Summary
holding that Dr. Smith's hedonic damages proof fails to survive Daubert analysis
Summary of this case from Stokes v. John Deere Seeding Grp.Opinion
No. 92 C 7815.
May 23, 1995.
Elliot R. Zinger Associates, Chicago, IL, for plaintiff.
Thomas Osran, Martha Barglow, Donald Zoufal, Corp. Counsel Office, Chicago, IL, for defendants.
MEMORANDUM OPINION AND ORDER
At about 9 p.m. September 14, 1992 City of Chicago ("City") Police Officer Hugh Robinson ("Robinson") shot and killed Lenardo Ayers ("Lenardo") in the vicinity of 68th and Loomis Streets in Chicago. Lenardo's mother Lillian Ayers ("Lillian"), both individually and as special administrator of his estate, brings this action against Robinson, Chicago Police Chief Matt Rodriguez and City itself to recover money damages under 42 U.S.C. § 1983 and various Illinois state laws.
At trial Lillian would hope to introduce into evidence expert testimony by Professor Stan Smith ("Smith") on the issue of hedonic damages. In Smith's book coauthored with Michael Brookshire (Economic/Hedonic Damages: The Practice Book for Plaintiff and Defense Attorneys (1990 with 1992/93 cum. supp.) ("Hedonic Damages ")) the hedonic value of life is defined as "the value of the pleasure, the satisfaction, or the `utility' that human beings derive from life, separate and apart from the labor or earnings value of life" (id. 164). Hedonic Damages describes the concept as encompassing everything from the fleeting joy derived "from staring adoringly into an infant's smiling face" to the more deep-seated satisfactions involved in "endur[ing] life's hardships" (id. 164). Smith claims to be able to put a dollar figure on those intangibles and, if allowed to do so, would testify that Lenardo's hedonic damages are approximately $2 million.
Defendants have moved to exclude that testimony under the principles set out in Daubert v. Merrell Dow Pharmaceuticals, Inc., ___ U.S. ___, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and their motion is now fully briefed and ready for decision. For the reasons stated in this memorandum opinion and order, defendants' motion is granted.
Anticipated Testimony
Lillian has not tendered any specific proffer as to the nature of Smith's anticipated testimony. But the principles and methodology upon which Smith relies — the so called "willingness-to-pay" approach to valuing human life (P.Mem. 7-15) — are reported in a number of articles in addition to Hedonic Damages 161-75: see, e.g., Lauraine Chestnut Daniel Violette, The Relevance of Willingness-To-Pay Estimates of the Value of a Statistical Life in Determining Wrongful Death Awards, J. Forensic Econ. 3(3) 75 (1990); Andrew McClurg, It's a Wonderful Life: The Case for Hedonic Damages in Wrongful Death Cases, 66 Notre Dame L.Rev. 57 (1990); Ted Miller, Willingness to Pay Comes of Age: Will the System Survive?, 83 Nw.U.L.Rev. 876 (1989) and a number of articles authored or co-authored by Smith (Smith, Should Judges Allow Testimony About Hedonic Damages?, Chi.Daily L.Bull., July 15, 1991, at 2; Gary Magnarini Smith, Hedonic Damages, Wis.Law., Feb. 1991, at 17-19 and 56-58; and Smith, Hedonic Damages in Wrongful Death Cases, A.B.A.J., Sept. 1, 1988, at 70-73). See also the reported cases in which Smith has previously sought to testify (e.g., Mercado v. Ahmed, 974 F.2d 863, 868-71 (7th Cir. 1992) and Estate of Sinthasomphone v. City of Milwaukee, 878 F. Supp. 147 (E.D.Wis. 1995)). All of those materials supply ample basis for evaluation, for Daubert, ___ U.S. at ___, 113 S.Ct. at 2797 has instructed:
The focus [of the admissibility determination], of course, must be solely on principles and methodology, not on the conclusions that they generate.
For current purposes the most useful explication of willingness-to-pay methodology is set forth in the sample testimony in Chapter 11 (pages 225-29) of Hedonic Damages and in the related section at pages 81-85 of its 1992/93 supplement. Those samples provide both a concise summary of the conceptual underpinnings of the economic model and an overview of the most recent empirical data. Hence that testimony will be used as a starting point, drawing upon the larger body of literature and Smith's own scholarship when necessary. Although Smith's coauthorship of Hedonic Damages leads this opinion to refer to that title throughout (rather than speaking of Smith alone), it is reasonable to treat the book's sample testimony as reflective of what Smith would offer if he were allowed to take the stand.
In the book's initial text the sample testimony deals with the damages to be awarded a 35-year-old white male named Jack Doe in a wrongful death action. Here is the testimony on hedonic damages, which follows testimony as to the economist's credentials and a calculation of economic loss ($605,795):
DIRECT EXAMINATION
Q: [Plaintiff's attorney] Do the losses that you have estimated thus far, wages, fringe benefits, and household services, all net of personal consumption, account for all the losses sustained by the decedent?
A: [Economist] No they do not.
Q: What additional losses are there?
A: Mr. Doe also suffered the loss of the pleasure of life, itself: the value that he would have expected to obtain from living beyond the value he may have attached to his net financial loss. This is a value which is recoverable in this jurisdiction according to my understanding.
Q: Is there a name that economists sometimes use to refer to this value?
A: Sometimes this loss is referred to as a loss of hedonic value.
Q: What does the word hedonic mean?
A: It comes from a Greek root word meaning value or satisfaction or pleasure. Economists use the word to refer to the non-earnings-based value of life, the value we get from living as opposed to working. People get value or satisfaction or pleasure from living, even though not all moments are pleasurable. But by and large, unless we are suicidal, we regard life as satisfying.
Q: Is this non-earnings-based value, or hedonic value, a significant figure?
A: Yes it is. In this instance it is $1,709,842 when applied to Mr. Doe. Studies have shown that where estimates of lost earnings are available as well as estimates for the non-earnings value of life, the hedonic value ranges up to several times earnings.
Q: Are you saying that economists have made measurements of the value of life beyond lost earning capacity?
A: Yes they have. There is an extensive body of literature published in scholarly journals estimating this value in several different ways.
Q: What are those ways?
A: First, economists have examined the value of life expressed by consumers in their expenditures on safety. If you buy an air bag for your car, you are spending money to reduce the probability of injury and death. You are placing an implicit value on your life in so doing. Secondly, economists have studied what certain risky occupations receive as extra compensation for the risk they present. When someone receives an extra $1.00 an hour as a security guard in a high-risk neighborhood, this represents a premium for the extra hazard to life. Finally, government agencies analyze the impact of lifesaving regulations and the costs associated with such regulations. All in all, there are dozens of estimates in the literature regarding the value of life published over the last several decades.
Q: What do those estimates show?
A: The estimates show, typically, that we value life in the several million dollar range.
Q: Can you describe to us an example of how these studies are conducted?
A: Yes. Assume that a person purchases a safety device for $700 and that device reduces the probability of his death from 7 in 10,000 to 5 in 10,000. By reducing his chance of dying by 2/10,000ths, or one chance in 5,000 at a cost of $700, economists would say that he valued his life at $3,500,000. In effect, if 5,000 people spent $700 each on air bags, one life would be saved at a total cost of $3,500,000. These figures are not the actual figures from air bag studies, but just sample figures to show how the analysis is done, in simple terms.
Q: Is this estimate for the total value of life?
A: Yes, these estimates include several components of value that must be subtracted out in order to arrive at the net satisfaction value. Specifically, we must subtract the net lost earnings value, the loss of household services, and the value of financial security for the statistically unknown, or anonymous, person being considered in these economic studies of the overall value of a human life.
Q: Could you explain this in more detail?
A: Yes. The value of life estimates are, in many instances, for the lives of anonymous persons. From that total value, we must account for what the unknown, statistically average person earns, contributes in household services, and attributes as a value of financial security. These figures add up to an estimated $800,000. Netting this amount from a reasonably conservative estimate of the value of life leaves approximately $2,700,000 which I treat as an undiscounted value for the anonymous, statistically average person.
Q: What adjustments do you then make to the undiscounted $2,700,000 figure?
A: I estimate the value of life per year of life expectancy by dividing by the remaining life expectancy of an average person, which is approximately 45 years according to the life tables. This leaves a $60,000 per-year value in 1989 dollars. Although most economists estimate that the remaining life expectancy of the lives of the people covered in the studies ranges from about 35 to 40 years, I used a more conservative figure of 45 years, which lowers the per-year estimate, because it is the value for an average person in the U.S. census.
Q: What other adjustments do you make?
A: I then take into account the age, race, and sex of the decedent by calculating the losses through the life expectancy of Mr. Doe, using the same growth and discount factors we discussed earlier.
Q: So you figured out the hedonic value of life based on the value that is attributed to statistically average people and then adjusted it to the specific characteristics of the decedent, namely, age, race and sex?
A: Yes sir.
Q: What further adjustments did you make?
A: Economists do not know as yet how to adjust further, to take into account that the decedent was married, for example, or that he had two children. These factors and all the other factors regarding the quality of life of the decedent can only be taken into account by the jury, in my opinion.
Q: So your figures can be adjusted upward or downward depending on other factors that a trier of fact may wish to take into account?
A: Yes sir.
Q: No further questions, your honor.
CROSS-EXAMINATION
Q: [Defense attorney] Turning now to the intangible losses that you tried to estimate, what is the word you used for them?
A: [Economist] I called it a hedonic loss, the loss of the satisfaction of living.
Q: Are you saying that because Jack Doe lost out on the hedonistic opportunities he had, he should also be compensated, without taking into account any of the misfortunes he might have sustained in his life such as sickness of a loved one?
A: The term is hedonic, not hedonistic. The two are different. I have estimated the value that he placed on the satisfaction of living, which includes his fair share of life's misfortunes.
Q: What is the purpose of the studies that economists have made that you refer to? Were they done to value life in court?
A: No, not particularly, but they can be used to do so. Government studies regarding average wages in various professions are not done to value lost earnings in court, but they are routinely used to do so.
Q: Are you saying that because I will pay $700 for an air bag to reduce the probability of dying by 1 chance in 5,000 that I will sell my life for $3,500,000?
A: No, if I asked you how much you valued your life, the answer would probably be an infinite amount. But by looking at the fact that if 5 million people were to buy air bags, they would be spending $3,500,000,000 and in that group, approximately 1,000 lives would be saved. Whose life is not known with any certainty at all, but collectively as a group, 5 lives were saved at a cost of $3,500,000 each. This is the basis on which economists say that members of the group place a $3,500,000 value on life.
Q: Don't you save more than a life when you buy an air bag? Don't you prevent injury, and don't you perhaps save property also in many of these devices such as smoke detectors?
A: Yes sir, that is true, and to the best of the ability of the analysis, the value of injury and property saving is taken into account.
Q: Do you expect us to believe that the average person, in buying safety devices, knows the risks that he is averting?
A: Studies have shown that for common risks, average individual perceptions are quite realistic; for uncommon risks, however, such as for nuclear plant accidents, risks are misperceived.
Q: Is your benchmark figure the result of averaging?
A: No, it is not a formal weighted average of the results. But I believe it is a reasonable estimate of the value that we may, on average, place on life.
Q: So it is speculative whether Mr. Doe valued his life at this figure, isn't it?
A: No, it is not speculative. Speculative damages usually refers to the cause of the damages. Here we know that there was a loss from the death, we are not speculating. And in estimating the loss, we are not speculating any more than if we were estimating the earnings losses of a deceased 18 year old, who had just graduated from high school with no earnings history. In that instance as in this, we would be going to tables that show average earnings for males of certain ages.
Q: Dr. Economist, do you know whether Jack Doe enjoyed life as much as an average person does?
A: I don't know whether he did or not. I have estimated the value that I believe he placed on his life based on the information that I can take into account. I can't help the jury to determine whether, due to other factors, Mr. Doe valued his life less or more than my estimate.
Q: What if I told you Mr. Doe never bought an air bag?
A: That doesn't necessarily mean that he had a casual disregard for the value of life; he may have been very careful crossing the street and shown in other ways how he valued his life. Not all of his behavior must be consistent; I would agree though, that if he frequently took risks for no particular reason, that might be relevant.
Q: Why isn't the hedonic value of life in the government accounting system like wages are?
A: The government does not account for household services in Gross National Product, either, as we discussed before. That exclusion does not mean that the value of such services, or that the value of living, is zero.
Q: What is the multiple between the lowest and the highest figure in the studies you examined?
A: The multiple is something like 5 for the more accepted studies.
Q: Five times from low to high! I find that a rather broad range, Professor Economist. No further questions, your honor.
REDIRECT EXAMINATION
Q: [Plaintiff's attorney] Is it speculative to state that we place a value on our lives?
A: [Economist] No. We, governments, industry, and individuals are constantly in the process of weighing risks and expenditures, implicitly valuing life in the process.
Q: When a policeman negotiates for extra pay due to the extra risk of death on his job, does he ask for, on average, as much as you or I?
A: No, studies of risk premiums paid to certain occupations probably underestimate the value of life, since those jobs attract people who show more than average willingness to place their lives at risk. This concept was first put forth by Adam Smith, who was perhaps the first economist, in the late 1700s.
Q: How do you explain the broad range of results?
A: Well, if you do studies of how much money high school graduates earn, you will find that some of them wind up barely making a living, and some of them, as entrepreneurs, wind up making millions of dollars. I don't find it unusual that in the studies of how various groups value life there is a significant variation. We don't all value life the same, I expect.
Q: Are you saying that the hedonic value of Mr. Doe was $1,709,842?
A: No. Only God knows the true value of Mr. Doe. I have made an estimate based on what economists can take into account. The trier of fact can adjust my estimate based on additional factors.
Q: So the fact that Mr. Doe enjoyed a caring wife and two lovely children is not specifically taken into account in your analysis, is it?
A: No, it is not.
Q: No further questions. Thank you for your testimony.
Next the testimony in the Hedonic Damages supplement deals with a hypothetical permanent injury case. What is reproduced here is the portion relating to the hedonic damage calculation generally. It incorporates into the analysis a recent survey (Ted Miller, The Plausible Range for the Value of Life — Red Herrings Among the Mackerel ("Plausible Range "), J. Forensic Econ. 3(3) 17 (1990)) that is intended to cure the theory of one of its major headaches: the wide range of values represented in the empirical data (see, e.g., William Landes Richard Posner, The Economic Structure of Tort Law 187-89 (1987)) ("statistical estimates of the compensating differentials that people demand to assume a risk of death or serious injury vary so greatly that it is not yet clear that the method is yet usable"):
DIRECT EXAMINATION
Q: Now, Dr. Economist, have you estimated damages other than those resulting from the function of Mr. Terry as what I would call an economic machine, simply producing money to live on?
A: Yes, sir. I have estimated an entirely different category of damages.
Q: This is both in the workplace and the home?
A: This is his loss of enjoyment of life separate from loss of earning capacity.
Q: In general how is this done?
A: In general, let me talk you through three sources of those data. By the 1980s, particularly, we had a wealth of wage differential studies that showed us what workers were willing to give up in wages to work in safer jobs: jobs that would reduce their probability of death on the job. Assume we use the example of top floor window washers on the 50th floor versus first floor window washers. Assume also that we found out from research that each first floor washer was willing to accept $300 less in annual wages on average, and that first floor window washers had a one in ten thousand less probability of death on the job, because if they fall off they don't fall anywhere. Then, we know what ten thousand of those first floor workers will give up in wages. They are willing to give up $300 each, or $3 million in total, so that one of them will not lose his life. They value life, those ten thousand workers collectively, at $3 million, or $300 each times 10,000 workers.
Well, I have studied those wage differential studies to come up with a total value of life, including all the components, of about $2.5 million. There are also a number of consumer studies in which we find out what consumers are willing to spend for safety devices, such as a car air bag system or smoke alarm systems in homes. From those studies, assume we know, for example, that consumers are willing to spend $500 for an air bag system in a car. If, because of air bags, one in ten thousand of those people will not die in a wreck who otherwise would have died, then ten thousand people spending $500 each have valued the life of an unknown American at $5 million. They don't know which one is going to be saved, but collectively they valued it at $5 million. We have a range of those studies.
Finally, in the 1980s we found out what Federal Government agencies were using as standards for the value of life when they set safety regulations, and those values average above $2.5 million. So, I've got three categories of sources that lead me to the conclusion that a conservative estimate of the total life value of a statistically average American — based on what Americans actually do, the wage decisions they make, the consumer decisions they make, and what government agencies use as value of life standards — that a conservative standard of total life value is just over $2.5 million in today's dollars.
From that I have to subtract the economic machine part of that $2.5 million; not for Thomas Terry, but for an unknown average American, and I weight that by the percentage of working Americans who are women versus men. I calculate and deduct from the $2.5 million "whole life" value, the economic machine components for average wage and fringe benefit earnings and the average value of household services. When I do all that and apply it to the life expectancy of Mr. Terry, we get a lost enjoyment of life value based on what Americans do. We don't ask them, "What would you do to avoid being killed?" We look at what they actually do, what they're actually willing to give up to preserve an unknown American life. They don't know if it's theirs, but we know the risk and probabilities of death — in those riskier jobs or without the safety devices.
Based on all that — I know that's an overview — I calculate loss of enjoyment of life at a little over $50,000 a year based upon the willingness-to-pay literature in economics. That's the loss as if it were all lost, as if Mr. Terry died. That's what Dr. Psychologist gives me. The percentages that, in his opinion, are the percentages of that $50,000 that this man has lost because of what happened to him. Otherwise the numbers would be much higher.
Q: This value of human life concept, this is a new concept that you've just developed?
A: It's as old as the first economics book. Adam Smith in 1776 wrote a book called Inquiry Into the Nature and Causes of Wealth of Nations, which is the first "modern" economics book. He talked about what are called compensating wage differentials, and when I teach labor economics classes, we talk about the same thing. They are now sometimes called hedonic wage differentials, but what Adam Smith says is that workers must be paid more to induce them to work in riskier jobs. He didn't have skyscrapers at the time, but if you're going to make them wash windows on the 50th floor, you must pay them a higher wage rate and that differential may exist forever because there's a whole bunch of people who aren't going to work on the 50th floor. Again, Adam Smith may have been thinking of coal miners, but it was the same principle. So, the concept is as old as the first economics book. It was not until the 1980s, however, that we had enough studies, modern studies, of wage and consumer behavior to enable people like me to put that in terms we can use in courtrooms.
Q: Have any articles or publications or pamphlets been written specifically dealing with this concept?
A: Yes.
Q: And can you give us a few of those?
A: One book that deals with that specifically is called Economic/Hedonic Damages, The Practice Book for Plaintiff and Defense Attorneys. That's the book I suppose you're holding. In the Journal of Forensic Economics, there is an article entitled, "Hedonic Damages and Personal Injury: A Conceptual Approach." It's the approach we're following in this case. There have been a number of other papers and articles given at professional meetings on these very concepts and it's what many forensic economists are spending research time on now, as a matter of fact.
Q: So, is your conclusion as to the overall value of life of this unknown American we talked about, this $2.5 million?
A: That's the total life value I used.
Q: Can you discuss in a little more detail the types of studies about differing wage rates versus different risks of life, how they generate values of life from the actual decisions of workers in the work place?
A: Yes, sir, and I can really repeat what I just said, quickly. We have studies, particularly in the 1980s, and most were actually performed for purposes other than courtroom testimony, that match wage differentials for jobs of varying risks with the statistics that, by the 1980s, we had on deaths on the job, by industry and by occupation. Those studies and databases, by the 1980s, allowed us to see that, for example, those first floor window washers would earn $300 less a year. I'm using that number as an example, but this is how the studies were performed. Risk statistics would show those top floor people had a one in ten thousand greater probability of death. So if ten thousand first floor people will spend $300 each — by spend I mean give up in wages, to be safer — if ten thousand of them spend $300, they're valuing an American life at $3 million.
Now I'm using $2.5 million. But I'm looking at those studies from the 1980s that actually average above $5 million.
Q: Are there other consumer studies that relate to this particular problem we're presenting to the jury?
A: Consumer-based studies would first examine the cost of a safety device. Assume it was $200 for a smoke detector system in a home, and we, again by the 1980s, had the probabilities on accidental deaths with or without smoke alarms. Well, if ten thousand consumers spent $200 each for the system, they're spending $2 million in total to save one in ten thousand persons. They don't know which one of them would have died, but they're collectively valuing life at $2 million. So, these studies were the second foundation for my opinion.
Q: I think you mentioned earlier some federal studies in this area. What studies of the federal regulatory agencies or bodies, if any, did you rely upon in reaching your conclusion about the $2.5 million figure?
A: I'm going to read you the most current information on what federal regulatory agencies use as their value of life standards when they set safety standards. The value of life standard used by the Nuclear Regulatory Commission is the highest, and that is $5 million — remember I used $2.5 million. The Occupational Safety and Health Administration standard is exactly $2 million. The Consumer Product Safety Administration for unknown adults uses a standard of $1.7 million; for an unknown child, $1.95 million. The Federal Highway Administration standard is one-and-a-half million. It wasn't until the late 1980s that these values came out in published articles. If one averages all of those regulatory standards, the average is 2.7 million. Let me try to explain what that means. If the Occupational Safety and Health Administration standard is $2 million, which it is, that means that if they could order a company to install a safety device or make its 10,000 workers use a safety device (a guard on a machine or a respirator system or maybe specially designed goggles), they would do so if that piece of safety equipment costs $200 a worker and reduces the risk of death by one in ten thousand. If that device had cost $201 for each of 10,000 workers, OSHA wouldn't have made them do it. They would not force a $2.1 million expenditure; that exceeds their $2 million standard.
Q: In any field of endeavor people criticize the work of other people and I'm sure in your profession as an expert you have been criticized; you have maybe criticized other people. What criticism of this loss of enjoyment of life approach have you considered in evaluating your approach to what this value is?
A: To my knowledge I have read all of the articles and all of the books published to date that have anything to do with this area and again, by the way, the original applications for these studies were in regulatory economics, not the courtroom connection. Many criticisms involve technical details; for example, in looking at different risks of death, how good are the data? What are the best sources? The best single review of these studies was done by economist Dr. Ted Miller, who works for the Urban Institute in Washington. He attempted to adjust the studies that had been published — these wage studies and consumer studies — based on some of the criticisms. His conclusion from all of the studies was an average whole life value of $2.2 million in 1988 dollars, which is just over $2.5 million in 1991 after adjustment for inflation. Indeed, a major criticism of some other economists working in this area comes from me. Whole life values are not hedonic loss values. The "economic machine" component of life value must be subtracted.
Although Lillian's counsel has not supplied this Court with Smith's hedonic damages calculation in this case, given the above methodology it would appear that the figure would fall in the $2 million range. But the precise number is less important to the analysis here than what the input provided by counsel reveals as to the legitimacy of the approach in Daubert terms.
Daubert Principles
Daubert has injected the trial judge into the expert testimony arena in a more active sense than before by abandoning the strict "general acceptance" standard of Frye v. United States, 293 F. 1013, 1014 (D.C. Cir. 1923) in favor of a more flexible approach based on the Federal Rules of Evidence ("Rules"), especially Rules 702 and 403. Though the Rules interrelate under the Daubert analysis, each will first be looked at separately.
Rule 702 provides:
If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise.Daubert, ___ U.S. at ___ -___, 113 S.Ct. at 2795-96 describes the "scientific knowledge" requirement as designed to compensate for the wide latitude that experts might otherwise have in rendering opinions, by providing some assurance that those opinions have a reliable basis and are grounded in the methods and procedures of science, as opposed to subjective belief or unsupported speculation. As for the helpfulness requirement ("assist the trier of fact"), it calls for what is in essence a specialized relevance inquiry that requires scientific testimony to "fit" the case at hand (id. at ___, 113 S.Ct. at 2795, borrowing from Judge Edward Becker's opinion in United States v. Downing, 753 F.2d 1224, 1242 (3d Cir. 1985)). In making the twofold determination of scientific knowledge and fit, the district court should consider such matters as (1) whether the theory or technique can be and has been tested, (2) whether it has been subjected to peer review and publication, (3) its known or potential rates of error and (4) its general acceptance — a non-exhaustive list of "general observations" meant to guide, not bind, the analysis (___ U.S. at ___ — ___, 113 S.Ct. at 2796-98).
Since Daubert our own Court of Appeals has approved trial courts' use of its analysis (or the equivalent) to exclude (1) a "curb side" medical opinion and other unproved hypotheses (Porter v. Whitehall Labs., Inc., 9 F.3d 607, 614 (7th Cir. 1993)), (2) a diagnosis based on observation where experts in the field would have conducted a medical workup, a review of the existing literature and a more extensive examination of the patient's occupational and medical histories (O'Conner v. Commonwealth Edison Co., 13 F.3d 1090, 1106-07 (7th Cir. 1994)) and (3) the opinions of two physicians whose tendered proof was largely anecdotal (Bradley v. Brown, 42 F.3d 434, 438 (7th Cir. 1994)). More generally, Wilson v. City of Chicago, 6 F.3d 1233, 1238-39 (7th Cir. 1993) (citation to Daubert omitted) has exhorted district courts to apply Daubert in these unwavering terms:
This Court, sitting by designation, was a member of the O'Conner panel.
The elimination of formal barriers to expert testimony has merely shifted to the trial judge the responsibility for keeping "junk science" out of the courtroom. It is a responsibility to be taken seriously. If the judge is not persuaded that a so-called expert has genuine knowledge that can be genuinely helpful to the jury, he should not let him testify.
And most recently, see Gruca v. Alpha Therapeutic Corp., 51 F.3d 638, 642-643 (7th Cir. 1995), criticizing a district court for abdicating its "gatekeeping" responsibilities.
Rule 403 is the second major hurdle that the proffered testimony must surmount (Daubert, ___ U.S. at ___, 113 S.Ct. at 2798; Buscaglia v. United States, 25 F.3d 530, 533 (7th Cir. 1994); United States v. Brown, 7 F.3d 648, 654 (7th Cir. 1993)). That Rule reads:
Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.Daubert, ___ U.S. at ___, 113 S.Ct. at 2798 underscored the importance of a thorough Rule 403 balancing to the admissibility of expert testimony by quoting the following passage from Jack Weinstein, Rule 702 of the Federal Rules of Evidence Is Sound; It Should Not Be Amended, 138 F.R.D. 631, 632 (1991):
Expert evidence can be both powerful and quite misleading because of the difficulty in evaluating it. Because of this risk, the judge in weighing possible prejudice against probative force under Rule 403 of the present rules exercises more control over experts than over lay witnesses.
No post -Daubert Seventh Circuit decision addresses the admissibility of expert hedonic damages testimony. In the past our Court of Appeals has sent mixed signals on the subject generally and on Smith in particular (contrast Sherrod v. Berry, 827 F.2d 195, 206 (7th Cir. 1987) ("[t]he testimony of expert economist Stanley Smith was invaluable to the jury in enabling it to perform its function of determining the most accurate and probable estimate of the damages recoverable for the hedonic value of [plaintiff's] life"), vacated and remanded on other grounds, 856 F.2d 802 (7th Cir. 1988) (en banc) with Mercado, 974 F.2d at 868-71 (7th Cir. 1992) (expressing "serious doubts" about Smith's underlying methodology and noting the lack of consensus among experts in the field)). This opinion of course takes those decisions thoroughly into account to the extent that they are consistent with Daubert.
As a technical matter, a vacated opinion becomes no opinion at all. In this instance, however, the status of the quoted language is less clear, because the penultimate paragraph in the majority opinion on en banc rehearing reads (id. at 807-08):
Because we reverse and remand for a new trial, we need not discuss the district court's other evidentiary rulings or jury instructions. We leave these questions for the district court on remand to decide in light of this court's prior discussions of those matters, specifically those found in our earlier vacated opinion.
Daubert Analysis
Because Smith's methodology is itself divided into several parts, it is most convenient to scrutinize it in the same terms for Daubert purposes. This opinion therefore addresses seriatim each of what, although he does not label his approach in these terms, are essentially the five principal parts of Smith's calculation — (1) benchmarks, (2) adjustments, (3) pedigree, (4) empirical data and (5) underlying assumptions — in accordance with the Daubert factors.1. Benchmarks Hedonic Damages begins its text calculation with a benchmark figure that purportedly represents the value of a statistical human life as determined through willingness-to-pay methodology. In the Jack Doe hypothetical that figure is $3.5 million, which is said to be based on a group of 21 studies conducted between 1974 and 1988 (Ann Fischer, Lauraine Chestnut Daniel Violette, The Value of Reducing Risks of Death: A Note on New Evidence ("Fischer et al."), 8 J. Pol'y Analysis Mgmt. 88, 90 (1989), reprinted in part in Hedonic Damages 169). Graph A following this opinion plots the results of those studies, illustrating data that ranges at a relatively even spread (mean = 3.1, median = 2.8, standard deviation = 2.7) from a low of $450 thousand to a high of $8.5 million. After criticizing the methodology behind a number of studies with very low estimates (smokers = $520 thousand, police officers = $870 thousand), Fischer et al. 96 concludes that "[t]he most defensible empirical results indicate a range for the value-per-statistical-life estimates of $1.6 million to $8.5 million (in 1986 dollars)." In turn Hedonic Damages derives its benchmark by characterizing the "preponderance of the results" as "rang[ing] from the high several hundred thousands well into the several millions" (at 168) and by labeling $3.5 million as that range's "central tendency" (id. 170). Hedonic Damages' supplement begins with a lower benchmark of $2.5 million, finding its main source in Plausible Range. Smith describes that article as the best single review of the empirical data, and Lillian's counsel have attached a copy to their memorandum on the current motion. Plausible Range takes 67 willingness-to-pay studies conducted between 1969 and 1990, makes a number of so-called "adjustments" and concludes that the value of a statistical life is between $1.5 and $3.5 million — with $2.2 million representing a statistically sound mean and median (Smith's $2.5 million simply adjusts that figure for inflation through 1991). Graphs B and C respectively depict the sets before and after "adjustments."
Plausible Range's treatment of a subset of 37 wage-risk studies (depicted in Graph D) illustrates its "adjustment" methodology. Six major adjustments are made to the raw data set. First, 12 of the 16 results over $3.5 million are divided by 3, which squeezes them together and brings all 12 within the $1-to-$3 million range (Graph E). Plausible Range does that because those studies use data compiled by the Bureau of Labor Statistics ("BLS"), which are said to be "biased" by a factor of 3. Second, Plausible Range removes a portion of the underlying data from 2 of the remaining 4 studies over $3.5 million and recalculates the results, which squeezes those 2 points together and moves them within the target range (Graph F). Third, Plausible Range multiplies 3 of the very low results by 2.2 (Graph G), assertedly because those studies used data compiled by the Society of Actuaries ("SOA"). Fourth, Plausible Range trashes 7 studies altogether, all but 1 of those discarded figures falling either below $1.5 million or above $3.5 million (Graph H). Fifth, Plausible Range leaves unscathed 13 studies clustered around $1, $2 and $3 million (Graph I). That five-step process leaves 30 data points, most of which have been altered by a factor of 2 or 3. Finally, Plausible Range adjusts for what it calls "risk underestimation." One of the persistent criticisms of the willingness-to-pay model is that people are poor perceivers of risk. That is, when faced with a safety threat most persons will either under- or over-estimate the actual risk. That critique goes to the heart of the theory and threatens to undermine almost all of the empirical research (McClurg, 66 Notre Dame L.Rev. at 103-04; Joanne Linnerooth, The Value of Human Life: A Review of the Models, 17 Econ.Inquiry 52, 53 (1979)). Plausible Range estimates that in the employment context people tend to underestimate risk by about 20% and, to compensate for that, multiplies the remaining 30 values by 1.234 (Graph J). That calculation appears to be based on a single 1978 survey of 74 University of Oregon students. This Court confesses to difficulty in following the even more convoluted adjustments made to the balance of Miller's 67 studies. But the effect is to bring their extraordinarily disparate results within the $1.5-to-$3.5 million range.
It is worth noting that Hedonic Damages 211 (the original text) considered BLS data to be the leading source in the field.
Daubert, ___ U.S. at ___, 113 S.Ct. at 2795 requires that expert testimony be based upon the methods and procedures of science. For the moment this Court will set aside its doubts as to the potential validity of a willingness-to-pay approach to the value of life to begin with, but will instead assume that the doctrine can under appropriate constraints qualify as a "science." But to locate its $3.5 million benchmark within the broad range of values contained in the willingness-to-pay literature, Hedonic Damages employs a simple eyeballing technique. Eyeballing may have the advantage of ease, but it surely lacks scientific reliability in the sense of producing consistent results (id. at ___ n. 9, 113 S.Ct. at 2795 n. 9). Anyone can look at the same range and come up with a different figure. It also contributes to gameplaying. Note, for example, how quick "Dr. Economist" in the sample testimony (presumably Smith himself) is to exploit the method's malleability by suggesting that he could have picked a higher number (like $5 million) and that by opting for the lower figure he is somehow rendering a "conservative" opinion. Maybe so, but a conservative opinion in that sense does not equate to a scientific one. Someone who states on the basis of a dull pain in his right knee that he thinks it is going to rain less than .1 inch expresses a conservative, but surely an unscientific, opinion.
Plausible Range spends a good deal more effort in deriving its $2.2 million mean. From the end result alone (Graph C) there is room for the conclusion that the willingness-to-pay model values life at approximately $2.2 million (though even the manipulated ultimate spread, almost equally spaced between a figure close to $1 million and one over $3.5 million, is troublesome as the predicate for picking a number in midrange — or indeed any number — as the value). But the real problem is in the classically Procrustean way that author Miller gets to that stage: Any objective reader can recognize the methodology as one that has made whatever adjustments were necessary to bring the raw data within a target range. Adding, subtracting, multiplying and dividing with a predetermined result in mind cannot fairly be labeled science. To put it bluntly, Plausible Range strikes this Court as a prime candidate for an Oscar in the most-misleading-title category.
Nor is that the only difficulty with the suggested results. To reexamine what has just been assumed arguendo, the entire process of selecting and adjusting willingness-to-pay data has proved unreliable because of the widely divergent views among economists concerning what does and does not constitute not a sound study. Thus the author of Plausible Result, writing a year earlier (Miller, 83 Nw.U.L.Rev. at 882 nn. 29 30), reported a 1985 survey of 17 willingness-to-pay studies by another economist that had concluded that the best three of those priced the value of life at approximately $8 million, almost four times the Plausible Result figure. Those "best three" studies are among the ones whose results Plausible Result divides by 3. It is also beyond cavil that the dozen-or-so economists who used BLS and SOA data in their own wage-risk studies disagree sharply with Plausible Result as to the quality of those statistics. In sum, neither the $3.5 nor the $2.5 million benchmark rests upon any scientific method or procedure, so that testimony regarding either one is inadmissible under the scientific knowledge prong of Rule 702.
It may be worth mentioning — and rejecting — Smith's use of regulatory statistics to bolster his argument that willingness-to-pay data congeals around $2.5 million. Government regulations are really a different kettle of fish for two reasons. First, it is not at all clear that agencies use willingness-to-pay or any other purportedly scientific approach to estimate life values. Beyond the more general observations that can be made about bureaucratic and political processes and the way in which decisions are reached at those levels (Mercado, 974 F.2d at 871), the willingness-to-pay literature itself illustrates the way in which government agencies abandon science to secure politically acceptable compromises (Miller, 88 Nw.U.L.Rev. at 886-87 (footnotes omitted)):
The Occupational Safety and Health Administration's (OSHA's) use of risk reduction values in regulatory analysis led the agency into a major conflict with the Office of Management and Budget (OMB) in 1985. OSHA initially attempted to use a value of $3,000,000 per anonymous life saved. The OMB decided that this value was too high and suggested that OSHA use a value of $1,000,000. Eventually, OMB and OSHA reached a compromise of $2,000,000 per life saved.
Second, in the regulatory context the range of estimates is enormous and Smith (like the author of Plausible Result) is simply cherry-picking (McClurg, It's a Wonderful Life, 66 Notre Dame L.Rev. at 106 n. 214) (citations omitted):
Although they do not always reveal their methodologies, federal agencies have set life values as low as $70,000 and as high as $132 million per life.214
2. Adjustments
By definition the willingness-to-pay model estimates the value of a statistical life — a nameless, faceless member of society. This Court's trial jury will have a very different task: It must value the life of a specific individual (Lenardo), the quality of whose life may have been substantially richer or poorer than the statistical mean. Applying Hedonic Damages' definition of the hedonic value of life to persons in radically different situations illustrates the point. Compare under that definition (1) a person who is sick with one who is healthy or (2) someone very old with someone very young. Or compare (3) a person with a loving family with one who has none, or (4) someone with many friends versus someone who has none. Try it with an arguably tougher class of criteria — say (5) wealth, (6) race, (7) intelligence, (8) education, (9) sex, (10) character, (11) reputation — and the complexities become plain. That process reveals a lack of fit (Daubert, ___ U.S. at ___, 113 S.Ct. at 2796) because the willingness-to-pay data upon which Hedonic Damages relies goes to one thing (value of a statistical life) while the jury is called upon to determine a potentially very different figure (value of Lenardo's life).
Hedonic Damages purports to address that problem by adjusting the benchmark to account for certain specific characteristics of the plaintiff. Here is the calculation in the Jack Doe hypothetical (Hedonic Damages 170-72):
$3,500,000 benchmark less 800,000 lost earnings, fringe benefits --------- and household services equals $2,700,000 hedonic value for average person divided by 45 life expectancy of average --------- person equals $60,000 per year hedonic value for average person
Hedonic Damages then assumes a life expectancy of 75 years and calculates the present value of Jack Doe's hedonic life with a 1.29% rate of growth and a discount factor of 3.13%.
Thus Hedonic Damages claims that the final figure ($1,709,842) accounts for Jack Doe's age, race and sex. That too is misleading, and in combination with the relatively low probative value of the testimony also calls for its exclusion under Rule 403. As described, the adjustment calculus makes an allowance for age and perhaps for income as well (its use of a 1.29% rate of growth is tied to Jack Doe's projected earnings and an assumption that the rich have a greater capacity to enjoy life). D.Mem. 7 n. 3 suggests that Smith has backed off of the latter position, stating there that "[i]n a deposition given in another case on November 17, 1994 [citation omitted] Smith reiterated the position the position [sic] that rich and poor have the same value of life." If so (and Lillian has offered nothing to the contrary), it is clear only that the adjustments address the decedent's age. As for the sex of the decedent, it would appear to be taken into account only in the indirect sense that the factor is reflected in the actuarial tables — and that would appear to be true a fortiori of the factor of race.
Whether sex or race (or both) matter to life fulfillment is an interesting question, but one into which Hedonic Damages does not delve. Its authors' representation that it has done so is grossly misleading. Most fundamentally, Hedonic Damages has gone only a small part of the way toward addressing the fit problem. Age is 1 of 11 factors identified at the beginning of this section as potentially separating Lenardo's hedonic value of life from the statistical mean, but the list could well have been several times as long. In sum, the low probative value of such testimony (ill-fitting data) is substantially outweighed by the danger of unfair prejudice (a false appearance of tailoring to the individual case).
3. Pedigree
Another troubling aspect of Smith's testimony is his use of Adam Smith's 1776 work An Inquiry into the Nature and Causes of the Wealth of Nations (Edwin Cannan ed., University of Chicago Press 1976) to lend pedigree to his theory. Adam Smith is of course the undisputed godfather of modern economics and The Wealth of Nations his crowning achievement. Unfortunately for Stan Smith, the surname Smith seems to be about the only thing they have in common.
Very little in the classic Wealth of Nations supports the Hedonic Damages method of valuing life, and much counsels against it. Adam Smith was a skeptic when it came to the ability of people to perceive risk (example: lotteries, id. 120-21), and he also observed that in certain contexts the "fear of misfortune" is overbalanced by "the hope of good luck" and the fancied "occasions of acquiring honour and distinction," particularly on the part of the young who thus do not make rational economic judgments about wages and risk (examples: soldiers and sailors, id. 122-23). Both of those observations remain perceptive today: As to the former, its continued accuracy is evident in the exceedingly poor odds and enormous popularity of state lotteries, while as to the latter, one need only view many of our armed forces' television commercials that portray the physical challenges of serving and are also designed to appeal to a young person's sense of adventure.
Adam Smith did comment that certain laborers (example: coal miners, id. 112) demand higher pay as a result of employment risks and incorporated that into a general theory of wages. But the leap from that observation to a theory for valuing life was not taken for almost 200 years: It is recognized that the genesis of the willingness-to-pay method of valuing human life is T.C. Schelling's The Life You Save May Be Your Own (reprinted in the Brookings Institution's study Problems in Public Expenditure Analysis (Samuel Chase, Jr. ed., 1966)). In that piece Schelling set out the model's basic framework, and nothing has really changed in that respect since. Schelling's purpose was to suggest a method for conducting cost-benefit analyses in the context of proposed safety regulations. It was another 20 years before the further leap into the courtroom was first taken, and this time the economist was Stan Smith (Sherrod, 827 F.2d at 205-06).
As a method for measuring statistical life in a regulatory context, then, the theory's genealogy traces back only to Schelling, while as a method of calculating damages in court Stan Smith is (or hopes to be) the father of what is really an infant industry. By seeking to portray as a genealogical credential the exceedingly tenuous connection between his willingness-to-pay methodology and Adam Smith's Wealth of Nations, Stan Smith coats his novel use of a quite recent economic theory with a vintage veneer that it does not deserve. Because the sample testimony involving Adam Smith's classic work thus provides little support for Stan Smith's theory and much unfair prejudice when used to influence a lay jury, it too is inadmissible under Rule 403.
4. Empirical Data
After Schelling had suggested a new way to measure the value of life, a number of economists set out to determine what results would flow from the application of that theory to the real world. Over a 20-year period they found that the willingness-to-pay model values life somewhere between $500 thousand and $9 million (see Graphs A and B).
Even if this Court were to find that the methodology underlying those studies constituted "science" as that term is properly understood, it would still have to exclude them under the helpfulness standard of Rule 702. It would really be no use to a jury to hear that others had placed the statistical value of life at greater than $500 thousand and less than $9 million (especially when told of how those numbers had been arrived at — see Mercado, 974 F.2d at 870-71). And it is frankly bogus to massage those numbers, as both Hedonic Damages and Plausible Result have done, to create a deceptive appearance of precision rather than the true picture of an enormous spread in "value." That is both fatal in Rule 702 terms and a basis for exclusion under Rule 403's balancing test.
5. Underlying Assumptions
As to the basic assumption underlying the economic model — that the amount of individuals' willingness-to-pay for small reductions in the likelihood of death reflects how society values life — this Court shares many of the concerns expressed by our Court of Appeals in Mercado, 974 F.2d at 868-71 and mirrored in much of the academic literature (see, e.g., McClurg, 66 Notre Dame L.Rev. at 103-06; Linnerooth, 17 Econ.Inquiry at 53-54; and sources cited in each). Those criticisms focus on (1) the assumption that people have freedom of choice in deciding to confront risk, (2) the assumption that people perceive risk accurately, (3) the nonmonetary factors that drive many consumer purchases (e.g., advertising) and employment decisions (e.g., civic pride), and (4) the political aspects of government regulation (e.g., budgets, lobbyists). This Court agrees that those considerations go a long way toward undermining the basic premise behind the willingness-to-pay model.
On the other hand, a great deal has been written on the subject over a number of years, some of it in respected academic journals. P.Mem. 9-12 n. 3 lists 58 articles, papers and other works. That listing too is deceptive, for a good deal of it is fluff, but a half-dozen-or-so of the listed articles have been published in discriminating periodicals (see bibliography in Vasile Tega, Management and Economics Journals (1977)). For current purposes it is unnecessary to decide whether there has been enough testing or peer review of the underlying theory to constitute scientific knowledge. Stripped of its empirical moorings, Schelling's basic insight into risk and the value of life cannot possibly assist the jury in calculating damages, so it is plainly inadmissible under the helpfulness prong of Rule 702 — regardless of its fate under Rule 702's scientific knowledge requirement.
One other consideration, already touched upon, plays out for the most part under Rule 403 and helps to explain why willingness-to-pay methodology may be an appropriate guide for regulators but not for courts and juries traversing the difficult terrain of valuing life. Those two activities have very different foci: Regulators deal in averages, while courts deal with specific cases. Thus a statistical mean may have validity in the former context, while at the same time it simply creates the already-mentioned deceptive appearance of precision in the latter.
* * * * * *
In sum, viewed in term of "scientific knowledge" the hedonic damages proof tendered by Lillian has thus failed to survive Daubert analysis. So Lillian's motion has failed on the battleground selected by the parties.
Other Specialized Knowledge?
As already indicated, both sides' submissions have taken the position that Smith's testimony is "scientific" in nature and is therefore properly analyzed under Daubert. But at least in the present state of the art (?), a substantial argument might be made for disqualifying that testimony from the "scientific" label in the first instance. That however would not necessarily end the analysis, for Rule 702 also speaks of "other specialized knowledge" (it may be noted that "specialized knowledge" is the label attached by Mercado, 974 F.2d at 871 in upholding a trial judge's decision to bar Smith's testimony).
By its express terms Daubert speaks only of "scientific" testimony and not of "other specialized knowledge" (___ U.S. at ___ n. 8, 113 S.Ct. at 2795 n. 8). It is an interesting question, not yet resolved by the lower federal courts post- Daubert, whether Daubert analysis extends to the other Rule 702 branches ("technical knowledge" is of course the third category referred to there) and, if so, with what possible adaptations. But that question need not be addressed here, not just because the parties have limited their quarrel to "scientific knowledge" but also because the factors discussed in this opinion demonstrate the inadmissibility of the Smith testimony from any perspective.
Conclusion
All five principal parts of Smith's proposed testimony on the issue of hedonic damages are inadmissible under Rule 702 or Rule 403 or both. Defendants' motion to exclude the testimony is granted in its entirety.