Any close observer of the regulatory process has learned by now that the government often requires the expenditure of a huge sum of money--sometimes billions of dollars--to save a single human life. She has also learned that there are many regulatory options available that would produce the same benefits at a far lower cost. She has learned these things largely from a table prepared during the 1980s by an economist at the Office of Management and Budget named John Morrall. This table shows the cost per life saved of various risk-reducing federal regulations. According to the table, this cost varies dramatically from regulation to regulation, from a low of $100,000 per life saved to a high of $72 billion. One-third of the regulations on the list reportedly cost over $100 million for every life they save.(1)
These numbers are ubiquitous in the literature on risk regulation. Numerous scholars have relied on Morrall's table, as well as earlier and later versions of it, in arguing that the costs of regulation often exceed its benefits, that many more cost-effective strategies exist for reducing risk, that regulation sometimes increases overall risk, and that regulatory priorities are not set in a rational manner.(2) Morrall's calculations, in short, have been used to support every one of the most prominent current critiques of the regulatory system. They have also played a significant role in political debates over regulatory reform, including most recently the debates surrounding the Republican proposed Contract with America.(3)
Despite the pervasive reliance on Morrall's table in scholarly and political discussion of risk regulation, no one has determined--or even asked--whether the fantastic costs reported by Morrall are open to question. In this Article, I fill this gap. I argue that John Morrall's table is in the nature of a modern urban legend, a vivid, plausible, "false-true tale," circulated broadly, embellished with local detail, and believed implicitly.(4) Like other modern legends, such as stories about rats served as hamburger and alligators living in the sewers, John Morrall's story is strange and believable. It reflects the "hopes, fears, and anxieties" of modern life.(5) And it is false-true.
Indeed, as I explain, there is ample room for disagreement with the picture of the regulatory system drawn by Morrall. For starters, a large percentage of the regulations appearing at the bottom of Morrall's list--the allegedly costliest regulations--have never taken effect. Many of these rules were rejected precisely because the agencies in question determined that their benefits did not justify their costs; at least by the regulatory reformers' lights, then, these regulatory decisions should be counted as successes. By topsy-turvy logic, however, they have become known as regulation's greatest failures.
Equally striking is the disparity between the agencies' (often implicit) estimates of costs per life saved and Morrall's estimates of such costs. Morrall's estimates are inevitably higher than the agencies' implicit estimates, sometimes as much as 1000 times higher. These huge disparities appear to arise mostly from two adjustments Morrall made to the agencies' estimates of regulatory benefits. First, in some cases, Morrall. rejected the agencies' estimates of risk and relied instead upon what he regarded as more reliable--and inevitably lower--estimates.(6) Use of lower estimates of risk led to lower estimates of lives saved, which led to higher costs per life saved. Second, in all cases, Morrall applied a technique that has come to be known as "discounting lives": He reduced the estimates of the number of lives that would actually be saved by a regulation in the future by ten percent for every year expected to pass before the lives were saved.(7) For example, whereas Morrall estimated that the Occupational Safety and Health Administration (OSHA) rule limiting arsenic exposures in the workplace(8) would actually save 11.7 lives per year,(9) this estimate of lives saved dropped to approximately 0.35 lives per year after discounting.(10) By deflating the estimates of lives saved, discounting greatly inflated the costs per life saved of the rules that were designed to protect against the risks of long-latency diseases and long-term ecological harm. Again, these are the rules that appear at the bottom of Morrall's list.
Without these adjustments to the agencies' estimates of regulatory benefits, the costs per life saved of all but two of the regulations on Morrall's list fall to less than $5 million.(11) This cost compares favorably to current estimates of the monetary value of a human life, based on the wage premiums reportedly accepted by workers in risky occupations: Studies commonly cite a range of $3 million to $7 million for the value of a statistical life.(12) Thus, even if one accepted the rather stringent and controversial condition that market exchanges are the appropriate measure of the value of a human life, one would be inclined to conclude that the costs per life saved of the regulations on Morrall's list--as calculated based on the agencies' estimates of regulatory benefits--fall well within the bounds of reasonableness.
Moreover, even these lower costs are inflated insofar as they reflect only one regulatory benefit: the prevention of quantified cases of human sickness and death. In practice, this means that the only regulatory benefit captured by Morrall's table, and by my own figures based on the agencies' calculations, is the prevention of premature human death from cancer. Yet, as we shall see, the regulatory programs that fare poorly in Morrall's table do not just prevent cancer. They prevent many human illnesses that cannot be quantified; they prevent ecological harm; and they prevent harms to values that are widely shared, such as autonomy, community, and equity. Described in this light, most of the regulatory programs at the bottom of Morrall's list are not just-barely cost-justified, they are bargains.
One will want to know whose vision of the costs and benefits of the regulatory system--Morrall's or mine--is correct. As I explain, however, one cannot choose between these very different pictures of the regulatory system without first making important and contested choices about values. The decision whether to discount lives saved in the future, which explains so much of the difference between Morrall's numbers and my own, ultimately turns on one's views of the worth of lives saved today relative to lives saved tomorrow. Even the assessment of physical risk reflects choices about values. The decisions whether, for example, to credit the results of studies showing heightened risks in laboratory animals and whether to fix attention on the highest plausible estimates of risk turn on one's view of the proper response to scientific uncertainty. Finally, the very decision to report the cost-effectiveness of federal regulation as a function of costs per quantified human life saved presupposes a particular, and particularly narrow, vision of the purposes of risk regulation. Value choices of this kind underlie Morrall's numbers, as well as my own. To say that one of these sets of numbers is true, and the other false, is thus misleading. It depends on one's hopes, fears, and anxieties.
After examining the values inherent in Morrall's figures, I discuss the lessons that might be learned from the wide and uncritical embrace that Morrall's table has received in the literature on risk regulation. Certainly, many who have relied on Morrall's figures share the values implicit in those figures and would not be displeased to learn that the figures embody those values. But this is by no means universally true. Indeed, some of the same authors who have relied on Morrall's numbers in concluding that the current regulatory system needs repair have also argued (sometimes in the very same articles) against the basic assumptions underlying Morrall's numbers. Why then has no one before now questioned the basis for the breathtaking costs Morrall cites? I suggest that the widespread acceptance of Morrall's table shows the perils of precise quantification. A table like Morrall's, with its estimates to the third decimal point, has the appearance of objective fact. In reality, however, Morrall's table is like a deep well into which all of the most prominent criticisms of the cur-rent regulatory state have been poured. It is not surprising that those criticisms may, in turn, be drawn back out of the well. But this tells us nothing we did not know before the table was prepared.
THE STANDARD ACCOUNT
The last two decades have witnessed unprecedented attention to the costs and benefits of the regulatory state. Regulations designed to reduce risks to human health have received particularly close scrutiny. Nevertheless, one can find few efforts to discover how well the system for regulating risk generally performs. Certainly, there have been numerous studies of the costs and benefits of specific regulations" as well as many broad-gauged critiques of the inefficient tendencies of particular categories of regulation, such as technology based regulation.(14) Yet very little empirical evidence exists regarding the costs and benefits of the risk-regulating system as a whole.(15)
In 1986, an economist at the Office of Management and Budget (OMB) named John Morrall attempted to fill this gap. Noting that "there is little systematic information describing the kinds of risks the government has chosen to regulate or the effectiveness of these interventions,"(16) Morrall set out to study the cost-effectiveness of federal regulations designed to reduce risks of premature human death. He chose forty-four regulations in all, based on the availability of "reasonably complete information" on costs and benefits.(17) The centerpiece of Morrall's article is a table showing the cost per life saved of these forty-four...