumption distorts their calculation of benefits, and once again, the distortion favors the regulated community.
Morrall also arbitrarily chooses to reduce benefits. Rather than relying on agency calculations to calculate benefits, Morrall chooses numbers from published studies that significantly reduce benefits. When agencies present a range of possible benefits, he arbitrarily chooses from the bottom of the range, without explaining his methodology. Morrall adjusts agency numbers on health effects, risk assessments, and effectiveness, in each case swaying the balance in favor of less regulation. That Morrall should do independent analysis of factors is not the issue, but Morrall's choice to manipulate the equation without an explanation or rationale leaves observers guessing as to how the numbers were attained.
Ignoring Unquantified or Non-Life-Saving Benefits
Morrall and Hahn also choose to ignore benefits that agencies did not quantify or benefits that can be quantified but not given a dollar value. For instance, Morrall gives OSHA's formaldehyde rule a price tag of $72 billion per life saved. Though the formaldehyde rule may only save one life a year, it has many other significant benefits that are not included in Morrall's calculation, including, according to Parker, "reduced or avoided burning eyes or noses, sore or burning throats, asthma attacks, chronic bronchitis, allergic reactions, dermatitis and skin sensitization. OSHA notes that over 500,000 American workers are regularly exposed to formaldehyde at concentrations that have been found to cause one or more of these illnesses or discomforts." By focusing only on lives saved, Morrall's calculations miss the harm that a regulation is actually averting, one that resists his simplistic calculations.
Hahn ascribes "a zero value to any benefit which the government's regulatory impact assessment did not quantify and monetize." Hahn also ignores benefits that were quantified and monetized by agencies "but which failed to fit within his Procrustean categories of recognized benefit -- reduction of cancer, heart disease, lead poisoning and accidents, and benefits of reduction a handful of air pollutants -- even as he insisted that he was using the government's numbers." Hahn wrongly asserts that leaving out these "non-standard" benefits will have no impact on his calculations.
Parker gives the example of EPA's Great Lakes Water Quality Guidance, which reduces "the discharge of persistent, toxic and bio-accumulative pollutants" in the Great Lakes. The compounds reduced by the regulation are associated with very serious risks, including "neurotoxicity, fetotoxicity, endocrine disruption, hematological impairment, reproductive dysfunction, sensory and equilibrium disturbances," among others. Despite the clear benefits of such regulation, this measure fails cost-benefit analysis because a great many of its benefits cannot be monetized. These non-cancerous risks are difficult to quantify:
Unlike cancer, which is widely assumed to have a linear dose-response down to a zero exposure level (making the calculation of population risk from aggregate exposure data relatively simple), non-cancer endpoints generally have non-linear risk thresholds -- which means that, to calculate a population risk from any given discharge, you have to know not only the exposure of the population to the pollutants issuing from the sources targeted by the particular regulation. You also have to know the cumulative exposure to individuals in the population to these and other interacting pollutants from other sources.
According to Hahn, the only benefit from this regulation is the reduction of fatal cancer to sports anglers and Native American subsistence fisherman.
Looking just at the cost-benefit analysis, the cost of the regulation would seem to far outweigh the benefit. An entirely different picture emerges with knowledge of the health effects of the hazards the guidance addresses. Fortunately, EPA's consideration of the rule was not limited to quantified benefits.
Accounting for such benefits requires a more in-depth dynamic analysis, which cost-benefit analysis cannot capture, leading Parker to refer to the methodologies of CBA's greatest proponents as a "simplistic scorecard."
Not a Neutral Tool
Even given the many uncertainties of cost-benefit analysis, proponents still argue that it acts as a neutral tool. Yet, as David Driesen points out, "if CBA only makes regulation weaker, and never strengthens overly weak regulation, it cannot improve priority setting and consistency in the manner its proponents envision." Driesen lays to rest the argument of CBA's neutrality by dissecting the use of CBA both in practice and theory. Driesen finds that both in OMB's implementation of cost-benefit analysis as well as in the assumptions of the cost-benefit analysis itself, CBA is weighted in favor of the regulated industry and against health, safety and environmental protections.
Driesen focuses his look at cost-benefit analysis on the role of the Office of Information and Regulatory Affairs (OIRA), a subagency of the Office of Management and Budget (OMB) charged with carrying out cost-benefit analysis through Executive Order 12866. According to a Government Accountability Office (GAO) report, between June of 2001 and July of 2002, OMB "significantly affected 25" environmental, health and safety regulations. If cost-benefit analysis is in practice a neutral tool, then OIRA's use of cost-benefit analysis to review regulation would sometimes strengthen protections and sometimes weaken them. Driesen found that none of OIRA's changes made environmental, health or safety protections more stringent, and 24 out of the 25 weakened protections. Even if cost-benefit analysis is theoretically a neutral tool, in the hands of this administration, it is certainly biased against strong public protections.
Further evidence comes from the records of agencies whose statutes demand the use of cost-benefit analysis. Courts have interpreted the Toxic Substance Control Act (TSCA) and the Federal Insecticide Fungicide and Rodenticide Act (FIFRA) to require CBA. The result has been a nearly complete halt on regulatory activity. According to Driesen, "EPA has not banned a single chemical under TSCA since the United States Court of appeals for the Fifth Circuit interpreted the statute as requiring that bans pass a cost-benefit test." FIFRA has suffered much the same fate. Cost-benefit analysis has put the breaks on needed regulatory protections.
For other statutes governing EPA, including the Clean Air Act and the Resources Conservation and Recovery Act, cost-benefit analysis is used in an "indeterminate position," meaning it is considered when forming regulation but does not serve as a specified criterion. For regulations promulgated under these statutes, OMB historically has applied cost-benefit analysis unevenly, providing lengthy analysis when considering a new regulation but quickly approving deregulatory actions. Driesen uses the example of regulating particulate matter. For example, notes Driesen, "OMB engaged in protracted argument with EPA in the early 1980s over whether EPA must prepare a CBA of a possible tightening of the particulate matter National Ambient Air Quality Standard (NAAQS), but it cleared EPA revocation of the hydrocarbon NAAQS in two days with no formal CBA."
Though overall OMB generally approves agency rules without changes, when it comes to reviewing EPA rules, "OMB often significantly changes between 45 and 75 percent of the rules it reviews." This high rate shows that at least in practice OMB is hostile to environmental protections.
Driesen was able to find only one case in which cost-benefit analysis has led to increased regulation: the reduction of lead in gasoline. Yet in this case, the initial implementation of the regulation occurred without performing cost-benefit analysis. CBA was only applied after the regulation had been in place for a number of years in order to support a more stringent standard. Heinzerling, Ackerman and Massey have shown that if the regulation had not already been in place, the empirical evidence would not have existed to justify the further reduction of lead in gasoline.
Cost-benefit analysis can be used in a variety of ways by regulators. Some agencies use cost-benefit analysis in what Driesen calls an "indeterminate position," meaning the agency considers cost-benefit analysis but does not use it as a criterion for determining regulation. OMB tends to see cost-benefit analysis as a criterion under which the cost of implementing a regulation can never exceed the benefit. Another option is that cost-benefit analysis is used as a criterion under which cost must always equal benefit, optimizing the efficiency of the regulation. Driesen shows that in each case cost-benefit is not a neutral tool and will always favor the regulated community over the health, safety and environmental regulation.
The Indeterminate Position: Even if cost-benefit analysis is used in an "indeterminate position," weighed equally with other factors, it will still side against health, safety and the environment because cost-benefit analysis requires a greater expenditure of government resources and a delay in implementation of important safeguards. In lieu of cost-benefit analysis, agencies consider regulation based on the technological feasibility of implementing a regulation (including the cost of compliance) or an assessment of the health effects of a given regulation. "Cost benefit analysis combines all of the difficulties of both of these forms of analysis and creates an additional complication -- it requires quantification of benefits and, whenever possible, the assignment of monetary values to each of those benefits," according to Driesen. This method is inevitably more costly and time-consuming for the agencies and delays enforcement.
Delays in regulation are not neutral. They always benefit the regulated community at the expense of those exposed to the potential hazards by increasing the amount of time individuals are exposed to adverse conditions while giving the regulated community a longer time to avoid compliance.
Benefit Cannot Outweigh Cost Criterion: The criterion that benefit cannot outweigh cost is inherently not neutral. It acts as a one way ratchet, reducing benefit when cost is too great but never demanding an increase in benefit. If cost falls below benefit, this criterion does not require a more stringent standard. But if cost outweighs benefit, agencies are forced to weaken their standard in order to comply. This understanding of cost benefit is the one generally employed by OMB and the administration.
Cost Equals Benefit Criterion: If benefits are optimized, then cost should always equal benefit. CBA using this criterion at first appears to be a neutral tool. It could create either more or less stringent standards based on the conditions. If marginal benefit is greater than the marginal cost, it could recommend a stronger, more costly standard. Optimizing benefit may be more neutral overall, but it is not neutral compared to existing standards. Key provisions in the law require full protection of public health and the environment. In comparison to this standard, optimizing benefit is not a neutral theory because, as Driesen says, "this optimization criterion would not make regulation that already fully protects human health and the environment more stringent, but it would sometimes make it less stringent, so it is certainly not neutral relative to a health-protective standard." By maximizing efficiency, this criterion could also allow for the death of innocent life or allow harms to go unabated.
Even if cost-benefit analysis is applied in a relatively neutral way, the underlying methodology involves value choices that cannot be neutral. A cost-benefit analysis requires choosing a specific methodology to make a comparison of benefits and costs. Various ways of calculating benefits can have drastically different outcomes. Driesen explains:
For example, CBA proponents do not ask how much would a company have to pay a victim to get her to agree to die of cancer contracted after breathing in the fumes from the company's plant. Rather, they asked how much would a potential victim pay the factory to avoid a risk.
Choosing a methodology involves a non-neutral value judgment.
Misguiding Our Priorities
Not only is cost-benefit analysis not a neutral tool; it fundamentally gets it wrong. Cost-benefit analysis does not reflect our country's values or priorities. In Lisa Heinzerling, Frank Ackerman and Rachel Massey's "Applying Cost-Benefit Analysis to Past Decisions," the authors seek to show what would have happened if cost-benefit analysis had been applied to some of our landmark environmental, health and safety regulations. They investigate the reduction of lead in gasoline, a proposed regulation that would have allowed damming in the Grand Canyon, and the regulation of occupational exposure to vinyl chloride, a chemical used in producing PVC. Their conclusion in all three cases is that cost-benefit analysis would have gotten it wrong, depriving us of some of our most important health, safety and environmental protections.
One of their most compelling examples is that of vinyl chloride. Vinyl chloride is a known carcinogen used in making PVC. In 1974, when OSHA sought to regulate vinyl chloride, substantial evidence existed about vinyl chlorides toxicity, especially its link to a rare form of liver cancer, angiosarcoma, but little was known about the safe level of exposure or how many people had or would die from angiosarcoma through exposure to vinyl chloride. At the time, there were only 13 known cases of angiosarcoma deaths from vinyl chloride exposure. Still, OSHA chose to take a precautionary stance and sought to lower the allowable exposure level to 1 ppm over an eight-hour period. Previously, industry had allowed an exposure of 200 ppm "time-weighted average" with a maximum allowable exposure of 500 ppm.
By statute, OSHA does not perform cost-benefit analysis and must enforce the most stringent policy "feasible." If it had performed CBA when determining an exposure limit for vinyl chloride with the knowledge they had at the time, CBA would have come out in favor of a much weaker standard. Heinzerling, Ackerman and Massey compared the estimated cost of compliance at the time with the estimated value of a life in order to determine how many lives OSHA would have needed to think it was saving to justify the stringent regulatory standard.
The estimated cost of compliance with the vinyl chloride regulation was thought to be $200 million per year (though it turned out to be much lower). For the value of a human life, the authors used two different estimates: the highest value of a life based on current EPA calculations and adjusted for inflation, which is $1.81 million, and the much lower value of life used in the Ford Pinto controversy that occurred around the same time, which estimated the value of a statistical life at $200,000.
Only 7,000 people worked in the vinyl chloride industry. Using the Ford Pinto value, one out of every seven workers would have had to die to justify the stringent standard. That means that 1,000 people would have had to die each year to justify OSHA's regulation. If you take into account discount rates, the picture becomes even more dismal. At a 3 percent discount rate, 200 people using the high estimate for life value or 2,000 people using the lower estimate would have had to die each year for OSHA to justify the costs. At a 10 percent discount rate, 700 people would have had to die using the former estimate and 7,000 using the lower. Thus they conclude, "using a 10 percent discount rate and the value of life estimated in the 1970s, it would be necessary to show that every worker in the industry, every year, would have died in the absence of the standard, in order to justify the regulation in cost-benefit terms."
This dramatic example adds to the overwhelming evidence that cost-benefit analysis is not only a weak tool for determining public protections, but its "impartial" calculations can have severe and damaging impacts. In no way is it a blind arbitrator, equally weighing both sides of an issue. Rather it is a political tool, weighted to favor the regulated community that does not adequately address our regulatory priorities.