Oren Cass On The Problems With Cost-Benefit Analysis

See The New Central Planners. Cass is with the Manhattan Institute. Excerpts: 

"The approach of the National Transportation Safety Board in a conflict with the Federal Aviation Administration illustrates the devastatingly foolish mindset the order aimed to counteract, and the valuable role a cost-benefit analysis can play. As any parent knows, a child under two years old can fly without purchasing a ticket by sitting on an adult's lap. But in 1994, a DC-9 crashed while attempting to land in Charlotte, North Carolina. Thirty-seven of 52 passengers were killed, including a nine-month-old infant held by her mother. The NTSB concluded that "if the child had been properly restrained in a child restraint system, she might not have sustained fatal injuries," and thus issued a formal safety recommendation requiring that small children have their own seats and appropriate restraint systems.

In response, the FAA used the number of annual aviation accidents with survivors and the share of passengers under the age of two to estimate that the proposed regulation could save no more than five infant lives per decade. With more than 6 million such infants flying each year, and assuming a ticket price of $200 per person, the cost per life saved would exceed $2 billion. Further, as the FAA noted, the additional cost would lead some families to drive in situations where they otherwise would have flown, and deaths resulting from "diversions" to far more dangerous highways would cause many more infant deaths than the new rule could prevent. Amazingly, the interagency battle continued for 12 years before the NTSB, without changing its opinion, "reluctantly concluded that it cannot convince the FAA to take the action recommended" and closed the matter."

"Perhaps the greatest flaw in cost-benefit analysis runs parallel to one that bedevils the identification of externalities: the subjectivity inherent in deciding what factors to include. Take the EPA's recent effort to tighten limits on atmospheric ozone levels. Its cost-benefit analysis identified between $19 billion and $38 billion in annual benefits as compared to $15 billion in annual costs. But the comparison is apples-to-oranges. The EPA cast its benefit net widely, counting whatever public-health benefits it could infer from past epidemiological studies and finding two-thirds of its total not from ozone reductions but rather from the "co-benefits" of incidental reductions in other pollutants. For example, missed workdays for mothers staying home to care for asthmatic children — avoided thanks to cleaner air — are valued at $75 each.

On the other side of the ledger, only the most direct compliance costs facing regulated facilities are considered. Even then, the analysis relies on the appearance of as-yet-uninvented technology (and its hypothetically declining cost) to make compliance plausible and affordable. In 2011, when President Obama (facing re-election) rejected the EPA's first attempt to tighten the ozone standard, he explained, "I have continued to underscore the importance of reducing regulatory burdens and regulatory uncertainty, particularly as our economy continues to recover." But the EPA's final 2014 analysis of cost excludes any consideration of how the policy might affect direct economic indicia like energy prices, employment levels, and economic growth, to say nothing of the broader socioeconomic impact of restricting industrial activity across broad swaths of the country. No consideration is given to the damage increased regulation and uncertainty does to the economy's dynamism, or the opportunity cost of firms never started and ideas never pursued.

Such things may seem unquantifiable, but that challenge is no different from the one faced in the translation from air-pollution levels to premature death to a dollar value. When the president of the United States says his policies are saving "tens of thousands of lives each year" and producing "hundreds of billions of dollars in benefits for the American people," one assumes the claim must have a solid basis. But few who see that particular sausage being made would be inclined to ever consume it again.

The claim of lives saved, for instance, does not represent actual lives actually saved. In fact, EPA offers no evidence of the relevant pollutants at the relevant levels ever causing a single death. Instead, it relies on epidemiological studies showing that deaths tend to increase slightly (on the order of 1% to 2% in the case of ozone) on days when atmospheric pollution concentrations are significantly higher. Thus the suggestion is that, by reducing those concentrations, each person faces some minutely smaller chance of dying. When EPA reports its rule will avoid 630 premature deaths from short-term exposure, it means only that each American's risk of death will be reduced by 0.0002%.

To determine the value of that risk reduction, EPA must then use the "Value of a Statistical Life," calculated in large part from wage-risk studies that examine the wage premium given workers in high-risk occupations. For instance, as one Harvard University study sponsored by the EPA found, male blue-collar workers in higher-risk industries earned an additional $0.32 per hour. This was nearly three times the equivalent premium for female blue-collar workers. For white-collar workers, there appeared to be no wage premium at all. Statistical analysis nevertheless translated these findings into a "value" of approximately $10 million per life, roughly the current figure used by the EPA, thus the billions of dollars in savings.

Table 5-20 on page 305 of the 575-page Regulatory Impact Analysis for the EPA's proposed ozone standard shows a "monetized ozone-only benefit" of $6.4 billion for the prevention of premature deaths. When combined with its estimated benefits from the coincidental reduction of other pollutants, the total benefits of $19 billion to $38 billion exceed the $15 billion in estimated compliance costs. But there is no $6.4 billion. There is only a statistical relationship between ozone levels and mortality showing a lower level could lower risk of death for some people on particular days by a thousandth of a percent, and a finding that certain blue-collar workers in certain higher-risk industries receive slightly higher wages.

Perhaps that is the best possible translation of environmental harm into economic cost. The analysis does have some use, insofar as it provides a basis for comparing the relative harm of various pollutants or the relative cost-effectiveness of different mitigation strategies. But it does not follow that society should spend up to 6.4 billion actual dollars to achieve a 6.4 billion "dollar" benefit, or 15 billion actual dollars for the full 19 billion to 38 billion "dollars" of benefit in the EPA's analysis. Nor does it follow that an economist can confidently use these estimates to design a tax that will improve the market's efficiency."

Thanks to the past 40 years of environmental regulation, an extensive literature now documents its economic effects. For instance, a 2000 study in the Journal of Political Economy found counties in "nonattainment" with Clean Air Act standards (as many more would be under the EPA's new ozone rule) saw the construction of new plants in polluting industries decline by 26% to 45%. A 2001 NBER working paper found that, between 1972 and 1987, such counties lost 590,000 jobs and $75 billion in economic output. More broadly, a 2013 study in the Quarterly Journal of Economics found that, after the passage of the 1990 Clean Air Act amendments, "the average worker in a regulated sector experienced a total earnings loss equivalent to 20% of their preregulatory earnings." In Consequences of Long-Term Unemployment, the Urban Institute provides a helpful literature review of the many findings on how such economic outcomes lead to "declining human and social capital," "impacts on future labor market attachment," "impacts on physical and mental health," "effects on children and families," and "impacts on communities."

Beyond the concrete effects within affected industries, policymakers committed to giving benefits and costs equal attention could also examine the macroeconomic implications of increased regulation. The Mercatus Center, for instance, has recently launched RegData, a comprehensive database quantifying industry-specific federal regulations. The regulatory equivalent of studies measuring concentrations of pollution in the atmosphere, RegData allows analysts to identify relationships between regulation and outcomes like economic growth and productivity. One early study found that from 1997 to 2010, the least-regulated industries experienced productivity growth at twice the rate of the most-regulated industries."

"The Department of Energy, approvingly analyzing its own proposed energy-efficiency requirements for commercial refrigeration equipment, assumed that increased prices would produce no decline in purchases and further concluded the new requirements would increase both employment and wages."