Are All Lives Equal?: Why Cost-Benefit Analysis Values Rich Lives More and How Philosophy Can Fix It 9798986128610

Citation preview

Are All Lives Equal? Why Cost-Benefit Analysis Values Rich Lives More and How Philosophy Can Fix It  

Carneades

Copyright © 2022 Carneades Published by Carneades.org All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from Carneades.org. ISBN: 979-8-98-612861-0

   

DEDICATION     This book is dedicated to all of the supporters, patrons, subscribers, and viewers of the Carneades.org YouTube channel. Thank you for all the support, interesting comments, and engaging discussions over the years. Stay Skeptical Everybody!

   

CONTENTS     Acknowledgments

vii

 

List of Thought Experiments

ix

1

Let Them Eat Toxic Waste

1

I 2

How We Got Here Consequentialism and Cost-Benefit Analysis

  Pg 17

3

How to Value a Life

Pg 32

II 4

Insufficient Responses That’s Not What VSL Means

  Pg 49

5

Life is Priceless

Pg 61

6

Nothing Compares to Life

Pg 74

7

The Average Cover-Up

Pg 109

8

Why Some Lives Are Worth More

Pg 124

III 9

Percentage Willingness to Pay The Problem with Absolute

  Pg 149

10

Percentage Willingness to Pay

Pg 168

11

The Justice of Percentages

Pg 187

12

Unequalland (an Example)

Pg 200

13

The Lie at the Heart of Economics

Pg 227

14

The Value of Black Lives

Pg 243

15

Covid-19: Your Money or Your Life

Pg 263

16

Objections and Responses

Pg 296

17

Final Thoughts

Pg 344

 

Appendices

Pg 355

 

References

Pg 383

 

Disclaimers

Pg 405

 

About the Author

Pg 407

         

       

ACKNOWLEDGMENTS     Thanks to Michael Blake for workshopping a much earlier version of this book and providing a number of the objections that are addressed. Also, thanks to Elizabeth Richardson Vigdor for giving me the idea for this book and providing a number of the economist’s responses. Finally, much appreciation to Juliana Žamoit for the fantastic illustrations and cover art and to Barry Lyons for copy editing.  



LIST OF THOUGHT EXPERIMENTS       1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

Thanos and the Gambian Disease The Happiness Meter Buba and Fatou’s Policy Preferences Gambian Micromorts and a Gallon of Milk Sarjo’s Incommensurate Preferences John’s Strange Views on Abortion Mary’s Odd Opinions About Brexit The Speed Limit in Speedy Hills The SWAT Team Trolley Problem The Chief of Medicine’s Transplant Problem Camperville and Hungrytown Cowardsville and Braveton Farmerstan and Sickmavia An Economist’s Democracy Measuring The Fear of Piranhas The Zoning of Nimby Grove Haruna and John Become Fathers Koula and Arax Get a Raise The Worth of Mike, Sarah, Bintou, and Lamin Rich vs Poor Labor Markets Melati, Karen, and the Tsunami Marco, Ron, and the $100 Bill VSL and the Veil of Ignorance Millionaire’s Peak and the Community Park A Coal Plant Comes to Millionaire’s Peak Dirtsville Needs a Road Regulating BanksCorp Bell Curve Valley Mallmart and Costgo Lemonade Stand Happiness Mount Diamond and the Space Heaters Leadpipeville and the Bottled Water Redline Town and Greenyard Hills Rarity Fever and Apocalyptococcus

Ch. 1 Ch. 2 Ch. 2 Ch. 4 Ch. 6 Ch. 6 Ch. 6 Ch. 6 Ch. 7 Ch. 7 Ch. 7 Ch. 8 Ch. 8 Ch. 9 Ch. 9 Ch. 9 Ch. 9 Ch. 10 Ch. 10 Ch. 10 Ch. 10 Ch. 10 Ch. 11 Ch. 12 Ch. 12 Ch. 12 Ch. 12 Ch. 12 Ch. 13 Ch. 13 Ch. 13 Ch. 13 Ch. 14 Ch. 15

35 36 37 38 39 40 41 42 43 44

Rarity Fever Redux, now with Coffee! Tinyworld and Vaccine Equity Conspiracyland and the Pig People Plutocracy from Pluto Apocalyptococcus Redux, Dictator Edition Illsboro and Metrosick Farmers and Cowmen in Equihoma City Rene and the Tiny Tax Trickle Down with Anne Frances and Gipp Meritoworld and Statictopia

                   

 

Ch. 15 Ch. 15 Ch. 15 Ch. 15 Ch. 15 Ch. 16 Ch. 16 Ch. 16 Ch. 16 Ch. 16



 

 

1. LET THEM EAT TOXIC WASTE   In 1991, Lawrence Summers argued that wealthy nations had a moral imperative to dump toxic waste on the shores of least-developed countries because the lives of those people were simply worth less. This might be slightly concerning, though not unexpected, had Summers been a white supremacist writing manifestos in his basement. Unfortunately, at the time Lawrence “Larry” Summers was the chief economist of the World Bank, one of the largest multilateral development banks in the world, writing a policy memo. Even worse, his claim is not only supported, but required by the underlying tenants of economics. This conclusion was not due to any bias on his part but to a necessary consequence of basic economics. The memo argues that, due to the lower “willingness to pay” of individuals in least-developed countries for reductions in risk, societal benefit would be increased if dangerous chemicals and toxic waste were sent to developing countries, as individuals in wealthy countries simply value their lives more.[1] Basically, if someone is going to die from toxic waste, it is better if it is someone that places a lower value on their life. Here is an excerpt: “The measurements of the costs of health impairing pollution depends on the foregone earnings from increased morbidity and mortality. From this point of view a given amount of health impairing pollution should be done in the country with the lowest cost, which will be the country with the lowest wages. I think the economic logic behind dumping a load of toxic waste in the lowest wage country is impeccable and we should face up to that.” Afterward, Summers claimed that the memo was meant to be a sarcastic critique of the World Bank’s policies (and later claimed to only have approved it, not written it himself). He also claimed that this was a necessary result of the focus in economics on increasing total benefits for the world through trade liberalization at the expense of costs to the poorest nations.[2] This sentiment can be read into the final lines of the memo, which claim that arguments against these claims “could be turned around and used more or less effectively against every Bank proposal for liberalization.”[3]

Regardless of the intent of the memo, the conclusions are horrifyingly accurate. Economics inherently assumes that to maximize happiness for the whole world, you should put into place policies that benefit the wealthiest, as they are more willing to pay for those benefits. The goal of this book is to offer a solution to this fatal flaw in economics without abandoning the discipline altogether. Cold Hard Economic Logic Whether you think the memo was written as a critique of the basic economic principles that place lower value on the lives of the poor or a manifesto supporting such a view does not change the underlying economic logic. Summers is factually correct about what economics implies, even if he is morally wrong. When aggregating the value of saving lives using costbenefit analysis, the life of a rich person simply counts for more. These analyses focus on maximizing total willingness to pay, even when that disproportionately harms those with the least ability to pay. This is not to say that Summers or others are morally corrupt. It is merely to point out that the current practice of cost-benefit analysis places less value on the lives of the poor than it does on the lives of the rich. This is not an isolated case because the differential valuation of life by income is required by the underlying principles of economics. In 1995, the International Panel on Climate Change (IPCC), a body of the United Nations, estimated that the lives of individuals in poor countries were worth around $100,000, those in middle income countries were worth $700,000, and those in high income countries were worth $1.5 million.[4] Similar, more recent, calculations have been done by organizations ranging from the Gates Foundation to the U.S. government.[5] Even if we set aside concerns about the very project of valuing human life for the moment (never fear, we will revisit them in the second section of this book), we should be concerned that the life of one person in a rich country is worth the equivalent of 15 lives in a poor country. These disparities have only grown as global inequality has grown. In their 2017 paper “Income Elasticities and Global Values of a Statistical Life,” Kip Viscusi and Clayton Masterman calculate the value of a statistical life (VSL) for all countries that have World Bank income data (see Appendix B for a reproduction of this table from Viscusi and Masterman, 2017). According to the authors “we calculate the average

VSLs in lower income, lower-middle income, upper middle income and upper income countries to be $107,000, $420,000, $1.2 million, and $6.4 million, respectively.”[6] However, these are not the most extreme cases: the life of someone from Bermuda ($18.2 million) is equivalent to the lives of 405 people from Burundi ($45,000).[7] These differences are built out of the way economists create the value of a statistical life; they are baked in. Even a paper funded by The Bill and Melinda Gates Foundation, which greets website visitors with giant letters proclaiming “All lives have equal value,” has commissioned a paper that shows exactly the opposite.[8] This 2018 paper by Lisa Robinson, James Hammitt, and Lucy O’Keefe had a slightly smaller range than Viscusi and Masterman’s (though this is mostly due to not calculating values for very small, very rich countries like Bermuda), but the broad implications are the same.[9] Even the U.S. government, despite hiding the differences for its own population using averages, values lives in poor countries much less than rich ones when deciding how to allocate aid. [10] An ethical conclusion taught in most elementary schools, that all

 

lives have equal value, seems to have eluded the best economic minds of the world.   Ethical Consequences These methods have real consequences in the policies that they inform. The result of these methods is an economic imperative to invest in policies that save the lives of the rich over the lives of the poor—just as Summers advocated in his memo. To illustrate the impact of such policies, imagine that you are the chief economist at the World Health Organization. Imagine that two deadly diseases have hit the world, one affecting only Americans, another affecting only Gambians.[11] These diseases have no other economic effects other than killing people. They kill in exactly the same way, but they kill at different rates. In America, your risk of mortality from this disease is 0.003%, so that, based on current population estimates, it will kill 10,815 people before the entire population gains immunity.[12] In The Gambia, your risk of mortality from the disease is 50%, so that based on current population estimates it will kill 1,050,284 people, or half of the population. [13] Stop the American disease and you save ten thousand lives, stop the Gambian disease and you save one million. Unfortunately, in this thought experiment you only have enough funding to cure one of these diseases, and only the WHO has the resources to undertake this research; no other organization is going to step in if you do not act. Your initial reaction may be to save the lives of Gambians from the simple and intuitive claim that it is better to save as many lives as possible. However, as a public official you are bound to take the course of action that will maximize benefits to the world, not act solely based on your own intuitions. Therefore, you conduct a quick cost-benefit analysis. Using the willingness of individuals in each country to reduce their risk of death, you find that the value of a single American’s life is $9,631,000 while the value of the life of a Gambian is only $79,000.[14] Saving the lives of 10,815 Americans is worth $104 billion, while saving the lives of 1,050,284 Gambians is only worth $83 billion. To your dismay, your intuition was wrong. You are morally obligated to cure the American disease to maximize benefits. Therefore, with a Thanosian snap you choose to cure the Americans and let half of the population of The Gambia die. For every American you save, you let 100 Gambians die. You let one million

poor people die to save ten thousand rich ones, confident that you are implementing policies supported by a rigorous cost-benefit analysis. My intuition is that there is something deeply unethical about the above conclusions. However, current discussion of these issues has led to philosophers and economists talking past each other. There is a great deal of philosophy that dismisses the entire discipline of economic analysis out of hand as inherently immoral. Some philosophers would happily watch the economics department burn to the ground and argue that such an accident maximized good in the world. Similarly, there is a strain of economists that dismisses philosophy’s ethical concerns as impractical and divorced from reality. A not insignificant number of economists take after Aristophanes, thinking that philosophers have their heads in the clouds, caring only about amorphous immeasurable concepts, with neither an appreciation for the real-world consequences of their theories nor an understanding of what economists mean by the term “value of a statistical life.” In what follows I will take neither position but instead will attempt to bridge the gap between these two acrimonious disciplines. I acknowledge the utility of cost-benefit analysis as a data-driven decision-making tool, while respecting that any tool that advocates killing one million to save ten thousand needs to be recalibrated but not necessarily discarded.

Economists value saving the lives of the rich more than saving the lives of the poor.  

Outline The remainder of this book is divided into three sections. This first section serves as an introduction for dismayed philosophers and other noneconomists into how economics can arrive at such a troubling conclusion. It includes an explanation of how economists value things, how cost-benefit analysis is conducted, and how lives can possibly be valued. It also includes an investigation into the philosophical underpinnings of this economic project in the form of utilitarian and consequentialist arguments. The second section covers five previously proposed solutions to this dilemma, two from philosophers, and three from economists. The two solutions offered by the philosophers amount to throwing the baby out with the bathwater, and risk hiding and perpetuating existing biases. The three responses that come from economists simply disguise the problem while continuing the inequity. The philosophical responses come from the works of Frank Ackerman and Lisa Heinzerling (2004), Mark Sagoff (2004), Elizabeth Anderson (1993), and Joseph Raz (1986). These responses either fail to solve the initial problem, or amount to attempts at replacing costbenefit analysis with a system that is less transparent and has greater potential for bias. The first economic response is to claim that this problem is simply a miscommunication; philosophers just don’t understand what economists mean by these calculations. It is defended by Trudy Ann Cameron (2010). The second economic response, which is currently used by the EPA and the UN,[15] is to hide these disparities through the use of averages. The third economic response, offered by Cass Sunstein, argues for disaggregation even by factors that should be ethically irrelevant (2004). In the third and final section, I make the case for a new, better response that more accurately represents individual welfare.[16] My solution resolves the problem equitably by advocating for the use of percentage willingness to pay, instead of absolute willingness to pay. This solution would value lives based on ethically salient characteristics, such as preferences, instead of morally irrelevant facts like wealth. I demonstrate that the failings of the previous responses point toward my solution, which preserves the usefulness of cost-benefit analysis and allows for the disaggregation of lives by risk preference, without valuing the lives of the wealthy significantly more than the lives of the poor. This is done by using a percentage measure of willingness to pay instead of an absolute measure of willingness to pay. I

show that this small change in how we calculate benefits could have an outsize impact on the fight to reduce global inequality, racial bias, global pandemics, and justify policies that benefit the poor as much as they benefit the rich. This small change in the methodology of cost-benefit analysis will not only make it more accurate, it will make it more ethical too. There is no need to get rid of this useful, transparent, data-driven tool for policy making. However, there is a need to clean it of its current bias by removing ethically irrelevant factors, such as wealth, from the calculation. Using a percentage measure of willingness to pay appreciates the intuition that some people do value their lives more, without engaging in a systematic bias against the poor. Unlike the previously offered solutions, percentage willingness to pay is fair, accurate, and objective.  

 

 

           

I. HOW WE GOT HERE A brief overview of economics, cost-benefit analysis, and the value of a statistical life                  



 

2. CONSEQUENTIALISM AND COST-BENEFIT ANALYSIS   Philosophers are notoriously concerned with the big picture questions. What is justice? What makes something beautiful? Can we know anything? Other disciplines are then left to answer those pesky small picture “how” questions. Political scientists deal with the day-to-day intricacies of how to enact justice, artists spend their lives determining how to make something beautiful, and researchers attempt to operationalize our epistemic principles to learn new things. In order to understand where economics runs afoul of philosophy we should look to its philosophical underpinnings in ethics, more specifically in consequentialism. For the uninitiated, consequentialism is simply the normative position that a particular action is made good or bad on the basis of its consequences.[17] Regardless of the justification for consequentialism as a personal moral position, it is essential for certain kinds of policymaking, as we will argue in the following section. There are a range of types of consequentialism, with the most widely recognized being utilitarianism, the claim that we should maximize the amount of happiness, well-being, or utility in the world.[18] For the philosopher with the bird’s eye view, implementing this theory as a policymaker seems like a clear and simple project. You determine the possible choices you can make (e.g., build a hospital or build a school); look at their outcomes (e.g., more health or more education); count up the happiness created by each; and choose the outcome that provides the most happiness to the most people. Simple as this may sound, there are a range of detailed questions that are left to the non-philosophers, in this case the economists, to answer: how do we measure happiness or utility? How can we compare different types of happiness or benefits using a common scale to make rational decisions? Should we allow policies that cause pain and happiness so long as the happiness outweighs the pain? Those partial to a different strain of consequentialism can simply replace happiness with welfare or desire

fulfillment, the basic economic questions of how to measure whatever concept is chosen will remain.

To understand the project that philosophers have tasked economists (specifically welfare economists) with, imagine everyone has a happiness meter that goes from 1 to 10. Imagine further if we had a supercomputer that was able to calculate how any decision you made would affect the overall happiness of the world. The goal of a just government would be to enact policies that increase or maximize the net possible happiness in the world. Simplistic as it may sound, this is the basic framework that economists use. Measure the benefit that everyone gets from a policy, subtract their pain or costs, then add them all up. Enact policies that have a positive benefit, or the most positive benefit of a range of choices, and do not enact policies with a negative benefit or a net cost.[19] In many ways, modern cost-benefit analysis is an attempt to operationalize these basic consequentialist intuitions.[20] Measuring Value There are a range of potential goods that a consequentialist might wish to maximize, including pleasure, utility, desire fulfillment, and welfare.[21] The challenge for the economist is that these goods are

unclear, and difficult to quantify. What units can we use to compare the happiness that a boy gets from riding a roller coaster for the first time, and the happiness a woman gets from receiving her first promotion? Some philosophers, such as Anderson and Sagoff,[22] claim that these things may simply not be comparable, that the very project of aggregating the benefit from these goods is flawed (however, this comes with its own slew of problems for policymakers as we will see in section II). Economists, on the other hand, have a simple method for comparing these seemingly incommensurable goods: money.[23] Specifically, economists use a concept called “willingness to pay” in order to measure benefit.[24] This comes from the intuitive notion that you get some inherent pleasure when you receive something for less than you would be willing to pay for it. Your benefit is determined to be the amount that you would have been willing to pay for a particular item minus whatever you ended up paying for it. This basic economic principle seems quite intuitive at first glance. If you would be willing to pay $5 for a hamburger and I sell it to you for $4, you have $1 of benefit left over. You have more benefit than if I had have sold you the hamburger for $4.50 and less than if

I had sold you the hamburger for $3, or given it to you for free. The opposite of benefits, referred to by economists as costs, can also be measured in these terms. How much cost is incurred when you drop your computer and break it? The same amount that someone would need to pay you to make you indifferent between dropping the computer and having the money (possibly enough money to fix your computer or to buy a new one but possibly more if there are files on that computer that could not easily be replaced). Note that this may not be equal to the original value of the computer, particularly if your willingness to pay for the computer was greater than what you actually paid for it. There is something very important to understand about willingness to pay as a measure of benefit. Willingness to pay is constrained by ability to pay.[25] What this means is that you cannot be willing to pay more than you have. You cannot be willing to pay $5 million dollars for a hamburger if you only have $5,000. Therefore, your maximum willingness to pay for anything is constrained by your wealth. This limitation is particularly

important, as it is this very limitation that eventually leads to the issue of unequal valuations of life. If willingness to pay more accurately measured

benefits and costs, these inequalities would not appear, as will be shown in section III.   Cost-Benefit Analysis Cost-benefit analysis (sometimes called benefit-cost analysis, CBA, or BCA) answers the second question for economists. Once we have a way to measure benefits and costs, how do we aggregate these benefits and costs into a decision tool that illuminates for policymakers the best course of action? Cost-benefit analysis is a tool for analyzing policies and determining if the benefits are worth the costs.[26] These are commonly used by governments and other public institutions worldwide to determine whether a particular policy is worthwhile. Boardman et al. describes the process as:   “a systematic cataloguing of impacts as benefits (pros) and costs (cons), valuing in dollars (assigning weights), and then determining the net benefits of the proposal relative to the status quo (net benefits equal benefits minus costs).”[27]   I can best describe this process as an operationalization of the classical utilitarian calculus, which attempts to enact policies that maximize overall happiness, or benefit, while minimizing pain or cost, and some (such as Sunstein) have framed the process in such terms.[28] Basically, the goal is to find the net benefits: the difference between the total good done by a policy and the costs of the policy (as compared to the next best option). According to this framework, there is a moral (or at least economic) imperative to enact the policy with the greatest net benefits.[29] This framework is particularly useful for policy makers because it provides an objective, data-driven method for analyzing the effects of a particular policy. In the absence of objective measures and tools, it is common for humans to make choices based on their intuitions. This process is not only opaque, but it can mask a range of unconscious biases for certain kinds of polices or beneficiaries that even the decision maker may be

unaware of. As much as a politician may claim to be unbiased, without objective analytic tools they are often consciously or subconsciously predisposed to choose policies that benefit people that look or act like them. Cost-benefit analysis is fully transparent, and if it contains any biases, they are clear for all to see. Intuitively, the attempt to maximize overall happiness or benefits seems in line with utilitarian sentiments. Costs and benefits must be placed into common units so that policies can be compared. Most commonly, this means that they are monetized, or converted into dollar values.[30] This subtle shift, from focusing on maximizing utility to maximizing dollar values is important. If absolute willingness to pay is a poor measure of utility, it loses the economic and moral imperative to drive policy. I argue in the final section that percentage willingness to pay is a better measure of utility than a dollar value absolute willingness to pay. Costs are calculated in terms of the “opportunity costs” of inputs (the benefit value of the next best use of those resources), or the willingness to pay to avoid the cost as discussed above. Benefits are calculated in terms of willingness to pay to receive that benefit. Costs for each individual are then subtracted from the benefits. This provides us with a number of net benefits for each policy. A policy that costs one person $1, but provides $2 of benefit to each of two other people would have a total cost of $1, a total benefit of $4 ($2 times two people), and a net benefit of $3 ($4 benefit minus $1 cost). However, this does not explain which of these three values (total cost, total benefit, or net benefit is the most important). The next question is how to determine whether or not to enact this policy given these values. There are two decision rules that might be

used to compare policies, the Pareto Efficiency criterion and the KaldorHicks criterion.   Decision Rules The Pareto Efficiency criterion claims that we should enact the policy that maximizes the total good in the system without hurting anyone.[31] It can be thought of as something of a Hippocratic Oath, where the policy should do no harm. Under this decision rule, no one can bear a net cost (e.g., we could not implement the policy above since one person bears a net cost of $1).

While this may be initially tempting, note that “harm” in this case is equivalent to cost. Using Pareto Efficiency, if one person had to pay a single dollar in taxes and got no benefits, but hundreds of lives would be saved, you would not be able to enact the policy because one person would be made worse off without being compensated. Therefore, this criterion is rarely, if ever, used in actual cost-benefit analysis (though it can be seen as a lower bound: if something creates benefit with no net cost to anyone, it is hard to argue with). As we will show in section II, this concern is also the reason that consequentialism (as opposed to deontology) is more fit to deal with these issues of policymaking. The Kaldor-Hicks criterion (or potential Pareto Efficiency), on the other hand, focuses on maximizing total benefit. Instead of summing the net benefits for each individual, it sums the benefits for the whole society.[32] If this net benefit is greater than the other policy options, including the status quo, then it should be enacted. If the net benefit is negative, it costs more than the benefits it provides, then it should not be enacted. Some versions of cost-benefit analysis require that benefits exceed costs by a certain percentage.[33] This should intuitively map onto utilitarian principles of maximizing the greatest good for everyone, not merely for any given individual. This might be justified either through reference to the arguments that support utilitarianism, or the economist’s claim that we should support any policy where at least in principle the winners could compensate the losers. In other words, there is enough gained by those that benefit to compensate those who bear the cost. If enough policies are enacted, the assumption is that everyone will eventually be a “winner” and therefore everyone will benefit. Philosophers have raised concerns with this argument, as often there are those who never benefit, and those that do benefit do not in fact compensate those that bear the cost. However, there are much more persuasive arguments for consequentialism offered by other philosophers than this simple economic explanation, as we will see in section II. Buba and Fatou In order to demonstrate how all of these concepts might be operationalized, consider the following example. Buba is willing to pay a maximum of $100 for a policy to be implemented. Therefore, he is considered to gain a benefit of $100 if that policy is implemented. Another way to frame this is that

Buba is indifferent between getting $100 and having the policy implemented. The underlying idea is that willingness to pay reveals the degree of benefit that an individual receives for a particular good. The more you are willing to pay for something, the more benefit you get from it. If this were the only consequence of the policy it would pass both the Pareto Efficiency Criterion (as someone is made better off, and no one is made worse off) and the Kaldor-Hicks Criterion, (as the sum of all benefits, $100, exceeds the sum of all costs $0). Willingness to pay can also be used to measure costs, or negative benefits, by determining how much someone would need to be paid in order to accept a particular policy. If Fatou only accepted this same policy (or more accurately be indifferent between the policy and the status quo) if she were paid $25, then we would consider her to be bearing a $25 cost. This would also be the case if Fatou had to pay an additional $25 in taxes for such a policy. If Fatou and Buba were the only people affected by the policy, then the policy would have a positive net value (because $100 – $25 = $75, and $75 is positive). Therefore, this policy would pass the

 

Kaldor-Hicks criterion (as the total net value is positive), but not the Pareto Efficiency criterion (as someone, Fatou, is made worse off). As before, it is crucial to note that willingness to pay is constrained by ability to pay. In other words, someone with only $200 to their name can have a maximum willingness to pay for a policy to be enacted of $200, while someone with $100,000 could have a maximum benefit of $100,000, or be willing to pay $100,000 to be indifferent between that policy being implemented and the status quo. This constraint leads to concerning implications for which policies appear to have greater or lesser benefits. There are a range of other nuances of cost-benefit analysis that have interesting and troubling conclusions. One of which has been written about extensively in the philosophical literature is the discounting of future benefits.[34] Basically, future dollars are not worth as much as present dollars.[35] The intuition arises from the simple question: would you rather have $1,000 now or $1,000 in twenty years? Economists assume that most rational people would rather have the money now, and therefore the present dollars are worth more than the future dollars. How much you would need to be paid to wait 20 years can tell you just how much. This practice leads to more concerning conclusions such as the lower valuing of future lives due to discounting of future benefit streams. This can have a significant impact, especially when looking at environmental policies that have costs far into the future.[36] These are important issues, but another text would be required to adequately address them. For the purposes of the present text, we will solely focus on the issue that two lives now are not valued equally and its implications for economics and policy, not the impact of choices around discounting.

 

3. HOW TO VALUE A LIFE   Most people’s initial reaction to the central problem of this book might be to ask, “Why are we valuing lives at all? How can something that is so morally entangled and important be reduced to a single dollar value? How could we even do it?” Things with a price are comparatively easy to value. A government might be able to purchase and deliver a single shot of a vaccine for $2. The cost for each vaccine is simply $2. However, as we saw in the last chapter, to justify that cost you need a corresponding benefit. What is the monetary benefit of the life that vaccine saves? There is a common intuition that we cannot place a value on any individual specific life. This intuitive concern that we should not value life is equivalent in some ways to the claim that assassination for hire should not be legal. Or, more broadly, that even if your enemies would receive millions of dollars in benefits from the government killing you, the government should not do it. One can only imagine the problems for the justice system and the general security of the population if the government killed anyone so long as enough people were willing to pay enough money to have them killed. The purpose of cost-benefit analysis is not to attempt to justify mob-mentality killings. However, no matter how immoral philosophers may think that economists are, when economists are talking of the value of a life, this is not what they mean. Rather, they are talking about the value of a statistical life (or VSL). A statistical life is saved when, on average, throughout a population, the risk of death is reduced sufficiently such that we can be confident that one person’s life will be saved by a particular intervention, though we do not know or decide who.[37] To make this clear, imagine a particular policy that reduces the overall chance of death by heart attack in a population of 100,000 by one percentage point. Simple multiplication finds that such a policy will result in saving 1,000 “statistical lives.” In other words, if everyone in this population has a 5% chance of death by heart attack, then approximately 5,000 people will die from this risk. If this 5% risk were reduced to 4%,

then 1,000 lives would be saved because only 4,000 people would die. The policy does not say that particular people will be saved or killed, merely that, on average, a certain number of lives are likely to be saved or lost. This is not sanctioning assassination for hire or government executions of the unpopular. Instead, it is a case of looking at the overall effects of a policy on mortality risk for a population.[38] But how do economists come up with a value for such a life? And why are these values so different from country to country? How is such a benefit to be measured in a way that can be compared to the cost of the policy? To have comparable units, a statistical life is monetized. More specifically, a certain level of risk is monetized and then aggregated up to an overall value of a statistical life. There are two common methods that economists use to calculate these values: contingent valuation and wage risk.[39],[40] Wage risk uses something called reveled preference, or how individuals actually behave in the market. It looks at behavior to determine how much individuals would need to be paid to accommodate a small increase in mortality risk. Contingent valuation, or stated preference, attempts to uncover the same information through the use of hypothetical survey questions. A full explanation of the mathematics underlying these techniques is not necessary for the present project.[41] Therefore, here are several simplified examples to motivate the necessary intuition and expose some of the weaknesses of each. Wage Risk Studies Would you do a job that might kill you? How much of a raise would you need to do such a job? One way of measuring the value of a life is by seeing how much an employer needs to pay someone to do a job that might kill them. Wage risk studies (also known as hedonic wage studies) are built on the idea that workers need to be paid more to take a riskier job. The wage premium (the difference between wages in jobs that are comparable in everything but their mortality risk) that a

The details of how economists calculate the value of a statistical life matter for understanding why some lives are worth more than others.

  worker needs to be paid to risk their life is then multiplied by the risk of death in that job to discern the value of a life in a particular market. Several jobs with varying levels of risk are combined in a regression to model what an individual must be paid to reduce a particular percentage risk of death.[42] For a simple example, imagine that there are two jobs being offered in a market that are otherwise identical, but one has a 1 in 1,000 risk of death each year, while the other is completely safe. Workers are aware of this risk, so there is an expectation that the business offering the riskier job will need to pay workers more in order for them to be willing to take the risk. Say the safer job is offering a $20,000 annual salary, and the risky job is offering a $25,000 annual salary. The difference between these two otherwise identical jobs in a perfectly competitive market would be the risk premium since all other factors have been controlled for. In other words, that $5,000 is the amount that you need to pay someone to accept an increase in their risk of death on the job from nothing to 1 in 1,000. From this point, to determine the value of a statistical life, we simply multiply the $5,000 by the 1,000 to get $5,000,000. Since people in this market must be paid $5,000 a year to reduce a 1 in 1,000 risk of death, they are valuing their lives at $5,000,000. Note that this is not claiming that if a business paid someone $5,000,000 they would be willing to die, but rather that a policy that reduced annual risk by 1 in 1,000 for a population of 1,000 people would be worth $5,000,000 in benefit to that entire population because on average it would save one life. There are a couple of assumptions inherent in this method. One is that we can find two jobs that are identical with the exception of mortality risk. This is unlikely, because a job that has a mortality risk likely also has a risk of injury. To address this, economists use multivariate regressions and hedonic pricing to control for other factors and to hold other features of a job equal. These methods are tangential to the current discussion, so I will not elaborate on them. The other assumption is that the market is functioning properly. This second assumption is more controversial, as few markets are completely free of some type of market failure. However, as the example in the next paragraph shows, disparities between the value of a statistical life of poor and rich people will arise even in a perfectly functioning market, so I do not

focus on it here. The point is that even if we had a perfect market that behaved according to economists’ assumptions, the issue of disparate valuing of lives would emerge, so the assumptions are not at fault. The key takeaway from this discussion is that in an economy with higher wages, wage premiums will be higher. Someone making $2,000,000 a year will likely require more than a $5,000 a year bump to take a 1 in 1,000 risk. Similarly, someone making only $2,000 a year, will likely be willing to take a 1 in 1,000 risk for much less than $5,000. Since higher incomes mean higher wage premiums, the lives of those with greater income are valued more. These intuitions are born out in the data.[43] In other words, if you do not control for income in your data, you will inevitably value the lives of those with more income higher. In section III we will see the philosophical case that income is ethically irrelevant to such valuations.   Contingent Valuation Contingent valuation is a process for valuing goods using survey data, also known as stated preference. Basically, this method amounts to asking people how much they would be willing to pay to reduce a particular risk, or how much they would need to be paid to incur an additional risk.[44] To be clear, no one is being asked how much they would be willing to pay not to die right now. Instead, we’re asking how much they would be willing to pay to reduce a particular risk of death. When used to determine the value of a life, individuals are asked what they would be willing to pay for a specific reduction in mortality risk. You might be asked how much you would be willing to pay in additional taxes for a program that reduces annual cancer deaths by 1 in every 1,000 people. If the average response was $7,000 a year we would consider the value of a statistical life based on these data to be $7,000,000. In other words, if we could implement a policy that costs less than $7,000 per year per person, but reduced cancer deaths by 1 in 1,000, we should do it, as the benefit would be positive: people would be getting something for less than they are willing to pay for it. Once again, it should be clear that if people have more money, their lives will be worth more. Since willingness to pay is constrained by ability to pay,[45] someone with only $5,000 in annual income could, at most, say

that they would be willing to pay $5,000 to reduce a 1 in 1,000 risk of cancer (though such a high valuation seems unlikely, as it would probably lead to a much higher risk of death from starvation, having spent all their money on taxes). Someone in a developing country making under $500 a year would be even further constrained, simply by the basic assumption of cost-benefit analysis that willingness to pay is a good measure of benefit. These studies are often claimed to be less reliable methods for determining the value of a statistical life, as their results have significant variation. There are also a range of issues around survey biases, and drawing conclusions from hypothetical instead of real choice data is not optimal.[46] However, when measuring the value of a statistical life these estimates are often consistent with the wage risk assessments. According to Boardman et al., “Estimates of the VSL [Value of a Statistical Life] based on wage-risk studies and those based on CV [Contingent Valuation] studies produce similar estimates to each another.”[47] As with wage risk studies, even if we conducted these studies perfectly and no statistical bias was present, the disparity in life valuation would arise as willingness to pay is constrained by ability to pay.   Cross-Country Valuation This provides a method for determining the value of a statistical life in countries where it is easy or common to conduct wage risk or contingent valuation studies. However, according to Viscusi and Masterman “Economists have devoted substantial attention to calculating the VSL in the United States and other developed nations, but there is a relative paucity of studies measuring the VSL in poorer nations.”[48] Viscusi and Masterman calculate the value of a statistical life income elasticity, or the change in the value of a statistical life based on a change in income through a metaanalysis of other studies of the value of a statistical life in various countries. They then use this elasticity to

Cross-country valuations of life explicitly use income to determine the value of a life, valuing the lives of the poor less.

calculate the value of a statistical life in any country with income data. Robinson, Hammitt, and O’Keefe use a similar methodology in their paper, though they offer several different potential VSLs for each country using different methods. Despite the variation in these methods, they still show that richer countries have a higher value of life by several orders of magnitude. These studies explicitly argue for a linkage between the value of a statistical life and income because that link is inherent to current methods for calculating the value of a statistical life. Since the value of a statistical life is based directly on income elasticity, it should be clear why lives in poorer countries are valued less. Where wages are lower, wage premiums are lower. Where ability to pay is lower, willingness to pay must be lower as willingness to pay is bound by ability to pay. These differences are built out of the way that economists create the value of a statistical life: they are baked in. These authors are not bigoted or biased themselves. Rather, the methodology that economics requires of them is biased against those with less wealth. With more wealth comes a greater ability to pay, and therefore greater value of life. This greater value leads to policies that save a few rich lives being chosen by cost-benefit analysis over policies that save a greater number of poor lives. As I argue in section III, this use of a morally irrelevant feature (wealth) is what results in the ethically troubling conclusion that the lives of the rich are simply worth more than the lives of the poor. However, this is not impossible to resolve. We only need to control for wealth by dividing the benefit by the wealth of the individuals and comparing instead the more accurate and less discriminatory measure of percentage willingness to pay. Now that I have presented a basic outline of the economic logic, which gives rise to this concerning conclusion, I will explain the responses that have been offered so far to deal with this issue. Three come from the field of economics; two come from philosophy. Unfortunately, none of them are sufficient to allow for unbiased, accurate, and fair policymaking.

 

           

II. INSUFFICIENT RESPONSES How philosophers and economists have tried and failed to ignore and cover up the unequal valuation of lives                



 

4. THAT’S NOT WHAT VSL MEANS   With this understanding of how statistical lives are valued and used to make decisions about policy, we return to the original problem based on these principles of economics: poor lives are worth less. If you ask an economist about this problem, their initial answer will likely be to claim that they are not actually in the business of valuing lives. Really, this is just an unfortunate feature of the way that we talk about reductions in mortality risk. “We are not saying that you can put a price on life, life is priceless! We are just putting a price on individual willingness to pay for a small reduction in mortality risk. The only ethical problem here is invented by philosophers that don’t understand basic economics.” Economists are partially correct. There are some philosophers who either do not understand the economics behind this process or find any quantification of benefits to be unethical (we will see them again in the next two chapters). However, that is not my critique. My critique is that valuing lives differently, whether you call them statistical lives or small reductions in mortality risk, forces, or at least pressures, policymakers to enact laws and projects that focus on saving the lives of the rich over the lives of the poor. Because it is immoral to codify discriminatory economic methodology that inaccurately takes wealth into account when determining benefits, the use of absolute willingness to pay to determine the value of statistical lives is itself immoral. In what follows, I outline Cameron’s argument that this is all a misunderstanding and demonstrate that, while this argument may be effective against some philosophical concerns, in the face of mine it is nothing but an attempt to cover up an immoral conclusion with archaic language. A Grand Misunderstanding Governmental disproportionate valuation of the lives of the wealthy over the lives of the poor significantly increases inequality while flying in the face of a range of deeply held notions about the ethical importance of equality. Many philanthropies make it a central tenant of their mission the claim that all lives are worth the same. As I noted in the first chapter, the Gates Foundation’s website proudly proclaims “All lives have equal

value,”[49] despite recently publishing a paper to the contrary.[50] Peter Singer famously argued that location or proximity should have no impact on the moral value that we place on a life.[51] The declaration of independence confidently asserts “We hold these truths to be self-evident that all men are created equal” despite U.S. policies that aver the opposite.[52] With a range of moral arguments at our disposal defending this claim, how can we arrive at the conclusion that, when you run the numbers, some lives are actually worth much less than others? In answer to this question, some economists have offered the following response to the initial problem. This is simply a misunderstanding between philosophers and economists. There is no real issue here because this is just a case of misleading nomenclature. People have an issue with the concept of a statistical life simply because it was poorly marketed. Economists do not really mean that there is a value that can be placed on life. By this position, when philosophers talk of the ethical considerations of the value of a life, and when economists talk of it, they are inherently talking past

 

  each other. According to this view, the only reason that the value of a statistical life is controversial is bad branding. Proponents of this position claim we should not measure these values in terms of lives but in terms of risk, something that is much easier for people to wrap their heads around because it lacks a corresponding philosophical concern. According to them, if we make our units easier to understand, the ethical dilemma will disappear.   Death by a Million Micromorts Cameron defends this view in her paper “Euthanizing the Value of a Statistical Life.” In this paper, she argues that the reason that the value of a statistical life appears to be an ethical issue is the units. According to Cameron, “Quoting the value of risk reductions in units of statistical lives is akin to quoting the price of milk by the tanker-truck load, rather than by the quart.”[53] As noted in the previous chapter, the value of a statistical life is generally measured in small increases or decreases to mortality risk, either through surveys or wage risk studies (what would someone need to pay you to take a small risk). You are not asked on a survey what you would pay to save your life but what you would pay for a small reduction in risk. You are not offered a lucrative job with a certain risk of death, rather a slightly better

paying one with a small risk of death. Therefore, we should talk about risk reduction, not statistical lives. Cameron claims that the problem is that these risk valuations are aggregated into a much larger value, the validity of which is difficult for individual citizens to judge. While you might be able to tell if the price of a quart of milk is too high, you would struggle to judge if someone selling you a truckload were giving you a good deal or not (assuming you aren’t in the milk business). If risk reduction was valued in more manageable terms, we would not be offended. Based on this, she proposes a new unit, the micromort, or the willingness to pay to reduce the chance of death by 1 in one million.[54] Based on this, if your value of a statistical life were $7,000,000, then your micromort would be $7. Therefore, if we had some policy that would

reduce mortality risk by 1 in 1,000 in a population of 200, we might say that such a policy gives everyone in that population 1,000 micromorts of benefit, or $7,000 of benefit each, for a total benefit of $1.4 million. This also makes the country difference seem much more manageable in absolute terms. An American might be willing to pay $7 for a 1 in one million risk reduction, but someone in Guinea-Bissau might only be willing to pay 10 cents. According to Cameron, if we frame our policy decisions in terms of micromorts instead of statistical lives, we can avoid the ethical implications of valuing a life and focus on valuing the risk of death. This is because the concerns really arise from the intuition that life is priceless and cannot be given a value. Therefore, if we simply provide it with a different name and unit, the troublesome intuition will dissipate. Cameron goes on to argue that this is a case for disaggregating these micromorts by demographic features, as some individuals may have different risk burdens and differential willingness to pay to avoid risk. I revisit this debate about aggregations and averages in Chapters 7 and 8. Cameron’s view that this is simply a branding problem is common among economists. Robinson, Hammitt, and O’Keefe go out of their way to make sure they are not running afoul of the grand claims about the equality of life made by their funder, The Gates Foundation. In the introduction to their paper they claim:   “VSL is not the value that the individual, the society, or the government places on averting a death with certainty. Rather, it represents the rate at which an individual views a change in money available for spending as equivalent to a small change in his or her own mortality risk.”[55]   The second of their claims is correct, however the first is misleading at best as (due to the way that cost-benefit analysis works) an individual’s willingness to pay for a reduction in mortality risk directly effects whether or not a government will choose to invest in a program that could save their life, or the life of someone richer that has a higher willingness to pay. Therefore, as I will show, regardless of how you frame the value of a statistical life, if it is created from absolute willingness to pay, you will

create a methodology that prioritizes investing in projects that save a few rich lives at the cost of many poor ones.   A Life is a Life Cameron’s argument that this is simply a case of bad branding is unconvincing because the unsettling conclusion will still arise using different units, despite a more palatable framing of the issue. While this might assuage some objections that life cannot be valued, it fails to explain how life can be valued so differently. This is not even to mention how such disparate valuations can pressure decision-makers to invest in projects that let many poor people die to save a few rich ones. Take the thought experiment from the first chapter dealing with diseases and the World Health Organization. The conclusion was not concerning because we valued life in principle but because we chose to save ten thousand Americans at the cost of letting one million Gambians die. If Cameron’s argument is sufficient to resolve all issues with the value of a statistical life, then we would expect this unsettling conclusion not to arise. However, merely changing the units will not change the outcome. We can state the original experiment in terms of micromorts and still be equally disturbed by the conclusion since this change of label does not alter the underlying calculus. Explained in terms of micromorts, we might frame the problem as the following. We can adopt one of two policies. One provides a benefit of a reduction of 30 micromorts to each individual in the United States, for a total of 10 billion micromorts, and the other provides the benefit of a reduction of 500,000 micromorts for each individual in The Gambia for a total of 1.05 trillion micromorts. Multiplying these by the value of a micromort of $9.631 for Americans and $0.079 for Gambians, we find the same net present values as before. Curing the American disease clocks in at $104 billion in benefits and the Gambian disease comes in at only $83 billion in benefits. Regardless of how we label them, we still choose to let half of the population of The Gambia, over a million people, die because we wanted to save ten thousand Americans. While changing the units may dispel notions about the inherent immorality of valuing any life, it fails to solve the underlying problem that our current methods suggest, which is that we should adopt policies that drastically increase the risk of death for poor countries

Rebranding the value of a statistical life does not change the underlying calculus that leads CBAs to recommending policies that save a few rich people while letting many poor people die.

so long as they slightly decrease the risk of death for rich countries. We should still export our toxic waste to poor countries because they are willing to pay less to save their lives regardless of whether they are paying in micromorts or statistical lives. If someone has to die from toxic waste, it should be poor people, whose deaths will cost the world the least. Rebranding only disguises the unsettling conclusion pointed out in the Summers memo; it does not resolve it. However, Cameron’s work does point us to a better solution. We need to find some way to change the way willingness to pay is measured so that it does not take wealth into account but still provides an accurate picture of how individuals would allocate scarce resources. If we could do this by simply changing the underlying calculation, without altering the broad methodology, as Cameron does, we could preserve the utility and objectivity of cost-benefit analysis, while avoiding the ethical dilemma. In the next two chapters I will argue that Cameron’s points will have power over the philosophers who argue that life is priceless and incomparable. However, I am not trying to stop the practice of valuing a life (or small reductions in mortality risk). Rather, I hope to make it more accurate and more just. In the third section of this book, I propose exactly such a solution. Instead of dividing by one million, we can divide by an individual’s wealth to calculate their percentage willingness to pay. Not only will this solve the initial problem of differential valuations of life, but it will make our methodology fairer and more accurate. Before we can get to such a solution, however, there are other positions to examine. The next two responses come from the philosophers who are of the mind that the entire project of cost-benefit analysis is deeply flawed.

 

5. LIFE IS PRICELESS   Many philosophers (and many others in public policy) think the best way to solve this problem is to simply burn the economics department to the ground. For some of these philosophers, the very project of valuing a life, or attempting to compare lives to money, is a non-starter. For these critics, the idea of cost-benefit analysis was flawed from the start. It is unsurprising that one would arrive at an immoral conclusion as the entire project was deeply immoral to begin with. In this chapter I examine and respond to the claim Cameron was concerned with: that the very project of valuing a life is immoral. In the next chapter, I look at a more nuanced objection from philosophers who take issue with our ability to compare qualitatively different kinds of benefits through monetization. There are two directions that one might come at the concern that valuing a life is inherently wrong. The first is to claim that the problem is an issue of bad measurement. Lives appear to be valued unequally because risk tolerance is something that is very hard to measure, and if we were able to measure it accurately, the issue would disappear. If we cannot measure it accurately, the only moral thing to do is stop the practice altogether. The second claim is a moral one that we should not value a life at all, ever. Basically, it states that cost-benefit analyses should get out of the business of valuing any non-market goods like lives. This position claims that it is immoral and inaccurate to value anything that can’t be sold on the market because life is priceless and benefits gained from something that is not sold can’t be captured in a monetary value. If we never value lives, we will never value some lives more than others. While each of these positions points us toward a better answer to the problem, they are each fundamentally flawed, as I will show. Bad Measurement One common critique of the value of a statistical life is that we are bad at measuring it. As I alluded to in Chapter 3, there are concerns that the methods used for valuing a life inaccurately estimate an individual’s willingness to pay for a risk reduction. There are several specific methodological issues that philosophers and

Some philosophers find the entire field of economics immoral and think the only ethical solution is to simply burn the economics department to the ground.

economists have raised to attempt to better measure the value of a statistical life. If we are inaccurately measuring how much a life is worth, it is possible that the reason for the disparities in countries is this inaccuracy. If we cannot improve these measurements, we should not value life at all. However, this would only solve the issue of disparately valued lives if a perfectly accurate measurement of willingness to pay would lead to all lives being of equal value. Defenders of this claim still support the use of absolute willingness to pay in principle; they simply claim that we are too inaccurate in our measurements to use it. Once again, this points to the correct solution, that there is something wrong with our measurement, but it does not go deep enough. As long as we continue to use absolute willingness to pay instead of some metric that is controlled for wealth, cross-country disparities will still exist. Wage risk studies assume a competitive labor market where workers have perfect information about their potential workplaces and the amount of risk present there.[56] Boardman et al. enumerate a number of issues with this method, such as the fact that people are bad at understanding very small probabilities, risk-seeking individuals naturally gravitate to these types of jobs and so may need to be paid less, and these studies are not generally representative of the full population.[57] Since we are only looking at riskseeking individuals, or individuals in certain sectors of the economy, we may drastically underestimate the willingness to pay of an average individual for a reduction in risk. To resolve this concern, Ackerman and Heinzerling argue that we should use estimates for unionized workers. Unionized workers are generally more risk averse, and therefore more similar to the general population. Additionally, unions have more resources than individual workers and so are better able to mathematically value risk. Using this method leads to a value for a statistical life as high as $42.3 million.[58] Contingent valuation also has its detractors due to a variety of surveying issues such as anchoring bias and the broad economic concern that measuring stated preference is not as justified as measuring reveled preference. Basically, people are bad at knowing what they would do in a given situation, so it is more accurate to see what they would actually do in that situation, than to ask them survey questions. Others are concerned with both of these methods because the variation of the responses is so wide.[59]

If similar methods provide you with a wide range of answers, you might be concerned that there is an issue with your underlying methodology. Broadly, there are a number of reasons to think that our methods for valuing statistical life are not actually capturing an individual’s willingness to pay for a small reduction in risk but instead are simply picking up any number of potential survey biases.   Accuracy and Equality While these are important issues, they unfortunately cannot resolve the disparity in the value of a life. Even if we were able to perfectly discern the wage premium for workers in a perfectly competitive market, incomes would continue to differ by country, and therefore the willingness to pay for a reduction in risk would necessarily vary since willingness to pay is bounded by ability to pay. As noted in Chapter 3, someone making $2,000 a year will need to be paid much less to risk their life than someone making $2,000,000 a year, regardless of the actual, underlying risk tolerance. As we can see, the value of a statistical life is actually measuring wealth, not risk preference. The same applies for contingent valuation studies. If you are asked what you would pay annually to reduce a 1 in 1,000 risk, someone with an annual income of $10,000 can say up to $10,000, while someone with an annual income of $700 can say $700 at the most. Even with the wide variety of responses, no matter what technique or response is used in a developed country, the amount for a country with a lower average income will necessarily be lower. This is not because poor people are natural risk takers and wealthy people are

naturally risk averse. It is that the way we measure risk tolerance through absolute willingness to pay is necessarily tied to wealth. Ackerman and Heinzerling are correct that there is a measurement problem at work here, but they underestimate its scope. While this argument exposes some issues with the value of a statistical life, it does not reveal what about the process is causing our final estimates to be different by country or income, though it offers clues that the issue may lie in the constraints on willingness to pay. All of our methods are limited by the fact that willingness to pay is confined by ability to pay, and people in developing countries have less ability to pay. Once we have resolved this concern, we may return to these smaller measurement issues, though as discussed in Chapter 6, the costs of dispatching with such attempts to measure will outweigh the benefits. However, until absolute willingness to pay is fixed, what economists call “risk preference” will be nothing but a proxy for wealth, and rich lives will always be worth more.   The Market for Lives The second claim is that lives are simply not the kinds of things that should be valued. If we do not value lives, then we will never have a situation where one life is valued more than another. There are two arguments for this position: we should not value lives because we should not value anything that is not traded on a market, and we should not value lives because we should only value costs, never benefits. Market goods are easy to value; we simply take the market price of the particular good. According to this claim, the reason why we struggle to correctly value a life is because it is not something that is traded on the market. Because we do not live in a world where you can pay to have someone killed or pay for immortality, we cannot put a price on life and death. Therefore, this attempt to put a value on something that cannot be bought or sold is meaningless. There are two issues with this solution. First, as Cameron accurately points out, we are not talking about lives in terms of being able to pay to have someone killed or guarantee immortality. We are talking about risk reduction. Risk reduction is a market good. You can buy a helmet or a seatbelt. You can pay for exercise equipment and products to help you quit smoking. You can buy healthy food. None of these will guarantee you will

live forever, but they will reduce your risk of death. The value of a statistical life is measuring something that we do buy and sell, not the certainty of death, but the chance of death. The second concern with this position is that, while this may resolve the immediate problem of valuing Gambian lives less than American lives, it will cripple cost-benefit analysis as a method of determining the best course of action, and particularly it will cripple interventions that attempt to save lives. If we do not value lives and other non-market goods, but value everything else, the calculated benefits of lifesaving policies will be substantially decreased, leading to fewer policies that save lives and more policies that maximize monetary benefits. Such a system would have the opposite consequence that philosophers hope that it would: it would devalue life and lead to fewer programs that save lives. Refusing to value projects that save lives do not mean we avoid the issue. It means that we think lives are worthless. It seems like saving lives is worth at least some benefit. People seem to be willing to give up their money if they believe that it will save a life somewhere. Without some value for saving a life, you could have two mutually exclusive policies, one that creates $5 of market net benefits, and one that has a market net cost of $5 but saves a statistical life. If we only look at the monetary estimates, we will always choose the policy that lets someone die. It is quite possible that such a change will simply fail to direct policy makers to enact programs with the potential of saving lives by devaluing them. While this may succeed in making American lives and Gambian lives worth the same, it only does so by making them worthless, biasing policy toward economic benefits over human ones.   The End of Benefits Another reason one might argue for not valuing life is to claim that we should not value any benefits in a cost-benefit analysis. This argument is made by Sagoff in his book Price, Principle and the Environment. He claims that economists should just stick to the costs of a program and let the decision makers address the benefits separately.[60] This is motivated by the underlying idea that willingness to pay is a concept that claims to measure benefits or welfare, but only does so by fiat. As Sagoff claims, “The Thesis that goods should be allocated to those willing to pay the most for them

because this maximizes welfare—when WTP [willingness to pay] is the measure of welfare—is then tautologically true. This principle draws as perfect a circle on earth as can be found in Heaven.”[61] In other words, according to Sagoff, there is no argument offered that willingness to pay is a good measure of benefit. Instead, it is simply assumed by definition. Since the benefit of getting something for less than you are willing to pay for it is in doubt, all of the benefits are in doubt. Therefore, Sagoff argues, we should not measure benefits of a policy, only costs. Then, have a politician or voters decide if the benefits are worth the costs. For Sagoff, the goals of a society are clear, so we should focus on achieving those goals in the most cost-effective manner, not using cost-benefit analysis to question the goals.[62] As costs are generally the kinds of things that are sold on a market, they do not require the use of willingness to pay just basic prices. This avoids using willingness to pay altogether and instead allows politicians to weigh monetary costs, with non-monetary benefits. As with many of these solutions, Sagoff is partially correct. The problem is using willingness to pay as a measure of utility. However, as before, his solution fails to fully exterminate willingness to pay from the process of cost-benefit analysis. This is because willingness to pay must be used to measure costs as well as benefits. Lives can be a cost of a program as well. If I have a potential policy that will increase risk as a cost, we will still need to value a statistical life.[63] We will need to compare that to Sagoff’s other costs of preserving other species, or preserving the environment. Additionally, there are all kinds of costs that people can bear that are non-monetary. Some people may oppose a policy because they do not like it, or are morally offended by it. Even within the environmental sphere, some policies can be opposed to each other: species have been known to eat other species on occasion. These costs must be captured in terms of willingness to pay. I agree that willingness to pay is the problem, but Sagoff’s solution has failed to eliminate it from cost-benefit analysis; it has only hidden it. This also raises additional problems

by preventing expert testimony from impacting decision making. I will cover this concern in the next chapter. Once again, though insufficient, this solution points toward the real answer, replacing willingness to pay as a measure of utility. The claim that

that allocating goods on the basis of absolute willingness to pay will maximize welfare is only convincing if you already believe that absolute willingness to pay is a good measure of welfare. By making the minor methodological shift to percentage willingness to pay, we have a measurement that more accurately tracks welfare—and succeeds in valuing lives equally.

           

6. NOTHING COMPARES TO LIFE     Some authors take issue with the whole process of cost-benefit analysis, not merely the fact that it values life. For these authors, the entire project of a systematic enumeration of the benefits and costs of a policy is deeply flawed. These authors often come from other ethical traditions of philosophy, which question the consequentialist underpinnings of costbenefit analysis. While the authors in the previous chapter did not go far enough in their efforts to root out factors that cause the system to value poor lives less, the authors discussed in this chapter risk biased, uninformed decision-making by completely discarding the useful tool of cost-benefit analysis, throwing the proverbial baby out with the bathwater. There are two arguments for disposing of cost-benefit analysis in its entirety. The first can be found in Ackerman and Heinzerling, and focuses on the problems of the Kaldor-Hicks criterion. The second is offered by Elizabeth Anderson and is based on the work of Joseph Raz, arguing for a form of value pluralism, which claims that some things simply cannot be compared. These positions do offer replacements for cost-benefit analysis. However, both of these replacement decision frameworks are at least as biased as our current cost-benefit analysis, if not more. In this chapter, I make the case that these responses are not only insufficient on their face, but may in fact worsen the existing bias against poor lives. These authors are too quick to throw out cost-benefit analysis without realizing the harm and corruption that can come from public policy untethered to data or a rational decision framework grounded in evidence.   Do No Harm The first concern comes in the form of a critique of decision criteria. As discussed in Chapter 2, cost-benefit analysis uses the Kaldor-Hicks criterion to decide between potential alternatives. This prioritizes policies that maximize benefits for the whole society, even if some individuals experience net costs. This is contrasted with the Pareto Efficiency criterion, which requires that policies do no net harm to any individual: that no one is worse off than the status quo. In their book Priceless, Ackerman and

Heinzerling raise concerns about both of these decision criteria, and make the case that some criteria that is more permissive than Pareto, but less permissive than Kaldor-Hicks is required. They argue that Pareto Efficiency is intuitively convincing, but it is too restrictive, as most policies will incur some cost.[64] However, they find the Kaldor-Hicks Criterion to be too permissive, as it does not require the “winners” who benefit under a particular policy to compensate the “losers.” Basically, polices that increase some welfare without any losers are rare and, though justified, often maintain the status quo, while policies that allow some to gain at the expense of others are not sufficiently concerned with the distribution of the benefits. I offer two responses to this concern. One, explained here, is that such an intermediate criterion would both completely hobble cost-benefit analysis and fail to solve the initial problem of the disparate valuation of lives. The second, which I explain in Chapter 16 (responses 2.3 and 3.1), argues that the solution I offer in this text actually succeeds at fixing the concern about the winners not compensating the losers by ensuring that losers of the last policy are more likely to be compensated in the next one. Either is sufficient to defeat the concern offered here. At its heart, this is really an objection to the consequentialist framework underpinning the cost-benefit analysis methodology, as it questions whether we should maximize societal good if it means that some people will lose out. Without the Kaldor-Hicks criterion we could only compare policies that benefitted individuals at the expense of no one. This would severely cripple cost-benefit analysis as a tool because the vast majority of policies incur some net cost to someone. While it might be interesting to attempt to defend consequentialism in the face of this critique, it is tangential to the point at issue. The problem of disparately valued lives will emerge even in scenarios that pass the Pareto Efficiency criterion. Therefore, any criterion that is more permissive than the Pareto Efficiency criterion (as Ackerman and Heinzerling advocate) but retains the other components of cost-benefit analysis will have such issues. To see this, remember the example from Chapter 1 about the chief economist of the WHO. This example presented us with a choice between two policies: one that saved the lives of Americans, and one that saved the lives of Gambians. Neither of the policies included a cost (we could

imagine that all the costs are “sunk” costs, we are going to pay the scientists regardless, the only question is which benefit we should pursue). The status quo in this situation is that all of the affected people die.[65] Therefore, as originally stated, either policy is a Pareto improvement; we do not need to take anything from anyone’s initial allocation to implement them. Either saving the Americans or saving the Gambians is better than doing nothing. This means that even using the less permissive framework, which Ackerman and Heinzerling admit is justified, we are required to reduce American risk of death by 0.003% (and save 10,000 statistical lives) instead of reducing Gambian risk of death by 50% (which would have saved one million statistical lives). This means that even if we accept their critique and get rid of the Kaldor-Hicks criterion, we will still value rich lives more. There might be a broader issue with cost-benefit analysis and consequentialism around whether the Kaldor-Hicks criterion is justified, but it is not this issue that gives rise to the concerns of this book. The differential valuation of cross-country lives issue arises even when there are no costs and only benefits. The solution that I offer in section III has the advantage of not only resolving the concern of differential cross-country valuing of lives but addressing the concerns with the Kaldor-Hicks criterion, because percentage willingness to pay assures that those who bear a cost do so relative to their means, and successive iterations of

cost-benefit analyses will not continue to concentrate gains among the wealthy. The common justification of Kaldor-Hicks from economists is that, eventually, the government will enact policies that benefit everyone, so the costs that people bear for one policy will be balanced out by another policy that provides more benefits for those individuals. However, the criticism is that policies often repeatedly benefit the wealthy at the expense of the poor. If this bias does exist, then it may be the case that benefits always accrue to one group and costs to another. This is exacerbated by using absolute willingness to pay that values benefits to the rich more than benefits to the poor. Percentage willingness to pay partially alleviates this concern, as the more benefits someone has received in the past, the less new benefits count in the overall calculation. This forces governments to actually implement

policies that will balance out the net costs borne by the poor instead of simply implementing policies that only provide benefits to one group and costs to another.   The Problem with Monetization Another central tenant of the cost-benefit analysis is that all of the costs and benefits can be put into the same units, in this case monetary units. This is justified by the economic claim that preferences are complete and transitive. In other words, all states of affairs are either better, worse, or the same as other states of affairs, and these states of affairs can be ordered or ranked. Basically, for any decision, a government can list all of the possible futures that a decision might create and rank them against each other (ties are allowed). Under such a model, you cannot have a situation where policy A and B are equal to each other, but A is better than C, and B is worse than C. However, some have argued that certain goods are incommensurable, or they cannot be compared. Basically, that you can have two states of affairs where one outcome is not better, worse, or equal to another, which means we cannot rank all potential polices against each other. If this were true, it would mean that we could not monetize and compare these states of affairs because real numbers are always either greater than, less than, or equal to other real numbers. If this is the case, it would solve the issue of valuing lives, as we could simply count lives saved and dollars spent as separate quantities. However, it would mean the death of cost-benefit analysis as we know it. In her book Values in Ethics and Economics, Anderson argues for a kind of value pluralism where some values cannot be compared to others.[66] If there are some goods that cannot be compared, then the project of attempting to monetize all of the benefits and costs for comparison is flawed from the start. If you accept this position, you may have a way out of the original problematic conclusion. If we cannot compare the value of a life and other benefits, then the problem of disparately valued lives comes from trying to convert a life into something with which it cannot be compared: money. However, this does raise the issue of what tool could be used to compare the impacts of policies if cost-benefit analysis were abandoned (as it would need to be without a common measure to compare different

benefits). Anderson argues for a more localized democratic process where, instead of consulting experts on the economic implications of a policy, individuals impacted by that policy are consulted and they are given the ability to determine what the policy should be. According to Anderson, “This can be done by devolving power from distant regulatory bureaucracies to local, self-managed institutions or citizen action groups.”[67] Basically, instead of conducting a distant, academic, cost-benefit analysis we should simply ask the people on the ground what they think a policy should be. Sagoff similarly argues for “democratic deliberation” to replace cost-benefit analysis, along with his claims of just measuring costs. [68]

This appears to be an argument that could both resolve the original problem and provide a new public decision framework. However, I am concerned both with the claim that goods are incommensurable, and the new decision framework. In the remainder of this chapter, I first look at the argument supporting the claim that goods are incommensurable, and offer objections, then I address the claim that we should move to a more localized democratic process and show that this would increase existing biases and make the process less transparent and rife for corruption.   Incommensurability Anderson bases her claim that values are incommensurable on the work of Raz. In his book The Morality of Freedom, Raz defends the claim that there exist situations where particular states of affairs are incommensurate. In other words, our preferences about the states of affairs are not complete (one state of affairs may be neither equal, better than or worse than another). He defends this claim by arguing that there are some situations in which preferences are not transitive (if a is better than b and b is better than c, it is not necessarily the case that a is better than c), and that, in such cases, the preferences about these states of affairs must be incommensurate. If preferences are neither transitive nor complete, they cannot be enumerated on a single scale. Therefore, they cannot be monetized. Raz defines incommensurability as “A and B are incommensurate if it is neither true that one is better than the other nor true that they are of equal value.”[69] In other words, according to Raz, preferences are not complete: we cannot rank any one situation as either better, worse or equal to any

other. When comparing two states of affairs there is some fourth relation called incommensurate, where one is neither better, worse, or of equal value to the other. Since real numbers are always greater than, less than, or equal to other real numbers, states of affairs cannot be turned into real numbers regardless of if we are using monetary values, percentage of willingness to pay, or any other numeric measure of utility. Raz defends this claim by showing that preferences are not transitive: if we prefer a to b and b to c, it is not necessarily the case that we will prefer a to c. According to Raz, the failure of the principle of transitivity is a sign of incommensurability.[70] Preferences of this type would pose problems for the whole of economics, cost-benefit analysis included, as one of the fundamental assumptions of economics is that preferences are transitive: if you prefer x to y and y to z, then you prefer x to z.[71] To motivate this intuition, Raz gives examples of desires which appear to violate this transitivity principle. Imagine that Sarjo prefers eating a hamburger to eating a hotdog but is indifferent between either of these options and going for a jog. You might imagine Sarjo saying, “If we are going to get food, I would prefer a hamburger, but I don’t care if we eat or go for a jog.” In this case, Sarjo’s preferences are incommensurate, as they fail the law of transitivity. If Sarjo values a hamburger equally to a jog and a hotdog equally to a jog, then Sarjo should value a hamburger equally to a hotdog. Since he does not, his preferences are not transitive. The only way to describe his preferences is that they are incommensurate because they cannot be mapped onto numeric values. Some economists might claim that such preferences are irrational, but Raz argues that there is no underlying objective scale of value that preferences are mapping onto. Therefore, these preferences are not irrational but a demonstration that preferences are not transitive. Basically, claiming that any intransitive preferences are irrational is simply to beg the question against the proponent of incommensurability. According to Raz, “The mistake in this thought is that it assumes that there is a true value behind the ranking of options, and that the ranking is a kind of technique for measuring this value.”[72] Therefore, incommensurability is not irrational because preferences do not measure some deep underlying scale of value (because no such scale exists) and are therefore not transitive. Basically, if

Raz is correct, Sarjo would not be irrational, as the only reason you would think Sarjo is irrational is that you are already assuming that Raz is wrong.   Incommensurate vs. Indifferent If this were the case it would resolve our original dilemma. Lives would not be valued at all because we could consider reduction in mortality risk to be incommensurate with monetary gain. Is it better to have seven million dollars of value spread among the residents a small community of 1,000, or is it better for the risk of mortality in that community to decrease by 1 in 1,000? If some situations are incommensurate, examples like this seem like a strong contender. It could be argued that neither of these situations is better than the other, but they are not equal because lives cannot be monetized. If lives cannot be monetized, or compared to monetary gains, there is no need to value a statistical life in the first place. However, I am skeptical of Raz’s arguments for a range of reasons. First, it seems to me that while individuals may have preferences that violate the rules of transitivity, there is no reason in group decision-making to treat them as different from someone expressing the preference that all the options are equal. Second, while preferences may serve as a good proxy for utility, if our real concern is maximizing utility, Raz needs to show that states of utility are incommensurate, not preferences.   To illustrate the first point, take John. He prefers the policy that abortion is legal to the policy that abortion is illegal. However, he also prefers the policy that abortion is illegal to the policy that abortion is legal. Additionally, he is indifferent between abortion being legal and abortion being legal. His beliefs are incommensurate, as they violate the law of transitivity (where L is abortion is legal, I is abortion is illegal and > is the preference relation: L>I and I>L but it is not the case that L>L, as would be required by transitivity) and therefore the law of completeness (his beliefs are all three of greater than, less than and equal to each other). The problem for Anderson’s interpretation of Raz is that even if we grant that such a strange individual exists, and that they are rational, there is no reason to think that such preferences should inform public policy, as they do not show what policy a government could enact to make such an individual happy.

John’s beliefs should not inform public policy since it is unclear what he wants. The point here is that, if someone does not express ranked preferences, then they should be treated as if they do not have a preference. Imagine that you are in one of Anderson’s idealized citizen action groups and one person consistently demands both that we raise and lower the sales tax on a particular good. Since such an option is impossible, it seems to me that you may consider that person’s views to not be importantly different than being ambivalent between raising and lower the sales tax. The point here is that the separation between equally preferring two things and those things being incommensurate with each other is a distinction without a practical difference for deciding how a society should act. This concedes the possibility that someone could have such preferences but not that such preferences should inform policy since they are inconclusive of a single option. This seems to justify the idea that it is possible treat individuals with incommensurate views on policies as if they viewed the policies equally in some cases. But what about a case similar to the one that Raz offers? Imagine Mary. She prefers a Hard Brexit to a Soft Brexit, but is ambivalent about whether the United Kingdom leaves the European Union or not. Basically, she thinks “if we are going to leave, we should do it completely, but I don’t care if we leave or not.” Mary’s view seems importantly different from John’s in that she seems to be expressing some view on the subject, that we might be able to act on in some situations. However, her views also seem incommensurate, as they are not transitive. Where HB is Hard Brexit, SB is Soft Brexit, and EU is remaining in the European Union, she seems to have the preference that HB=EU and SB=EU, and therefore she should have the preference that HB=SB, but in fact she has the preference HB>SB. In cases like Mary’s it seems that there are two possibilities. Either she really has an opinion about the policies, which is getting masked by framing the preferences in terms of binary comparisons, or she is really expressing incommensurate preferences between them, and her views should be given as much credence as John’s. First, to determine if her true opinion is masked by the framing of the question, we need to simply give her all three options and ask her to rank them (allowing ties). If she actually ranks the Hard Brexit first, then No Brexit is tied with a Soft Brexit, that means she

actually was not indifferent between No Brexit and a Hard Brexit. If she ranks Hard Brexit and No Brexit as a tie and Soft Brexit below them, then she really was not expressing a measure of indifference between a Soft Brexit and No Brexit. In any of these situations, her preferences are not incommensurate, and we can monetize or make policy decisions based on them. If she ranked all of the options the same, then we can view her as having no preference. If she claimed that such a system could not allow her to express her views, she might actually have incommensurate preferences. At this point we could simply ask her to pick the policy she would enact if she could. If she picks any policy over the other two, her views are not really incommensurate. If between all three options Mary would choose a Hard Brexit, she is not really ambivalent between that and No Brexit. If she would choose No Brexit, then she is really not ambivalent between that and either of the other options. If chooses a Soft Brexit, then she really does not prefer a Hard Brexit. If she claimed that she could not chose a single policy to enact, then her views are incommensurate, but they are also not importantly different than John’s. We cannot make policy based on indecision. The point is that, from the perspective of making decisions on policy, there is no important difference between someone who is incommensurate between two options and someone who is ambivalent between those two options. This is not to say that we should discount the views of anyone who has any inconsistent preferences but rather that if your preferences are incoherent, there is no way to use them in a decision-making process no matter how democratic or locally sourced it is. While Raz may be correct that some people have strange incommensurate preferences, Anderson’s claim that we should treat those preferences any differently than indifference is unconvincing.   Do You Know What Makes You Happy? My second concern with Raz is that preferences are not what we really care about. While a consequentialist who cares about fulfilling desires might be worried if there are not objective rankings of desires, that is not what the proponent of the cost-benefit analysis cares about. Arguably the goal of cost-benefit analysis is to maximize utility as measured by willingness to

pay. While people may have inconsistent desires, it seems that utility can be measured on a single scale and follows the law of transitivity. Regardless of your feelings about hotdogs, hamburgers, and jogs, there is some fact of the matter about how much utility or happiness you get from each choice. People may be wrong about how much utility a particular activity gives them, but that does not mean that there is no objective measure of it. There is a fact of the matter about the amount of utility that a particular choice brings into the world. Utility is hard to measure, so we use things like willingness to pay to approximate it, but just because someone has inconsistent preferences, or does not know how much utility a give choice would bring them, it does not mean that utility is not an absolute scale. A situation which might be able to separate these two situations is one involving misinformation. Chapter 15 will look at misinformation specific to covid-19 and how that can impact choices, but for now we will focus on a more general example. Imagine a child who has never tried broccoli but has the belief that they hate it. Unfortunately, this belief is incorrect. If they tried broccoli, they would absolutely love it, but their preference, lacking that information, is to avoid it. The question remains: Should a parent who knows the child will love broccoli encourage their child to eat it? Which world has more happiness in it, the one where the child’s preference is achieved, or the one that actually makes the child happy? For an example with more weight, should a government promote exercise if it would create more happiness, even if the population does not want to exercise? Is a world where people exercise better than the one where they remain in ignorance? While there are interesting arguments to be had around which persuasive methods governments can use (mandates, nudges, public information, etc.), the question here is not around the tools but the goals. To avoid these concerns, imagine we are using the most innocuous public information that simply promotes a beneficial but unliked activity (without in any way requiring it). In other words, a government is subtly promoting something that will increase happiness, but that people do not generally choose to do. It seems that, at the very least, our assessment of the possible worlds should take into account an individual’s actual happiness, not the inaccurate desire fulfillment they think they will receive. Put another way, governments shouldn’t promote scam products that seem desirable but are

actually harmful and make you unhappy. Rather, they should provide consumers with complete information so that they can better understand what will make them happy and evaluate happiness based on actual results instead of wishful thinking.   Will Local Democracy Save Us? Even if you still hold that incommensurate preferences give us a reason to avoid cost-benefit analysis altogether, the question remains whether there is a viable alternative to analyze and adjudicate the impacts of a policy. Both Anderson and Sagoff propose eliminating a cost-benefit analysis that cannot account for incommensurate preferences and replacing it with a more democratic process. Their central concerns have to do with the aggregation of preferences since they are value pluralists who think preferences are incommensurate and cannot be aggregated. Anderson views economists being replaced by everyday people who best know their own minds, arguing for “Institutions of voice [that] allow people to articulate their concerns directly and thereby empower them to put new items on the agenda without depending on ‘experts’ to speak for them.”[73] As mentioned in the previous chapter, Sagoff, argues for a slightly larger role for economists allowing them to conduct analyses of costs, but precluding them from researching the benefits of these policies, with instead a preference for deliberative democracy.[74] The argument for these positions basically boils down to saying individuals have more than one thing that they may value or benefit. These benefits cannot be aggregated together, so people affected by these policies, not technocrats, must make determinations. A single algorithm that determines whether or not to enact a project based on a single value is insufficient. Therefore, we should eliminate a process that attempts to aggregate these choices into a single number or net present value. Instead, we should use local communities of practice and focus groups of those impacted by the policies to make decisions about which policies are worth the cost. Such a solution would solve the initial problem of disparately valuing lives on the surface, but I am concerned that this would just mask the issue and make things worse for lack of transparency.   Exile the Experts

I have three objections to the positions offered by Anderson and Sagoff that we should limit or exclude cost-benefit analysis from the decision-making process. The first is the concern that evidence-based policy making using cost-benefit analysis is more objective, effective, and grounded in truth than the alternative proposed by Anderson and Sagoff. The second is that the only reason that we are able to scrutinize cost-benefit analysis at all is that it is transparent and can be improved upon. Any process happening behind closed doors with only the people directly impacted by a policy is not transparent, and even if a problem of corruption or discrimination is exposed, we cannot fix the system to prevent future abuses in the way that we can fix cost-benefit analysis. This solution will not actually solve anything; it will merely mask it. Governments will still make decisions to prioritize rich lives over poor ones, but without a public, objective, system like cost-benefit analysis, no one will know why or how certain decisions were made. My final concern is that we may really have incommensurate preferences, but at the end of the day the government must make choices. These choices will put an implicit value on life, even if we do not state it explicitly. We do not really have a choice of whether or not to rank options against each other. We must rank them because we must act. Our only choice is whether to make those choices in the light of day or the darkness of a back room. First, not including a cost-benefit analysis in the discussion of a policy is akin to kicking climate scientists off of a government panel discussing policy options to deal with climate change. Both of these actions are supposedly justified by the belief that citizens know best about their own needs and situations, while scientists are “experts” who have too much power and don’t really know what people want or need. People experience the climate and the economy every day; they don’t need an expert to tell them what they should believe or prefer. People may have different values when it comes to the economy, and they may have different values when it comes to the climate. We do not need a rigorous scientific analysis of the benefits of cutting back on carbon or any economic policy because citizen intuition is the most important. The people most directly affected by these policies are the companies that profit from pollution, so they are the ones that should be in the room.

Hopefully it is clear why this is problematic. Yes, individuals experience weather every day, but they do not experience the entire climate. Their individual perspectives are biased by anecdotal evidence. An individual might conclude that because it was cold this year where they lived, there is no evidence for global warming. Yes, companies that pollute are the ones most directly affected by climate change legislation. But they also have a deeply vested interest in seeing a certain outcome that may not be beneficial for the majority who are indirectly affected. A scientist, on the other hand, is able to view the total evidence, take random samples, and draw conclusions about the entire population. Excluding climate scientists from the discussion of climate change leads to special interests trumping facts, and fallacies of generalization repressing scientific evidence. Similarly, while individuals can express their own beliefs about the economy, such as disgust about the foreigners taking all of the good jobs, or the firm belief that corporations should all be run by the state, economists can present unbiased data about how the labor market is actually changing, or how the private sector has been shown to be able to provide certain goods much more efficiently than the public sector. Cost-benefit analysis is one of the tools used by economists to make such determinations. If economists are not allowed to value benefits like the actual economic impact of immigration, or the value gained or lost through privatization, the only thing that we have



 

to go on are the often-biased views of individuals, which are frequently factually incorrect. Sagoff’s solution of limiting economists to only speaking about costs and not benefits may not be analogous to kicking science out of the room, but it arbitrarily silences an important part of the debate. It would be akin to allowing climate scientists in the discussion of environmental policies, but preventing them from speaking about temperature change. Only looking at the impact of climate change on rainfall but not temperature paints an inaccurate and incomplete picture. This would lead to a worse conclusion than simply silence from the experts, it would lead to making decisions based on misleading and incomplete information. Sagoff’s solution handicaps the experts to the point where their testimony may in fact be harmful. To put this objection simply, in a large country it is impossible for everyone to express their opinions on a particular policy. Therefore, we need a method to find out how a particular policy would affect individuals. We can either have economists randomly sample the population, gather data, and report aggregated information about the costs and benefits, and who bears them representing the entire population, or we can take public comment from the individuals with the privilege, time, and money to come and make a comment. Which of these will be more representative of the population? It should be clear that random sampling will necessarily represent a wider swath of the population, including those who do not have the time or money to make a public comment. This is not to say that public comment is a bad thing. Qualitative data may help tell stories that cannot be seen in aggregated numbers. But that does not mean that aggregated numbers do not represent a broader, more impartial story about the entire population. If you are trying to predict the weather, it is better to talk to a meteorologist with the full satellite picture who is able to see storms coming and going than only the people in a single neighborhood who can just see their small piece of sky. This is particularly important for the question at hand because the population under the purview of multilateral organizations is the whole world. It is difficult to even get a single representative from every country in a room together to discuss the impacts of a policy on their populations.

Due to limitations on visa processes, and the fact that such a meeting takes time and money they will necessarily be a privileged group. These individuals are not representative of the whole world, they are representative of the richest and most powerful in the world. This method of decision-making will exacerbate existing inequalities, not resolve them. The point of randomly surveying populations is to deal with exactly this issue. As much as Anderson may want these decisions made at a local level, that does not solve large collective action problems where everyone must cooperate to succeed, such as climate change. Nor will such a procedure resolve issues requiring global public goods such as cures for diseases, or vaccinations. It is better to gather representative data and present it in an unbiased way, than to attempt to get every single town in the world to have a local discussion and come to a conclusion about their carbon emissions, for example. Exiling the experts gives the powerful carte blanche to act as they wish, often at the expense of the poor, instead of making policies based on representative facts.   Sunlight as a Disinfectant My second concern is that the public will be unable to scrutinize such a nonspecific and opaque process to prevent corruption and discrimination. Small groups of citizens may come up with novel solutions, but they may also fail to recognize or appreciate all of the eventual effects of a policy change (particularly how that policy effects those not privileged enough to have a seat at the table). On the other hand, a systematic method for looking into all of the positive and negative effects of a particular policy change (such as cost-benefit analysis) allows the public to have confidence that everyone’s interests were taken into account instead of just the people who can afford to be in the room. If there is a problem with that process, such as the use of absolute willingness to pay to undervalue the lives of the poor, we are able to see and critique it. No such means of critique is available for the behind closed doors local democracy promoted by Anderson. This transparent system, where clear reports and values are published, is what makes it possible for the likes of Anderson, Sagoff, and myself to critique that process. If, however, the process is conducted with a small group of privileged participants in an unsystematic way, it would be easy for them to ignore underprivileged groups that might not be present in the

room. And those choices will never be subject to critique in the light of day since they are not made through a systematic process but instead through an unsystematic discussion. Institutionalization of the process makes us able to improve it. If the people in that room chose to devalue the lives of Africans in their internal, unseen calculus, we would not know. We would not be able to scrutinize and make recommendations. And even if we found out, the process could not be blamed, only the individuals. This prevents improvements to the system. People have many unconscious biases that affect their decision-making processes. Without a systematic and unbiased process like a cost-benefit analysis, it is likely that these biases will be used to make decisions but never brought to the light. Systematic processes may have biases as well, but so long as they are transparent, those biases can be found and eliminated. That is the goal of this book, to eliminate one such bias. There is an inherent bias against the poor in using an absolute instead of a percentage willingness to pay. We are aware of this because cost-benefit analysis is transparent. We can resolve this bias by a slight alternation to our methodology for calculating willingness to pay. There is no need to destroy the system to fix a small problem. Anderson and Sagoff’s solution may appear to work while aggravating the underlying concerns and simply improving the quality of the cover-up. Anderson and Sagoff’s solution will not resolve the issue. It will exacerbate it by allowing personal biases of the powerful to be the determining factors in policy making and then hide the process away, preventing any further scrutiny. Poor lives would still be valued less, but only the rich would be aware of it. Such a system is not merely discriminatory and corrupt. It makes progress toward improving the system impossible. I would rather have an unequal system that I can critique and try to improve to the point that it may be more unequal one day than an unequal system that is so opaque as to prevent any criticism.   The Dangers of Implicit Values of Life My final concern is that even if the benefits in one state are incommensurate with the benefits in another, a government must act, and by acting place an implicit value on life. This implicit valuation is dangerous, as it is not subject to the same public scrutiny as an explicit valuation. In the

World Health Organization example from Chapter 1, you must choose one disease to cure. Regardless of how you make that choice, you place an implicit value on life. If you choose neither and sit in indecision because you cannot put a price on a life, you will do a far greater harm. A public institution cannot hold two inconsistent policies as incommensurate or even equal. It must choose between them. Therefore, eventually, even if you have a group of people with different values present, the government will make a choice; they will rank the options. They will make a value judgement between the incommensurable values. Even if the government is stymied by indecision and chooses to cure neither disease, they are still placing a value on life (just a very low one). A room full of people who are incommensurate between the options must act and therefore place a value on life, even if this is not done explicitly. The question is not whether a government will act but how we choose what the government should do. To understand this, imagine a committee in the city of Speedy Hills, population 1,000,000. This committee is deciding on how high to raise the speed limits for residential roads. Leaving the speed limit at its current value of one mph would lead to a 0.0001% mortality risk annually (or one statistical life). Setting it at 10 mph would lead to a 0.001% mortality risk annually (or 10 statistical lives). Setting it at 20 would lead to a 0.002% mortality risk annually (or 20 statistical lives). But setting it at 30 would lead to a 0.01% mortality risk annually (or 100 statistical lives). In the population, if they set aside all safety concerns, people prefer to have a higher speed limit. They get no additional benefit from the status quo with the speed limit at one mph. They each are willing to pay $40 annually for a 10-mph speed limit (for a total benefit of $40,000,000). For a speed limit of 20 mph they are each willing to pay $150 annually (for a total benefit of $150,000,000). For a speed limit of 30 mph they are each willing to pay $200 annually (for a total benefit of $200,000,000 annually). These preferences represent how people feel if they ignore the safety implications of the speed limits, and only think about their other benefits. These four options (1, 10, 20 and 30) are the only options under consideration. The committee that makes such a decision may claim that they are not putting a value on a life, but no matter what they choose, they are doing so implicitly. If they retain the current speed limit, they are implicitly setting

the value of a statistical life at more than $7,500,000, if they valued a life less, they would set the speed limit higher. For example, if they valued a life at $7,499,999, they would set the speed limit at 20 mph because this would be the only option where the benefits ($150,000,000) would outweigh the costs ($149,999,980). This means that if the committee does not raise the speed limit, the benefits per life must be greater than the maximum benefit per life of the alternative options (20 mph at $7,500,000 per life) If they set the speed limit at 20 mph, then they are implicitly setting the value of a life at between $7,500,000 and $625,000. Their implicit valuation of a life cannot be above $7,500,000 because then the costs would outweigh the benefits (20 statistical lives at $7,500,000 each equals the total benefit of $150,000,000). The committee’s implicit valuation of life could not be below $625,000 because that is the point where benefits from setting the speed limit at 30 mph surpass the benefits from setting it at 20 mph. At 30 mph, $200,000,000 in benefit and 100 statistical lives at $625,000 ($62,500,000) of cost leaves a net benefit of $137,500,000. At 20 mph, $150,000,000 in benefit and 20 statistical lives at $625,000 ($12,500,000) of cost leaves a net benefit of $137,500,000. It does not seem rational to set the speed limit at 10 mph, as more benefit can always be gained at another level. But if they did, perhaps only comparing this option to the status quo, then they are implicitly setting the value of a life at less than $4,000,000 (10 statistical lives at $4,000,000 each equals the total benefit of $40,000,000). If the value of a life was greater than $4,000,000, the costs of the policy would outweigh the benefits. In other words, comparing just this option and the option of retaining the current speed limit, if the committee retained the speed limit at one mph, it would mean that the costs of raising the limit (10 deaths) outweigh the benefits ($40,000,000), meaning that each of those lives are valued more than $4,000,000. Regardless of what they decide, they are implicitly putting a value on a life. However, without a rigorous cost-benefit analysis, the public will never know what value was being placed on a life, or whether that value was greater for some communities and less for others.

In fact, if the people in the meeting are not explicit about the value they place on life, it is likely that their implicit biases will lead them to set a lower value on life for a community that they do not like, or which is different from their own community without even realizing it. The decisionmakers may claim their preferences are incommensurate between lives and money, but any action that they take creates an implicit ranking of the cost of mortality reduction against the monetary benefit of faster speed limits. The process proposed by Sagoff and Anderson, that decisions must be driven by those privileged enough to spend time working in local government is rife with the potential for corruption, is not based in sound science and will lead to the valuation of lives, just an unpublished implicit one. Cost-benefit analysis, conversely, is representative of the world, is a regimented process which clearly documents its biases, and is based in science instead of intuition. The job of a government is to make decisions. Even if people have confusing, inconsistent, or incommensurate preferences, decisions of public policy must be made. Even if the process is flawed, I would contend that it is better to have a transparent, scientific, representative process than one which is none of those things. Therefore, because alternative processes result in worse outcomes, we are stuck with something like the cost-benefit analysis. Governments will always make decisions that put an implicit value on a statistical life, whether or not it is calculated. This discussion has shown that it is not costbenefit analysis itself that is at fault but the methods used to calculate benefits. In the third section of this book, I argue that we can resolve the problems of cost-benefit analysis without resorting to opaque and biased methods of policy analysis. Anderson and Sagoff are correct that what needs to change is our methodology. However, they go too far by advocating doing away with cost-benefit analysis in its entirety when all that is required is the small change of moving from absolute willingness to pay to percentage willingness to pay. This minor methodological change retains all of the transparency of cost-benefit analysis while reducing the unjustified bias against poor people inherent in absolute willingness to pay.

 

7. THE AVERAGE COVER-UP   Imagine you are the leader of a SWAT team attempting to stop three empty proverbial runaway trollies on two tracks: two trollies are on a track on the left, 20 minutes apart, and one is on a track on the right. There is one person stuck on each track. You don’t have time to rescue the people or stop the trollies, but you do have time and enough materials to blow up exactly one trolley. If you blow up either left trolley, the other two trollies will be unaffected and both people will die (the trolley behind will just plow through the wreckage), but if you blow up the right trolley, only the person on the left track will die. It seems intuitive in such a situation to blow up the right trolley since this will save a life. Blowing up the leading left-hand trolley will give the person there 20 more minutes of life, but this does not seem important. It seems that the length of the life you

 

Saving the life of someone with 20 minutes to live is more beneficial than saving the life of someone with 50 years to live. If you can only blow up one trolley, blow up the right-hand one.

are saving matters. Saving someone who only has 20 more minutes to live is not the same as saving someone who will live for decades. How much life you are saving seems to be an ethically salient feature that you can legitimately consider when choosing who to save. This intuition is exactly the intuition that the proponents of averages discussed in this chapter wish to deny. The average advocates claim that blowing up the leading left-hand trolley and letting both people die is equivalent to saving one by blowing up the right-hand trolley. This chapter will also address the uncomfortable conclusion that while the current disparities in the worth of life may be unjustified, much smaller differences based on ethically salient features may be.   Means aren’t Mean One way to deal with uncomfortable disparities in valuations of life is to simply cover up those disparities by averaging the values out. While I treat this as an economic response, it is better described as a political compromise than a justified theory. As ideological purists, philosophers have an inclination to propose solutions that may have negative practical implications, but are justified in theory as we saw in the previous chapter. Politicians, on the other hand, have a proclivity for solutions that are practically and politically viable, even when those solutions are not justified by the underlying theories. In this chapter, we will look at one such economic solution that allows governments to sweep pesky concerns about truth under the rug in favor of something that is politically appealing, if inaccurate and unjust. Before getting into the argument for averages, it is important to note that it is possible to compute the value of a statistical life at the subnational level. Since the value of a statistical life based on absolute willingness to pay is closely tied to income and wealth, these subnational values will also show significant variation. Just as the value for poor countries will be less than the value for rich countries, the value of the life of a New Yorker will be greater than the value of the life of someone from Selma, Alabama. This means that even national governments need to deal with disparate valuations of statistical lives, though these disparities are likely smaller than the global ones. As we will see, while some such subnational differences

may be justified by ethical and economic theories, there are political realities that prevent them from being implemented. Governments are often required to value statistical lives in their policymaking but are also accountable to their people, who are frequently dismayed when their government says publicly that some lives are worth more than others. Therefore, many of these institutions have come up with another solution that simply covers up the problem without solving it: averages. Outrage at the differential valuing of life is to be expected for any government with sufficient transparency to actually explain their methods for cost-benefit analysis. To resolve these issues, organizations and agencies have often opted for using averages instead of disaggregated values of life. This involves averaging the measured value of a statistical life over very different populations with different risk preferences. One agency of the U.S. government that commonly uses the value of a statistical life in its calculations is the Environmental Protection Agency (EPA). According to the EPA’s website:   “EPA recommends that the central estimate of $7.4 million ($2006), updated to the year of the analysis, be used in all benefits analyses that seek to quantify mortality risk reduction benefits regardless of the age, income, or other population characteristics of the affected population until revised guidance becomes available.”[75]   The current policy is due in part to the backlash against the previous practice of the EPA known as the “Senior Discount.” This policy set the value of a statistical life for someone over 65 as 37% lower than everyone else for exactly the same underlying reasons in the trolly problem at the beginning of this chapter: shorter lives are worth less.[76] This change was met with sharp public outcry and was quickly retracted.[77] A similar and apropos scandal plagued the IPCC in 1995, as Ackerman and Heinzerling note “they were valuing lives in rich countries at $1,500,000, in middle income countries at $300,000, and in lower income countries at $100,000.”[78] Again, following public outcry, the next report averaged all lives out to one million dollars each.[79] If we just took averages of the statistical lives, we would resolve the problem of differential treatment of lives in cost-benefit analysis since all life would be valued the same by

definition. This section will examine whether this seemingly simple solution is justified.   Valuing an Average Life Taking an average for the American population currently results in the EPA’s estimate of $7.4 million in 2006 dollars.[80] Weighting Viscusi and Masterman’s country specific estimates by World Bank population data, I calculate a global average of $1.816 million. This would appear to solve our original problem. If all of the lives in the initial WHO problem were valued at $1.816 million, then the policy that saved more lives would win out. However, it still allows for lives to be valued and compared with other things that people value. There are a number of points in favor of this simple solution. First it actually addresses and resolves the problem. If all lives are valued at the average value of a statistical life, then when given a choice between two policies with the same cost that only impact mortality risk, we will always choose the one that saves more lives. Second, it is easy to calculate and measure. As demonstrated above, it does not take much to convert disaggregated values of a statistical life into an average. This means that no additional costly measurements need to be taken, and even in cultures where it is difficult to do wage risk or contingent valuation studies, we can still value mortality risk. Third, it retains the utility of cost-benefit analysis. We are still able to calculate empirically the net benefits of a particular program. Finally, it is politically palatable. Beyond avoiding the original moral question, it prevents politicians from being required to answer questions about why one person’s life is worth more than another’s. Intuitive as this claim may be, it requires an important assumption: that there are no ethically salient features that might make us think that one life is worth even a small amount more to society than another. As we saw with the trolley example above, this seems to miss some important concerns, such as the length of a life. In this chapter and the next, I make the case that there are in fact some ethically salient features of life that might make us value individual lives slightly differently in different cases. However, as I will argue in the next section, wealth is not one of these features. Therefore, absolute willingness to pay cannot be used as a valid measure of the value of a life. Instead, we must control for wealth using percentage willingness

to pay. This will lead to our valuing lives slightly differently, but when these differences are only based on ethically salient differences, they become drastically smaller.   Ethically Salient Differences Although the solution of averages is fair across groups, it is not accurate. This average is a political compromise that is not justified by the underlying consequentialism that supports cost-benefit analysis. There are some ethically salient features of a population that should feature in our valuation of life. By ethically salient features, I mean features of a segment of the population that we would be justified in using to determine which policies to apply to them. One clearly ethically salient feature that averages fail to take into account is preferences. Some populations may actually be more risk averse than others. Some communities might value money more than health, while other communities might value health more. If our goal is to implement democratically supported policies that conform to the will of the people, these preferences matter. On the other hand, a feature that is not ethically salient is eye color. We should not apply different policies to a population simply based on the eye color of that population. The problem of differential valuation of lives is not that lives are valued differently in principle, but that they are valued differently for the wrong reasons, i.e., based on features that are not ethically salient, such as wealth. If it is the case that there are some ethically salient differences that might impact our valuation of a life, then averages will not be an accurate solution to the problem. Averages obscure ethically significant pieces of information that should influence our decision-making. That is not to say that all pieces of information about an individual or a population are ethically important and should be considered in decision-making. Rather, it is to say that there exist some characteristics that we should take into account. If there are ethically salient characteristics of a population or individual that might lead us to another decision, we have a moral duty to incorporate that information into our analysis. Averages fail to incorporate this information. To demonstrate both the existence of such characteristics as well as their ethical importance, I offer the following three thought experiments.

Imagine that you are the chief of medicine at a hospital. You have two patients who are about to die and both need a heart, but you only have one heart available for transplant. One patient is 95 years old and has other conditions, which mean that he is expected to only live a few more weeks, even if he gets the heart. The other patient is twenty years old, otherwise healthy and is expected to live for another 60 years at least if she gets the heart. I would argue that, in this case (as with the SWAT trolley problem), life expectancy is an ethically salient feature of the patients. If our goal is to maximize the number of years lived by these patients, then it is clear who should receive the heart. The heart will give more benefit to the patient who will live for 60 years than the patient who will live for three weeks. A similar intuition is what led the EPA to briefly institute its “Senior Discount” on the value of life of older populations. The case for averages would claim that we should average out the life expectancy of the patients, and make a decision on the basis of this average. In effect, both patients have an average of 30 years to live, so there is no reason to give the heart to one over the other. Both lives have equal value. My intuition is that, if you find life expectancy to be a legitimate characteristic to use to make decisions, you would find this averaging abhorrent. The averaging hides ethically important facts about the case, and may lead to one making a decision on the basis of incomplete data. Even if you would choose to save the 95-year-old man, it seems that the life expectancy of the patients is at least somewhat relevant to your decision. Regardless of the choice that you make in this situation, it seems that it is important to leave this information disaggregated to prevent making a decision based on incomplete information.[81] Camperville and Hungrytown For a more direct application to cost-benefit analysis, consider the following situation. A small, rural county is made up of two communities. Let’s call them Hungrytown (population 200) and Camperville (population 100). The county council is planning to build a food bank and a homeless shelter, but they have not decided where to build them. Residents of Hungrytown really want a food bank: they would each be willing to pay $149 for a food bank, but only $1 each for a homeless shelter. The residents of Camperville, on the other hand, really want a homeless shelter and would each be willing to pay $298 for a homeless shelter, but only $2 each for a

food bank. Intuitively the county council should build the food bank in Hungrytown (disaggregated benefit of $29,800) and the homeless shelter in Camperville (disaggregated benefit of $29,800). However, if we average out willingness to pay, as the EPA or the UN would have us do, we find that the average benefit of a food bank or a homeless shelter is $100 to all residents of the county. Therefore, we should build both in Hungrytown because they have more people. Averages underestimate the benefit of a food bank to Hungrytown (at $20,000, when it should be $29,800) and overestimate the benefit of a homeless shelter to Hungrytown (at $20,000 when it should be $200). Averages also underestimating the benefit of a

homeless shelter to Camperville (at $10,000 when it should be $29,800). Based on this it should be clear that averages not only fail to take into account ethically relevant features, such as personal preferences, they are also less accurate, and do not recommend the most beneficial policies. Such an argument can apply to the value of statistical lives as well. Instead, imagine a power plant that provides good paying jobs but decreases air quality and a park that provides no jobs but improves air quality. And instead imagine Camperville has a very high unemployment rate that would gladly bear a slight increase in mortality risk in exchange for more jobs. Hungrytown has full employment and places a high value on their air quality. Averages would set the value they place on a statistical life as the same, and place both projects in Hungrytown, despite their preferences signaling otherwise. The point is that there are at least some ethically salient differences that we should take into account when valuing statistical lives. If any such differences exist, averages cause harm by limiting the ability of decision-makers to take preferences into account. These three thought experiments support the claim that there is ethically salient information that might impact the value that a government places on a statistical life. If such information exists, using it will make our costbenefit analysis more accurate and responsive to the will and preferences of the people. These averages not only make the tool of cost-benefit analysis less accurate; they fail to take into account the preferences of individual populations. However, averages do point clearly to a problem with costbenefit analysis as it is currently practiced: the inclusion of wealth, an ethically irrelevant feature. Current cost-benefit analysis uses absolute willingness to pay to calculate benefits. Absolute willingness to pay is limited to ability to pay. This means that any benefit calculation using absolute willingness to pay inherently includes wealth as a factor. Individuals with more money have a greater ability to pay and therefore will have a greater willingness to pay, not because of actual differences in preferences but because of the ethically irrelevant factor of wealth. In the third section of this book, I argue that wealth is not an ethically salient factor, but we can control for it in costbenefit analyses by using percentage willingness to pay. Others have made the case that averages are harmful and inaccurate. In the following chapter, I explain Sunstein’s case for using disaggregated

values of statistical lives at the subnational level. Doing so resolves the concerns presented in this chapter that there are at least some ethically salient factors that should be taken into account when valuing statistical lives. Sunstein also presents a solution to the original problem of disparate valuations of taking averages at the international level. This resolves the original concern without sacrificing ethically salient information about the population. However, I argue that, on the subnational level, Sunstein includes some differences that are not ethically salient and on the international level ignores others that are.

 

8. WHY SOME LIVES ARE WORTH MORE   I began this book with the concern that there was something deeply wrong with different lives being given different values. This chapter will defend the opposite claim: that there are some ethically salient features of a population that might lead us to value lives slightly differently. The problem then is not that lives are valued differently but that we are using ethically irrelevant features, such as wealth, to value them. This chapter looks at this argument for disaggregation in greater depth, as well as a fifth and final insufficient method for solving the original problem of differential valuation of international lives. Specifically, I focus on two arguments made in Sunstein’s paper “Valuing Life: A Plea for Disaggregation.”[82] Sunstein presents a unique view on the subject of disaggregation. In order to both address the issues of overly simplified averages and still avoid the initial problem of international valuations, he argues for disaggregation at the national level, but for aggregation and averages at the international level. In other words, individual countries should disaggregate their value of a statistical life based on demographic factors (i.e., saving the life of someone who would die in a few hours regardless should count differently than saving the life of someone with years to live), but international organizations should average and value all lives the same. In this chapter, I document his argument against averages at the national level, and his case for averages at the international level. I also make the case that, while he is right about some of the ethically salient features of sub-national lives, these should apply in international contexts as well. I also make the case that he fails to exclude the most crucial ethically irrelevant feature: wealth. The Case for Disaggregation Sunstein makes an argument for disaggregation of risk valuation by person at the national level but not at the international level. In other words, national governments should use different values of a statistical life for members of their own populations, but we should not compare these values internationally. To elicit the intuition, Sunstein asks us to imagine a regulatory agency that perfectly knows exactly how much everyone would

be willing to pay to reduce a particular risk. Individuals willing to pay more to reduce risk would buy safer, regulated, more expensive products, while individuals who had a lower willingness to pay to avoid risk would buy cheaper, unregulated, riskier products.[83] Similar to a market, this would maximize the surplus willingness to pay as individuals could consume products with exactly the amount of risk they were willing to pay for. Basically, those who could bear more risk would be free to spend their money on other things. However, such a system is impossible due to the lack of perfect information about everyone’s preferences and the fact that regulatory benefits can’t be provided to some without providing them to everyone. Although such a system may be impossible at the individual level, Sunstein claims that we can implement it at the demographic group level.[84] Sunstein argues that if a risk primarily impacts a group with a higher willingness to pay to reduce that risk than the average population, we are justified in using a higher value of a statistical life for them. Since more benefit is gained by risk averse populations avoiding risk, we can justify spending more on them. We might think of this in terms of speed limits. If a community is very risk averse, they should be allowed to set the speed limits in that community very low, even if that means they lose out on the benefits of getting places faster. On the other hand, if a community values fast travel over risk reduction, they should be allowed to set the speed limits in their neighborhood higher, even if that means they are more likely to die. Note that this is an imperfect example, as speed limits impact visitors who may have different risk tolerances, but the general idea should be clear. The following thought experiment provides a more detailed and specific example of why some lives should be valued slightly more. Cowardsville and Braveton Imagine two communities that have very different risk tolerances. One is a community of 10,000 people living in a town named Cowardsville. Nationally, people are willing to pay $7 to reduce a 1 in 1,000,000 risk of death (for a statistical value of life of $7 million). However, in Cowardsville, people are very risk averse and they are willing to pay $20 to reduce a 1 in 1,000,000 risk of death (for a statistical value of life of $20 million). Otherwise, residents of Cowardsville are perfectly representative of the nation (in terms of income, race, age, etc.). Now, imagine that there is

a risk impacting only residents of Cowardsville that causes a 1 in 1,000 risk of death. In order to prevent this risk, the government would need to implement a policy that costs $100 million and has no other benefits. If we used the national value of life, we would calculate the benefits to society to be $70 million (1 in 1,000 risk of death in a community of 10,000 means 10 deaths, at $7 million each) and the costs to be $100 million. Therefore, we would not recommend this policy, as it would have a negative net benefit. However, this is not the actual benefit that Cowardsville would receive. They would get a benefit of $200 million since they are very afraid of death. In other words, they would be willing to vote to pay up to $20,000 each in new taxes to reduce this risk. Therefore, being offered to only need to pay only $10,000 each in taxes would be a bargain deal for them. Sunstein’s argument is that we are justified in implementing such a program since the actual benefit to the community is greater than the costs. The $70 million in benefits is an inaccurate measure based on data that is not representative of the impacted population. What this means is that the people of Cowardsville actually do have a higher value of a statistical life than the national average. The benefit that they receive from a small reduction in risk is greater than the benefit that most of the population receives. It is important to note that Cowardsville is representative of the population in all other ways, so that this willingness to pay more for a reduction in risk is not solely due to them having more wealth. This means that their higher value of a statistical life is driven by real preferences, not wealth or some other ethically irrelevant feature of the population. As we saw in the previous section, the largest gaps in willingness to pay are often driven by wealth. That is why, as I argue in the next section, we need to control for ethically irrelevant factors that influence willingness to pay, such as wealth. To fully represent Sunstein’s argument, imagine another community of 10,000 called Braveton. The people of Braveton are particularly risk seeking. While nationally people are willing to pay $7 to reduce a 1 in 1,000,000 risk to life, the people of Braveton are only willing to pay $2 (but, as with Cowardsville, are otherwise demographically representative of the population of the country). Now, imagine that Braveton experienced a risk of 1 in 1,000, which would cost $50 million to prevent.

Using the national value of a life, we find that Braveton should enact this policy ($70 million in benefits, $50 million in costs). However, according to Sunstein, this is putting an undue burden on the people of Braveton, as they would only be willing to pay $20 million to prevent such a risk. Using the more accurate value of a statistical life for Braveton would not impose this undue burden on the people of Braveton who would rather spend their money on things other than risk reduction. This may seem cruel to this community, but in fact it seems representative of how communities have different values. One community that is risk averse

with respect to their health might want the government to build a hospital there, while another community that values education more might prefer that the government build a university. It is not the place of cost-benefit analysis to tell a community what they should or should not value, merely to aggregate those values into a decision tool. Sunstein argues that we should value risks by demographic groups so that if a particular group is more impacted by a risk than another, we can use a more accurate value of a statistical life for that group. This will assure that the government neither imposes undue costs or fails to eliminate risks that individuals would be willing to pay to eliminate. The point of both of these examples is that using averages will fail to accurately represent the wishes of the individuals in a particular country, by charging them for protection that they do not want, or failing to protect them when they would be willing to pay for it. Together with the arguments from the previous chapter, this is a reason to think that there are some ethically salient features of populations that would lead us to differentially value their lives. There seems to be an ethical argument that if we had to choose between a program that saved one statistical life in Cowardsville or nine statistical lives in Braveton, this analysis would point toward saving one life and letting nine die. The problem with such an example is that, as a matter of fact, such larger differences do not appear in communities with similar wealth. In fact, as we will see in the next section, while there are minor differences in actual preferences the majority of these difference that we see in valuations of statistical lives is driven by differences in wealth. Sunstein fails to identify that there are, in fact, some features of a population that are ethically irrelevant and should not be incorporated in this analysis. This is why I explicitly stated that the populations of Cowardsville and Braveton are otherwise identical to the rest of the nation (as without this, their disparate willingness to pay would almost certainly be based on the ethically irrelevant factor of wealth). Without controlling for wealth, Sunstein’s argument in favor of disaggregated values of statistical lives is in grave danger of doing the same thing that cross country valuations do: devaluing poor lives and overvaluing rich ones. To avoid these concerns, we must use percentage willingness to pay in order to control for wealth in such calculations. While percentage willingness to pay

as outlined in the final section does have some differences between the values of life for particular countries, these differences are very small, verging on nonexistent especially when compared to the same estimates found using absolute willingness to pay.   International Valuations Sunstein makes a note that this position may have harmful consequences in an international sphere, though he does not flesh out a full argument for this claim. Here I attempt to fill in the gaps in his position and offer such an argument. Despite his claim that different countries should use different values for statistical lives, he claims “this point should not be taken to support the ludicrous proposition that donor institutions, both public and private, should value risk reduction in a wealthy nation above equivalent risk reduction in a poor nation.”[85] In other words, while he supports using specific values of a statistical life for different countries, he does not support using these values for cross-country analyses.[86] In those cases, he is willing to compromise the principles he just argued for at the national level and use averages, as he claims that there is a justifiable difference. Sunstein justifies this claim by comparing who bears the costs of the government of a developing country to who bears the costs of an international philanthropic organization. The costs of an international philanthropic organization that engages in a risk reduction practice in a developing country are not, according to Sunstein, borne by the people in that country, and therefore the willingness of individuals in that country to pay for risk reduction is not applicable. On the other hand, since the government of a country is eventually funded by taxes from people of that country, that government has a responsibility for the people of that country to not bear an excessive burden. Basically, because people in a developing country do not bear the cost of philanthropic interventions, their willingness to pay for the risk reduction is not applicable to these analyses. The response relies on the fact that we compared the willingness to pay for the people of Cowardsville and Braveton to the actual cost. If there was no cost to the people, then any policy would be pure benefit. Either policy that reduces mortality in Cowardsville or Braveton would be justified even if they had a very low value of a statistical life. While this response does address the central problem in much the same way as the averages, there are

a number of issues with it, both including ethically irrelevant features of a population and failing to recognize the cost of international organizations that people in developing countries bear.   Multilateral Governments There are three important issues with Sunstein’s argument that make it untenable to solve the initial problem. First, the argument against disaggregation at the international level is only effective against private philanthropies whose costs are borne solely by the philanthropy, not multilateral organizations or philanthropies/bilateral organizations that partner with multilateral organizations or country governments. Second, while some kinds of disaggregation are based on ethically relevant information, others are based on ethically irrelevant information, and we need to only disaggregate by the relevant information. Third, even for private charities it is unethical to ignore the preferences of the populations since there may be non-monetary costs that they must bear. Setting aside the issue of completely private charities for now, Sunstein fails to explain the correct course of action for international government entities, also known as multilaterals. Multilateral governments like the United Nations are funded by member countries, which are in turn funded by taxpayers in those countries.[87] Therefore, multilaterals have a responsibility to take the preferences of all individuals into account since the interventions are eventually funded by those individuals, and they should not pay for something that they do not support. Additionally, private foundations often partner with multilaterals or country governments to fund projects. While this may affect the costs borne by the government, it does not change the fact that such projects are eventually responsible to the people. The original problem we looked at is not a problem for projects funded solely by the Bill and Melinda Gates Foundation, nor for any individual national government. The problem is for multilaterals (such as the WHO, the World Bank, the Internal Monetary Fund, etc.), the costs of which are eventually borne by the citizens of all member countries. There is no reason why multilateral organizations that implement policies that reduce risk and impact people around the world should be subject to any less scrutiny than national governments.

In defense of Sunstein, one might claim that multilaterals are importantly different than national governments for a range of reasons. Multilaterals generally have less power over citizens than national governments. They cannot enact laws, or put people to death for crimes. They cannot force the people of a country to fight in a war with another country. Therefore, due to these limited powers, people in developing countries do not have a right to have a say about what a multilateral does. Further, a Sunstein defender might argue that because multilaterals are mostly funded by rich nations, they are similar to philanthropies since the recipients do not bear much of the cost. The reason that we care about disaggregation of risk valuations is because we do not want to impose undue burdens on people who would not be willing to pay for a reduction in risk. If the individuals in developing countries pay very little into the system, then they would not be concerned about spending more or less on risk reduction because they would not bear the burden of the cost of risk averting policies.   The Power and Cost of Multilaterals However, these arguments are unconvincing for several reasons. First, while multilaterals do not have significant power over high-income developed countries, they do exercise great power over developing countries through conditional financing that can dictate policies. Often these bodies have more power than developing country governments to determine policy because their funding is contingent on certain policies being enacted.[88] Regional and international organizations do send military peacekeeping forces into countries.[89] They can use conditional funding to impose particular policies or political structures. They do all of these things under the guise of acting in the interests of the world, not just in the interests of their donors. If multilaterals have as much, if not more power as country governments in developing countries, there seems to be no reason to hold them to a lower ethical standard. Additionally, these organizations make the claim of acting as a world government. Whether or not they have the power to back up that claim, if they wish to make it, they should be subject to the same ethical constraints of a government. One would not claim that we should hold a particularly ineffective national government to a lower standard than any other national

government, simply because they are ineffective. They should be held to that standard as they purport to speak for the people. Simply because a government is not very powerful and inefficient at its task, if it purports to represent the people, as the United Nations, and other multilaterals do, it should be responsible to act in the interest of the people, and therefore consider the actual preferences of individuals in developing countries relative to risk. The claim that the cost is not borne by individuals of developing countries, and therefore they should not have a say in how the money is spent, also fails for two reasons. First, low-income countries may contribute low absolute amounts to these multilaterals, but these amounts are much larger when considered as a share of their overall budget. The Gambia contributed only 0.006% of its government expenditures to the UN last year, but in percentage terms, this was not much less than Germany, which contributed 0.009% of its budget.[90] While Gambian citizens may be contributing a smaller absolute amount, they are contributing a comparable share of their income in taxes. Second, the same claim could be made of poor states in the U.S., where states like Wyoming contribute around 0.1% of federal tax revenue.[91] If we find that the United Nations is effectively a philanthropy, even though all nations contribute to its costs, the U.S. federal government is effectively a charity for poor states. If organizations do not need to concern themselves with the preferences of members who contribute small amounts to their overall budget, then the federal government does not need to concern itself with the preferences of individuals in poor states. However, Sunstein does not think the federal government can be viewed as a charity where rich states help out poor states. He thinks that federal laws should be subject to review through cost-benefit analyses. Sunstein that thinks that poor American states, which (like poor countries) have their own governments, are subject to some constraints from a government above them, are somewhat represented by that government, and pay into it only in a small amount and deserve differential valuations of statistical lives (due to the concerns that arose in the examples of Braveton and Cowardsville). He is therefore committed to the same for developing countries. If the federal government has the responsibility to take the wishes of poor states into account, multilaterals have that same responsibility to developing countries.

Therefore, multilateral organizations should be subject to similar regulations since their relationship to member countries is not economically different from the relationship of the federal government to individual states.   Private Charities are not Immune A final concern with this line of argument about the economic relationship between multilaterals, charities, bilateral donors, and developing countries is that, even if people in developing countries bear none of the monetary costs, their preferences should be taken into account with respect to projects that provide pure benefit since they will bear the opportunity costs. People may value benefits other than reductions in mortality risk, and those preferences should be taken into account, even when those people bear none of the direct monetary costs. This is because some projects may preclude other projects. A region may only need one major university or hospital. If one community prefers one over the other, that matters, even if they pay for neither. To elicit this intuition, take the following thought experiment. Imagine two neighboring countries, Farmerstan and Sickmavia. Farmerstan places a high value on being agriculturally self-sufficient, having the best farm produce in the world, and the jobs that come from agriculture. Sickmavia on the other hand has been ravaged by disease and values health and long life over having strong agricultural production or economic prosperity. Now imagine that a wealthy philanthropy is deciding where to put two large research institutes, one working on disease prevention and one working on increasing agricultural productivity. The research center working on disease prevention will provide a substantially higher reduction in mortality risk in whichever country it is placed in (as it will be able to focus on the particular diseases endemic to that country). The agricultural research center will provide a more substantial increase in productivity and new technology to whichever country it is placed in. Only one of each facility is needed for this region of the world (i.e., if the disease research facility is placed in one country it is unlikely that another such facility will be placed in the other country). If we take Sunstein’s argument that we do not need to account for preferences of charity recipients at face value, we must defend the claim

that the charity is fully justified putting the disease research facility in Farmerstan and the agricultural research facility in Sickmavia, even if that is against those people’s wishes, and even if those communities bear the opportunity cost of not getting the other research facility. What Sunstein fails to realize is that simply because the people in a country do not pay for an intervention, it does not mean that they do not bear non-monetary costs. In this case the people bear the cost of losing the research facility that they actually wanted. If an intervention by a charity costs a local community jobs by giving away for free something that the community used to make in businesses, according to Sunstein, that community has no recourse.[92] Sunstein claims their preferences are unimportant, as they bear none of the direct monetary costs, but they do bear some costs because they bear nonmonetary or less direct costs. Ethically Irrelevant Features However, there is one final objection to this disaggregation, which we have mentioned in passing throughout this section and will be fully explained in the next section. While there are some things, such as risk tolerance, that we should have information about for a population before imposing a policy on it, there are other things that we should not know, or at the very least, should not use in our unbiased decision framework. Imagine once again that you are the chief of medicine of a hospital deciding between two patients in need of a heart transplant. It is relevant information for an objective decision whether one of them will die in a week due to another disease. It is relevant whether one wants to die and the other wants to live. However, it is irrelevant information whether one of them is your uncle, the daughter of an influential politician, or a member of a religion that you vehemently disagree with. Making decisions on the basis of ethically irrelevant factors is problematic, as your personal opinions or relations should not impact your decision making as a public official. The issue with Sunstein’s initial case for disaggregation is that it does not take into account that there are some pieces of information on the basis of which you should not make decisions. In fact, the potential for this personal bias on the basis of irrelevant

 

Some features of a population are ethically relevant to decision making and should be used in CBAs, like preferences.  Others such as family ties or wealth should not.

factors is exactly the reason that we would want to engage in cost-benefit analysis in the first place. The question that might be raised in response is: What factor is both incorporated into the valuation of a statistical life and is irrelevant to our decisions about risk reduction? Familial relations are clearly not factored into disaggregated valuations of statistical lives. The answer, as we investigate in the next section, is wealth. Wealth, in the form of ability to pay, is inherently factored into every calculation of the valuation of a statistical life, and yet it should be irrelevant to our decision of whether to save a life. When statistical lives are disaggregated, the lives of the wealthy will consistently be given a higher value for no other reason than they are wealthy. The amount of money you have should be irrelevant to a government’s decision on whether they should save your life, but using our current model for cost-benefit analysis it is the only really relevant factor. This is immoral, and drastically skews any social policy in favor of the wealthy. In the next section I not only demonstrate that this is the root cause of disparate values of a statistical life, I demonstrate how fixing this minor methodological issue, this flaw at the center of economics, can help address problems from the next pandemic, to racial injustice to global inequality. We can disaggregate the value of a statistical life by ethically relevant factors as Sunstein advocates, without biasing these data by using wealth to measure it.

 

   

           

III. PERCENTAGE WILLINGNESS TO PAY How a small methodological change can drastically reduce global inequality

 

           

9. THE PROBLEM WITH ABSOLUTE     Imagine that you are standing in line at a polling station to cast your vote. However, this year, it seems the poll workers are doing something different. Each person who walks away from the registration table has a different number of ballots. Most people have one or two, but some have three or four. One man in a suit walks away with a large stack of at least 20! A woman is turned away without a single ballot. When you get to the registration table, the poll worker explains that economists have taken over the government, and in order to fully understand the depth of your preferences they need to understand how much money you have. If you don’t have any money, you can’t be willing to pay anything for a candidate’s proposals, so your vote doesn’t count. But if you have lots of money, you can feel the impacts of a policy more, you can be willing to pay more for them, your preferences matter more, so you are given more ballots. They look up your IRS statement, give you two ballots and send you on your way. You wonder at the immorality of it all: Why should people with more money get more say in what the government does? Surely economists would realize how problematic such a system is. It is corruption, where you don’t even need to pay the bribes! In this chapter we will see that, not only would economists not realize the problems of such a system, the system they use right now already includes such inaccurate and unjust calculations that value the preferences of the rich more than the poor (even if those rich people never end up paying for anything). All of these failed solutions from section II have pointed in one direction: that we need to divorce willingness to pay from wealth. In this chapter I offer an explicit argument that the ethically problematic methodology that gives rise to immorally valuing the lives of the poor less is the calculation of benefits using absolute willingness to pay. In the remainder of the section, I defend a particular method for avoiding including absolute willingness to pay in cost-benefit analysis: percentage willingness to pay. I show that not only is a rigorous and objective cost-

benefit analysis still possible using percentage willingness to pay, it is both more accurate and more just.     The first task is to link absolute willingness to pay to our central problem. I offer the following simple argument in order to do this:   P1: Using absolute willingness to pay, the value of a statistical life is mostly determined by measures of income or wealth. P2: Using a method to determine the value of a statistical life that is mostly determined by ethically irrelevant features of the population is immoral. P3: Wealth and income are ethically irrelevant features of a population. P4: Using absolute willingness to pay, differences in the value of a statistical life are mostly determined by ethically irrelevant features of a population. (P1, P3) C1: Therefore, using absolute willingness to pay is immoral. (P2, P4) Premise four and the conclusion generally follow from the first three premises. Therefore, for the remainder of the chapter I will defend each of the first three premises in turn. Note that this argument does not determine a specific replacement for absolute willingness to pay. Although I advocate a specific alternative in the chapters to follow, I am not committed to the claim that percentage willingness to pay is the only ethical alternative methodology, or even the best alternative. I am simply making the case that it is an improvement on the status quo, and responding to the concerns of Paul Portney and others who claim that “BCA analysts use dollars to estimate benefits because there simply is no other way to directly measure the intensity with which people desire something.”[93] In fact, there are many alternative ways to mathematically control for wealth, or income, any of which may satisfy this argument.   You Are What You Own

The first premise claims that there is a link between the value of a statistical life for an individual (as measured by absolute willingness to pay) and their wealth or income. There are two ways this can be shown, theoretically and empirically. In what follows, I first examine the theoretical case for why someone with less wealth or income might gain less benefit (as measured by absolute willingness to pay) from a reduction in risk when compared with someone with less wealth. Then I will show that this is borne out using the data from Viscusi and Masterman’s paper. Theoretically, there is one feature of absolute willingness to pay that marks it as clearly dependent on wealth: the fact that absolute willingness to pay is constrained by ability to pay. If someone with only $100 to their name is asked on a contingent valuation survey how much they would be willing to pay to reduce a 1 in 10,000 risk, they cannot accurately say any more than $100. Therefore, the value of a statistical life for such an individual will never be more than one million dollars. An individual with even $10,000 in wealth asked the same question could give a response leading to their life being worth as much as 100 million dollars. Because the poor can never be willing to pay much, on average their lives will always be valued less. This makes sense, the more money you have, the more you would be willing to pay to reduce a small chance of death. As shown in Chapter 3, there is good reason to believe that both contingent valuation and wage risk studies place a higher value on the lives of the wealthy. It is important to understand that these do not result from underlying differences in preferences. Someone with $50 to their name, who is willing to spend everything they have to reduce a 1 in 1,000 risk is intuitively much more risk averse that someone with one million dollars to their name who is willing to spend only $500 to reduce a 1 in 1,000 risk. While it may seem that the poor person values their life more (they would give every cent they have to prevent that risk), the value of the poor person’s statistical life is only $50,000 despite them spending everything they had. The millionaire’s value is ten times that, despite only being willing to spend 0.5% of their fortune. These differences in the value of a statistical life are not due to differences in risk preference. If anything, the poor person is much more risk averse than the millionaire, despite absolute willingness to pay providing us with the opposite conclusion.

Here's another way to think about this. Imagine that you are lost in the jungle with two friends. You come to a river and see on the far bank some people with a boat that could help you out, but they don’t notice you. Unfortunately, you cannot swim and both of your friends are afraid of piranhas (which commonly live in these waters). As a good friend, you want to figure out who is more afraid of piranhas and send them across the river. One of your friends is quite rich. He pulls out a wad of many one hundred dollar bills and offers you one of them if he does not have to swim. Your other friend is poor. She gives you all the money in her pocket ($10.93) the last of her water and food (valued at $20), and offers to sell you the clothes off her back (valued at $30) if you don’t ask her to cross the river. Who is more afraid of piranhas? My intuition is that the person who gives all of her possessions is more afraid of piranhas than the person who only offers one of many hundreddollar bills. However, if you think that absolute willingness to pay is an accurate measure of utility, you are committed to the claim that the rich person is actually more risk averse to piranhas since he is willing to pay more (even though he is only willing to pay a small fraction of his wealth). Both of their absolute willingness to pay is determined more by how much money they have than by the intensity of their desire. Economists have admitted as much, with Portney claiming that a rich person and a poor person will be judged to gain different benefits from an intervention even if they both have the same intensity of feeling about that intervention.[94] Absolute willingness to pay is not a measure of benefit or intensity of feeling. It is a measure of wealth. It remains to be seen whether or not this is borne out in the data. While wealth may have a large impact on willingness to pay, it is not clear that it is the only thing leading to differences in the valuations of statistical lives. There are huge differences in the values of statistical lives presented by Viscusi and Masterman. If these differences are driven by risk preference and not wealth, then the point is moot. However, if we control for wealth and the differences largely disappear, then these cross-country differences are primarily driven by wealth, and we have strong support for our first premise. A primitive but intuitive way to do this is to use gross national income per capita for a country as a proxy for wealth and simply divide the value of a statistical life estimates by this factor. If the resulting estimates

are substantially different, then wealth may not be the driving factor. However, if they are largely the same, that provides strong support for the claim that, using absolute willingness to pay, the value of a

statistical life is mostly determined by measures of income or wealth. Taking the values provided by Viscusi and Masterman for GNI per capita, I find that when wealth is controlled for (by dividing the value of a life by the GNI per capita), the resulting values are almost identical. The largest value is 173.2, and the smallest is 170.6. In other words, the largest value is only 1.015 times larger than the smallest. This is particularly meaningful given that the largest value for a statistical life using absolute terms is 405.8 times larger than the smallest value. In other words, a methodology that controls for wealth would claim that saving 34 of the most risk averse lives in the world would be equivalent to saving 35 of the most risk seeking lives in the world. Though there is still some inequality here, it seems to be based on their preferences not their wealth, and a government might be justified in saving slightly fewer risk-averse people instead of slightly more risk-averse people. Using absolute willingness to pay, saving the life of someone from the most risk-averse country in the world would be equivalent to saving 406 lives in the most risk-seeking country in the world, which clearly seems unjustified, particularly because these differences are driven by wealth. The next premise explains why the small preference-based differences are morally acceptable, but the large, wealth-based ones are not. Based on this analysis, there is strong evidence that any cross-country differences between these values are due almost entirely to differences in wealth, not actual underlying preferences. Appendix D includes a table of all of these values for reference listed as percentages of GNI per capita for a reduction of a 1 in 1,000 risk. As the next chapter shows, this basic operation can be used to control for wealth and get almost identical values for statistical lives across countries, where the only small remaining differences are the ethically relevant underlying risk preferences.   Ethically Irrelevant The second premise claims that there are some features of a population that morally should not impact a decision of how much to value mortality risk

reduction. Simply defined, ethically relevant features are those that directly impact the utility of the individual or group that is receiving the risk reduction. If one group of people is much more afraid of dying from cancer than other members of the population, this fear is an ethically relevant feature when considering where to put a cancer treatment facility. Those who are more afraid of cancer will get more utility from having such a facility near them than those who are not. Factoring in ethically relevant features will allow a policymaker to maximize the benefit to the population. Ethically irrelevant features then are those that do not impact the utility of a group (as we will argue below, wealth fits this description since you do not have greater capacity for utility simply because you have more money). Using a decision method that relies on ethically irrelevant features will lead to policies that fail to maximize benefit to the population, and therefore are immoral. The following example demonstrates how basing a decision off ethically irrelevant factors can fail to maximize the benefits for a community. Imagine that you are the elected chair of a local zoning board for the city of Nimby Grove. There are two vacant lots in residential areas of your town that you need to rezone. Nimby Grove has decided that they are in need of a new waste treatment plant (which will increase mortality risk in the surrounding area by 1 in 1,000,000) and a new park. It is up to you to decide which lot to rezone for each amenity. Some ethically relevant features that you might take into account include the risk tolerance of the surrounding neighborhood, how many people might be impacted by the cost of the increased mortality risk of the waste treatment plant or the benefit of the park, or whether the surrounding areas are going to be rezoned for a highway in the near future (forcing people to move out, and thereby decreasing the benefit of the park or the cost of the waste treatment plant). These are all ethically relevant features because they impact the utility of the population directly impacted by the policy. However, there are other factors that should not impact your decision. These include features like whether one area contributed more to your reelection campaign, whether you or your family live near one of the lots, or if someone offers to give you $10,000 to put the park near their house. As I argue below, the average wealth of a neighborhood fits into this category or ethically irrelevant features. Putting a park in a community that contributed more to your reelection campaign will not help to maximize the

direct utility gained by those effected. The amount that a particular population gave to your campaign is ethically irrelevant to your decision to rezone the land near them because how much they contributed does not impact the happiness they get from the park or the cost they bear from the slight increase in mortality risk. Your family might benefit from a park near their house, but being related to a politician does not make people capable of experiencing more happiness than anyone else, so this feature is also irrelevant. Just because someone is willing to bribe you more does not mean that they are going to receive more happiness from the park, than someone who does not have the money to offer such a bribe. And clearly, simply because one neighborhood is richer it does not mean that they will get more happiness from a park or bear greater suffering for the mortality risk of a waste treatment plant. Despite what economists may assume, being richer does not give you a greater depth of feeling. Taking these features into account in an analysis of benefit will divorce the results of your analysis from reality and, in doing so, lead to recommendations that do not maximize benefit, as they inaccurately take irrelevant factors into account. This makes any such

 

If politicians are left to their own devices, they may take into account ethically irrelevant features of a population, such as campaign donations or wealth. If CBAs are effective, they should control for these features.

analysis both immoral because it fails to do the most good, but also erroneous because it inaccurately reports to be doing the most good.   Money Does Not Matter Now that we have shown that the value of a statistical life using absolute willingness to pay is primarily based on wealth or income, and that valuing benefits based on ethically irrelevant factors is immoral, the only remaining question is whether wealth and income are ethically irrelevant features of a population or not. In what follows I argue that wealth and income are factors of a population that are ethically irrelevant to cost-benefit analysis, and therefore valuing statistical lives using absolute willingness to pay is immoral. First, it is important to provide a short note on wealth versus income. Some measures of willingness to pay look at what a population would be willing to pay in terms of annual taxes in order to measure benefit. Other measures look at one-time benefits received. Both of these techniques limit one’s willingness to pay to one’s ability to pay. However, ability to pay is defined slightly differently in each case. In the case of an annual tax increase, willingness to pay is capped at income. However, with a single, one-time expenditure, it is capped at overall wealth. Whichever measure is used to calculate the limits of ability to pay in this particular cost-benefit analysis is the measure that needs to be controlled for. While the discussion so far has focused on wealth, in what follows, I argue that both wealth and income are ethically irrelevant factors, and therefore either method is inaccurate and immoral. When first discussing cost-benefit analysis in Chapter 2, we stated that philosophers had tasked economists with attempting to measure how happy any given policy would make each individual affected on a scale of 1 to 10, and then sum up that happiness. Whichever option had the highest sum was the option that maximized benefit, and therefore happiness. However, as I have shown, this is not exactly what happens. If two individuals were made as happy as possible by a given policy, their happiness should count for the same amount. If everyone else gets a scale that maxes out at 10, no one should get a scale that goes from 1 to 100 or 1 to 100,000. However, a person with only $10 to their name has a maximum willingness to pay of $10. A person with $100,000 in wealth has a maximum willingness to pay

of $100,000. A policy that provides complete and total happiness (i.e., that they would be willing to spend their entire wealth to enact) for one of these individuals counts 10,000 times more for the rich individual than the poor one. According to the underlying assumptions of economics, in this situation making one rich person perfectly happy justifies making 10,000 poor people as miserable as possible (i.e., enacting a policy that they would pay their entire net worth to prevent). This is the same issue that

arose in the voting thought experiment at the beginning of the chapter. The economist’s operationalization using absolute willingness to pay is not true to the philosophical underpinnings of consequentialism. There is no philosophical case made to justify the claim that the happiness of the wealthy should count for more. A charitable interpretation of this is that it is merely a feature of poor measurement on the part of economists. A more cynical view might claim that economists are actively tipping the scales drastically in favor of the wealthy and have created a methodology that values their happiness more (possibly due to any system that questions this failing to receive funding from the wealthy). Either way, the philosophical justification for cost-benefit analysis does not hold up using absolute willingness to pay. However, this merely demonstrates that no argument was given in favor of using wealth to determine benefits. It has yet to show that, in fact, wealth and income are ethically irrelevant factors to utility. It is important to note that I am not making the claim that “money does not buy happiness.” I am not arguing that increases in wealth will not lead to increases in utility. Rather, I am arguing that one’s wealth does not limit or cap the utility that one can experience, nor does having more wealth make physiologically one capable of experiencing more happiness than poor people. In order to illustrate this, take the following example of two individuals, which shows that greater wealth does not translate into a greater capacity for utility. Imagine two men: Haruna and John. Haruna and John have never met, but they both became fathers today, and it was the happiest day of their lives, they were as happy as the possibly could be. They both were by their wives’ sides when they gave birth and could not contain their joy when they looked into the eyes of their new first child. It seems intuitive that both Haruna and John gain a similar if not equal amount of happiness from this experience. Haruna and John are identical in their sentiments, identical in their underlying intensity of feeling. However, Haruna is a schoolteacher in Senegal, and only has $500 in wealth. John on the other hand is a schoolteacher in the United Kingdom and has $250,000 in wealth (which is around the median wealth for that country). If wealth is a relevant factor to maximum happiness, it would follow that, despite appearances, John is actually 500 times happier than Haruna on the happiest day of both their lives. Having less wealth means that Haruna’s capacity for happiness maxes

out at $500, while John’s capacity continues climbing all the way up to $250,000 for the best day of their lives. Put another way, imagine that, before the baby was born, both were told that the doctors could provide a procedure that would reduce the risk of death of the child by half, and then they are asked how much they would be willing to pay for such a procedure. As they both care more about this than anything else, they both say that they would be willing to pay their entire life’s savings for such a procedure. If wealth is a relevant factor to utility, then John cares 500 times more about his child than Haruna does. Using absolute willingness to pay, John gains 500 times the happiness as Haruna if such a procedure were covered by a government policy. However, this is clearly incorrect. John does not gain any more benefit than Haruna from a procedure that saves the life of his child. John is not happier than Haruna when his child is born. John does not value the life of his child any more. The government policy saving the lives of their children does not provide John any more benefit than it provides Haruna. Yet, if you think that wealth is a relevant factor when measuring utility, you are committed to the claim that John is really much happier than Haruna to see his child born, simply because he is richer. According to economists, wealth provides you with an ability to experience more happiness than poor people ever could. If you think that absolute willingness to pay is an accurate measure of utility, then you are committed to the claim that John gets a much greater benefit from a policy that saves the life of his child than Haruna does. Absolute Failure Based on this, it seems clear that absolute willingness to pay is an immoral and inaccurate method for measuring happiness. Current values of statistical lives are largely based not on ethically relevant features like risk preference, but on wealth, as the vast majority of these differences can be explained by differences in income. Making policy decisions on the basis of ethically irrelevant features that do not influence an individual’s happiness is immoral as it fails to maximize happiness. Wealth or income does not give one a greater capacity for happiness, or a greater benefit from a policy. Economists may retort that while absolute willingness to pay is an imperfect measure of happiness or utility, it is the best option that we have. [95] They may claim that further data is difficult if not impossible to collect.

As we cannot see into the minds of individuals, we can only really act based on their revealed preferences in a market, and absolute willingness to pay is the best way to do that. I am sympathetic to the concern that without a better alternative this argument falls flat. If these are the best data that we have, they may be better than simply basing decisions on biased opinions. However, there is a better method, as I will argue in the next chapter. This alternative does not require impossible measurement of the mind, but is more accurate and more moral. That alternative is measuring benefit using percentage willingness to pay.  

 

10. PERCENTAGE WILLINGNESS TO PAY   The rich and famous are notorious for flaunting the law, often because they would rather pay the fine for speeding, parking in a handicapped spot, or shoplifting than actually follow the law. Their wealth makes it so that these fines, which are meant to serve as a deterrent fail to change behavior, but this is not true everywhere. Finland has a unique method of solving this problem, by determining the cost of speeding tickets using a proportion of a person’s individual earnings.[96] This leads to speeding tickets as high as €116,000 for an executive making €14,000,000, and therefore is an effective deterrent against behavior that endangers the public safety, even for the wealthiest.[97] It also means that the poorest are not caught in a debtor’s prison with a flat fine too high for them to pay. This rule exposes an important fact about human psychology: percentage loss or gain is more comparable in terms of happiness lost or gained than an absolute measure of loss or gain. To further elicit this intuition, imagine two women, Koula and Arax, who just received raises. Koula will be making five times as much as she was making before her raise, while Arax will be making $5,000 more. If you were asked to guess how happy this raise made Koula, you could easily guess that such an increase would make her very happy. However, if asked the same question of Arax, you would be unsure. If she were making $100 before, this is fantastic news, but if she was making $10,000,000 before, this raise would not really matter to her. Being aware of the percentage income increase gives you more information about how someone is impacted than being given the absolute value of the increase because percentage change in wealth or income is a better measure of happiness than absolute change. In fact, some economists acknowledge that the monetization of benefits paints an inaccurate picture. Portney makes this exact point: “…the willingness to pay for the favorable effects of a project or policy depends on the distribution of income: a billionaire would be able—and therefore willing—to pay more than a pauper for the same



improvement in environmental quality, even though both cared about it with equal intensity.”[98] However, Portney goes on to claim that despite this clear flaw “BCA [benefit-cost analysis] analysts use dollars to estimate benefits because there is simply no other way to directly measure the intensity with which people desire something.”[99] The case I am making here is not just that the first claim Portney makes is correct (absolute willingness to pay is a bad measure of benefit), but also that we can completely refute the second. There is a better way to directly measure the intensity of desire: percentage willingness to pay. Absolute willingness to pay is unjust and inaccurate, but if it is the best option available, then my arguments are no better than those offered by Anderson and Sagoff. Without data-driven, objective methods for policy analysis, bias, and corruption are inevitable (as argued in Chapters 5 and 6). However, as I will show in this chapter, a fairer and more accurate solution is possible. Percentage willingness to pay requires only a small change to methodology, but is a more just, and more accurate system, which remains objective and requires no additional resources to collect. I start by making the case that this method is capable of solving the initial issues raised for absolute willingness to pay. I then show that this new methodology is both more moral and more accurate. Utility on the Margins As I have shown, there are some ethically salient features that should be taken into account when measuring the value of a statistical life, such as individual preferences (e.g., some people would rather risk their life for a reward, others would rather stay safe). These are ethically relevant as they have an impact on overall utility. However, there are other features that are irrelevant and should be ignored when determining policy action, such as familial relationship of the decision-maker or the wealth of those impacted by the policy (saving the life of a rich person does not increase global utility any more than saving the life of a poor one). Having more wealth does not increase the ability of an individual to experience more welfare, or utility. Economists and policy makers have noted this, but it has yet to make its way into cost-benefit analysis. According to Boardman et al., “Economists generally assume declining marginal utility of money. That is, as a person’s wealth increases, each additional dollar produces smaller increase in

utility.”[100] Finland is not the only country to punish people commensurate with their income, similar policies can be found in Germany, Sweden, Austria, and France.[101] Someone making $1,500 a day does not lose the same utility from a $500 speeding ticket as someone that makes $40 a day, but either individual speeding is just as dangerous for the public. From the public perspective, percentage punishments are both more effective and more equitable. Even Sunstein notes that willingness to pay is importantly disconnected from welfare, but fails to advocate for improving cost-benefit analysis by using a quantitative measure of welfare beyond the admittedly imperfect absolute willingness to pay: “Even if the monetized costs exceed the monetized benefits, the welfare gains to workers might be higher than the welfare losses to consumers. Because welfare is the master concept, and cost-benefit analysis is an imperfect means of assessing welfare effects this point cannot be ignored.”[102] Intuitively doubling one’s income or halving it provides a similar benefit or exacts a similar cost regardless of your actual income. This is with the potential exception of individuals who are barely surviving on their existing income, as halving it might kill them. However, the utility lost by these individuals is understated to an even greater extent by absolute willingness to pay, which claims that a person with only

a dollar to their name loses the same amount of utility from losing that dollar to someone with a million dollars to their name. Put another way, percentage willingness to pay claims that taking $2.50 from someone with only $5 to their name is equivalent to taking $500,000 from someone with $1,000,000. Absolute willingness to pay equates taking everything from 100,000 people with only $5 to their name as equivalent to taking $500,000 from someone with $1,000,000. Percentage willingness to pay may underestimate the loss of $2.50 if taking that away would mean that person could starve and die, but not nearly as much as absolute willingness to pay does for the 100,000 people who lose everything.   Percentage Willingness to Pay Percentage willingness to pay, unlike absolute willingness to pay, tracks our intuitive sense of welfare and is supported by the economic principle of the declining marginal utility of money. Using a percentage willingness to pay measure of utility, instead of an absolute willingness to pay measure of utility is not only feasible, it is more equitable precisely because it is more accurate. This measure resolves the initial problem of differential values of statistical lives, while still accounting for ethically salient preferences. Before providing an argument for why this method is better, here is a brief explanation of how it works. Percentage willingness to pay is calculated as absolute willingness to pay divided by total ability to pay. So, if someone was willing to pay $500 for a benefit out of their $10,000 of wealth, their percentage willingness to pay would be 5%. Chapter 12 will provide more in-depth examples of how this method can be used to value statistical lives, value benefits/costs to firms, and more in cost-benefit analysis. While this may seem to introduce a second concept to be studied (ability to pay) expanding measurement costs and potential error, ability to pay is at least theoretically also required to study absolute willingness to pay (as willingness to pay is constrained by ability to pay). Therefore, nothing additional should need to be measured. This works to control for either income or wealth, depending on what is being used to calculate the underlying absolute willingness to pay. If one is measuring annual benefits and using annual income to determine the constraint of ability to pay, then the denominator to calculate percentage willingness to pay is an

individual’s income. However, if you are looking at total benefit and using wealth to determine the constraint of ability to pay, then wealth can serve as the denominator. Whatever value is being used as the measure of ability to pay shall be used as the denominator in determining percentage willingness to pay.   Calculating Percentages To show how this methodology can be calculated and how it compares to absolute willingness to pay, consider the following imaginary study of contingent valuation. Imagine that there are only two countries, the United States and The Gambia, and in those countries there are two types of people. In the United States there are people who make $100,000 a year and people who make $10,000 a year. In the Gambia there are people who make $1,000 a year and people who make $100 a year. In our sample we have one of each of these people who are representative of their group, call them Mike ($100,000), Sarah ($10,000), Bintou ($1,000) and Lamin ($100). All of these people are asked on a contingent valuation survey how much they would be willing to pay annually in taxes to fund a policy that would reduce risk in their country by 1 in 100,000. Imagine we then discover that all of the people in the study have identical preferences in absolute terms, they would all be willing to pay $50 to prevent this risk. Based on this, we might conclude that both countries have a value of a statistical life of $5 million. This is problematic for two reasons: first Lamin, who is willing to pay half of his income to reduce the risk, seems to be making a very different choice than Mike. It seems that someone who would give half of their income to prevent a risk is more averse to that risk than someone who would only give one two-thousandth of their income, but according to absolute willingness to pay, they both value their lives equally. Second, contingent valuation studies often vary the exact scenario presented that may do little to impact Mike, but will restrain Lamin’s ability to maintain the same statistical life in absolute terms. Say that we are conducting such a study. We go on to ask what they would be willing to pay to prevent a 1 in 10,000 risk. If they were rational, in absolute terms, they would respond that they would be willing to pay $500 and in fact, the first three do. Lamin, however, cannot be willing to pay $500 because he does not have $500, he only has $100. Because

willingness to pay is constrained by ability to pay, we are forced to factor in the irrelevant characteristic of individual wealth in the valuation of Gambian lives. So, he says that he is willing to pay $100 (because the study does not allow him to be willing to pay more). Based on these new estimates, the United States still has a value of a statistical life of $5 million, but simply based on the way we asked the question, The Gambia has been downgraded to the value of $3 million. This downgrade has absolutely nothing to do with ethically relevant factors like preference, and everything to do with the ethically irrelevant factor of wealth. Now imagine the study asked what you would be willing to pay to reduce a 1 in 1,000 risk. Once again Mike and Sarah remain internally consistent with respect to absolute willingness to pay and claim that they would each pay $5,000. However, now both Bintou and Lamin are constrained by their wealth. They are rational and internally consistent in absolute terms, so, while they would pay $5,000 if they had it, they are limited to their incomes of $1,000 and $100, respectively. Once again, we are factoring in an irrelevant characteristic into our calculations: wealth. While the American value of a statistical life remains at $5 million, the Gambian one has dropped to almost a tenth of that at $550,000. Because we are using absolute valuations instead of percentage of wealth valuations, the way that the question is asked changes the result. The above example is particularly relevant to justify the notion that willingness to pay does not perfectly map onto utility. Bintou and Lamin are willing to give 100% of their income to prevent a 1 in 1,000 risk of death, whereas Mike is only willing to pay 5% of his income to do the same. It should not follow that Bintou and Lamin value their lives less than Mike does. If anything, we should conclude that they value their lives much more since they are willing to trade away all other goods to avoid this risk, while Mike is only willing to trade a small fraction of his goods. Absolute willingness to pay fails to capture the share of your resources spent on a particular investment. This is somewhat a feature of the way that the questions are asked. But, even if we design a survey that is not limited by anyone’s wealth, such as the initial example of a 1 in 100,000 risk, Lamin is still paying 50% of his income, while Mike is paying 0.05%, but they are assessed as valuing their lives equally. The fact that Bintou and Lamin in the final example have a lower willingness to pay for risk

reduction is not an indication that they would rather spend their money on other things. Instead, it is an indication that there is literally nothing else that they would rather spend their money on. If willingness to pay is not in some way standardized to income, this issue will persist. In Chapter 12 we will look at how these contingent valuations can be used to create percentage values of statistical lives. A similar comparison can be made with wage risk studies. Imagine two labor markets. Both of them have two jobs that are identical except that one carries a 0.1% risk of death. In the American labor market, the safer job pays $20,000 a year while the riskier job pays $22,000 a year. In the Gambia, the safer job pays $750 a year while the riskier job pays $1,500 a year. Using our standard absolute willingness to pay model, we would determine that Americans value a 0.1% risk at $2,000, while Gambians value it at only $750 (leading to a much higher value of a statistical life for Americans). However, we could also frame this as the Americans only need a 10% pay increase to accommodate this risk, while the Gambian requires a 100% pay increase to accommodate the risk. This would lead to a higher value of a statistical life for Gambians, who require a larger percentage pay increase to take the risk. This means that any survey using absolute willingness to pay is a poor measure of actual utility compared to percentage willingness to pay. You lose more utility from giving up your entire income than 5% of it and you gain more utility from doubling your income than getting a 10% raise regardless of the absolute number of dollars you are giving up or gaining. If what we are trying to maximize is utility instead of absolute willingness to pay, then focusing on percentage changes in income is simply a more accurate option.   Percentage Willingness to Pay is More Accurate In the remainder of this chapter, I offer a formal argument to support the claim that using percentage willingness to pay is a more accurate measure of utility than absolute willingness to pay. In Chapter 11, I present an argument that percentage willingness to pay is in fact a more ethical and just method of computing benefits, regardless of its accuracy. The argument for accuracy is as follows:  

P5: Becoming completely destitute is equally harmful to someone who has $1,000,000 and someone who has $10,000.   P6: Increasing one’s wealth by $100 provides different levels of happiness to someone who has $10,000,000 and someone who has $10.   P7: Absolute willingness to pay claims that both P5 and P6 are false.   P8: Percentage willingness to pay claims that both P5 and P6 are true.   P9: All else equal, a methodology that claims both P5 and P6 are true is a more accurate measurement of our intuitions about the relation of monetary benefits to happiness than one that claims P5 and P6 are false.   C2: Therefore, all else equal, percentage willingness to pay is a more accurate measurement of our intuitions about the relation of monetary benefits to happiness than absolute willingness to pay.   The key premises in this argument that need to be defended are P5, P6, and P9. Premises P7, and P8 should be clear simply from the definitions of percentage and absolute willingness to pay and the conclusion follows logically from the premises. The conclusion C2 follows directly from premises P7, P8, and P9. Therefore, the remainder of this chapter will focus on defending premises P5, P6, and P9.   Percentages Measure Losses Better To elicit the intuition for P5, imagine two middle aged women, Melati and Karen. They both own their own homes, but live mostly paycheck to paycheck. In other words, their entire wealth is in their houses. They both have scraped and saved their entire lives to own those homes. They both live on small islands in the Pacific. One day, an earthquake in the Pacific causes tsunamis that destroy both of their houses, leaving them both without a home, and therefore with no wealth at all. A common intuition is that they have both suffered the same harm: they both have nowhere to live

and are equally harmed by this event. However, Melati lives in Indonesia and Karen lives in Hawaii. Because of this difference, Karen’s house is valued at $1,000,000 but Melati’s house is valued at only $10,000. If absolute willingness to pay is correct, then Karen suffered a tragedy 100 times worse than Melati did. It would have been worth destroying the houses of 98 other women like Melati if we could save the house of one Karen in terms of total happiness in the world. However, if percentage willingness to pay is correct, then they have both suffered equally (with 100% losses). They have both lost everything and we should be indifferent between saving Karen’s house and Melati’s house since they will both cause the same amount of psychological harm. My intuition is Karen and Melati have suffered the same loss, and Karen does not have more capacity to suffer, simply because the market has placed a greater value on her house. If your intuition is that both Melati and Karen have suffered equally or close to equally, then percentage willingness to pay is a more accurate description of the world. Even if you think that Karen suffered slightly more, percentage willingness to pay is more accurate if you think the difference is closer to equal than 100-fold. Karen’s house could have been valued at $10,000,000 and Melati’s house at $1,000, at which point the proponent of absolute willingness to pay would be committed to Karen’s suffering being

 

Becoming homeless does not cause the rich more psychological pain than the poor simply because their houses are worth more.  Losing your home should be valued equally no matter where you live.

10,000 times greater than Melati’s, a claim that is very hard to justify. Based on this, percentage willingness to pay is a more accurate measure of happiness in the world than absolute willingness to pay.   Percentages Measure Gains Better In defense of P6, take the following thought experiment. Imagine two men: Ron and Marco. Ron is a millionaire with $10,000,000 in net worth. He spends lavishly buying $500 meals and $10,000 suits. Marco is homeless. He scrapes to have enough money to eat every day. Marco is a master of making a few dollars last an entire month. He is caught in a poverty trap, unable to increase his wealth because he must spend everything he has to just survive. He only has ten dollars to his name. Marco and Ron are walking along the same street and there is a $100 bill in the center of the sidewalk. Who will be happier if they find it? If absolute willingness to pay is a good measure of happiness or utility, then they will both receive equal happiness from a surprise $100 windfall. However, in reality Ron will drop it all on a tip at a luxury restaurant and barely remember he found it the next day. Marco could stretch that $100 out for several months, allowing him start his journey out of a poverty trap. The point is that Marco’s increase in happiness at finding the $100 is not equal to Ron’s. That $100 could turn Marco’s life around, but Ron will have forgotten it in 24 hours. However, using percentage willingness to pay, Ron gains only 0.001%, while Marco gains 1000%. Percentage willingness to pay would place Marco’s happiness much higher than Ron’s despite their both receiving the same absolute value of money because Marco’s utility at receiving that money is much higher. Another way to think of this is to answer the question, should a government give $99 to Marco or $100 to Ron? According the absolute willingness to pay, there is no question: we should give $100 to Ron because this maximizes absolute willingness to pay. However, it should be clear that this does not maximize utility or happiness. The most accurate way to measure such benefits is to control for wealth. Percentage willingness to pay shows that Marco’s gains in finding $100 on the street are much greater than Ron’s gains, and therefore is a more accurate measure of utility in the world.

Premise 9 is slightly more challenging to defend. It is important to remember that this is not claiming that percentage willingness to pay is the most accurate method for measuring utility or happiness, but rather that, with all else equal, a method that conforms to our intuitions about happiness both in terms of gains and losses is preferable to a method that does not. If we can improve our methods without losing the benefits, we should. It is important to note that this assumes that there are no benefits that percentage willingness to pay lacks but absolute willingness to pay retains. In other words, there are not some other scenarios like premises 5 and 6 for which our intuitions indicate that in fact absolute willingness to pay is a more accurate representation of reality. In Chapter 12, I defend the claim that all else is equal with respect to operationalization as percentage willingness to pay can be used just as easily as absolute willingness to pay through an example. In Chapter 16 I examine potential objections to percentage willingness to pay, including some that attempt to find scenarios where absolute willingness to pay is more intuitive. However, even if all else is not equal, that does not defeat the argument for C2, which explicitly states that it is the case only if all else is equal. If someone is able to offer a convincing argument for why absolute willingness to pay better corresponds to our intuitions about benefit, I am happy to hear it, but I have as of yet seen no such scenario. In the next chapter, I show that, in fact, percentage willingness to pay is not just more accurate, as argued here, but also a more ethical and just measure of benefit.    

           

11. THE JUSTICE OF PERCENTAGES     In 2016, a debate was raging about the route of the Dakota Access Pipeline, a large oil pipeline planned to travel through North Dakota.[103] Oil pipelines carry mortality risks as oil can leak out and seep into drinking water or cause other mortality risks.[104] The debate centered on the exact path of the pipeline. Initially it had been planned to cross the Missouri river near the comparatively wealthy and predominately white city of Bismarck but was later rerouted to cross the river just north of the lands of the Standing Rock Sioux Tribe.[105] Native groups from around the country came to protest the pipeline. There was substantial concern at the time that this action was implicitly valuing the lives of the wealthier white population in Bismarck at a greater value than the lives of the Native Americans to the south because they were wealthy and white.[106] Proponents of the pipeline justified the move because the population of Bismarck was much larger than the reservation, but it is conceivable that even had the populations been of similar size the mortality risk and therefore the value of the poorer Native American lives would have been implicitly valued as less. This story highlights two issues central to our argument here. First, transparency is key in such analyses. One issue with the philosopher’s response of doing away with cost-benefit analysis is that such a decision framework will mean that we do value lives disparately, but that we just do it behind closed doors—and implicitly as it was suspected with regard to the case of the Dakota Access Pipeline. We need public cost-benefit analyses to show why decisions are being made, and that they are based only on ethically relevant factors. Had such an analysis been made comparing the options the public could have seen whether it was the number of people, or the privilege/wealth of the individuals that impacted the decision (and the public could have scrutinized the decision if the populations were equal, but the lives of one were valued less). Second, there are some things, such as wealth, which we should not factor into calculations of statistical lives. Doing so will inevitably lead to the lives of the poor and marginalized being valued less than the wealthy. This second

point is the focus of the current chapter. Including wealth in a calculation of the value of someone’s life is immoral, and unjust.   The Justice of the Percentage Percentage willingness to pay is not merely a more accurate way of measuring the impact of a policy, it is also more equitable and just. I defend this claim with the following argument:   P1: Using absolute willingness to pay, the value of a statistical life is mostly determined by measures of income or wealth. P3: Wealth and income are ethically irrelevant features of a population. P10: Percentage willingness to pay controls for wealth and income when determining the value of a statistical life. P11: All else equal, a process for determining the value of a statistical life that controls for ethically irrelevant factors is more just and equitable than one that does not. P12: Percentage willingness to pay is a process for determining the values of a statistical life that controls for ethically irrelevant factors. (P3, P10) P13: Absolute willingness to pay does not control for ethically irrelevant factors of a population. (P1, P3) C3: All else being equal, percentage willingness to pay is more just and equitable than absolute willingness to pay. (P11, P12, P13) The premises P1 and P3 are drawn from the argument in Chapter 9, which claimed that absolute willingness to pay is immoral. These are defended in earlier chapters. Premise 12 follows clearly from premises P3 and P10. Premise 13 follows from Premises P1 and P3 (a method that is mostly determined by wealth or income inherently does not control for them). Similarly, the conclusion follows from P11, P12, and P13. Therefore, the premises that we will defend in what follows are P10 and P11: the

claims that percentage willingness to pay does control for wealth and income, and the claim that controlling for these factors makes a method more equitable and just. This argument shows that absolute willingness to pay is not only immoral, but that percentage willingness to pay is a more just method of conducting cost-benefit analysis. Money Controls As claimed by P10, percentage willingness to pay controls for wealth and income. There are two ways that I will show that this method is successful: conceptually and empirically. Conceptually it should be clear how this is done. By dividing an individual’s absolute willingness to pay by their total ability to pay, we factor out the impact that their total ability to pay had on their willingness to pay. Instead of answering the question of how much money an individual would be willing to spend on a particular good (a question that is inherently tied to the amount of money that an individual has), this answers the question of what share of an individual’s resources they would be willing to spend on a particular good. While the first question depends on your wealth, the second question does not. Studies of percentage willingness to pay could be conducted without ever assessing an individual’s income or wealth but by simply asking what portion of what they have they were willing to spend on a particular good. This is potentially an overly simplified method that may not capture the entirety of the relationship between wealth and utility, but by controlling for wealth it does a better job than absolute willingness to pay. Therefore, conceptually not only does percentage willingness to pay clearly divide income and wealth out of the equation, it can be calculated without them, by simply asking what percentage of your resources would you spend on to gain a particular benefit. Despite conceptual strength, if it is the case that percentage values of a statistical life continue to be strongly determined by income or wealth, then we have failed to control for income in practice, even if we have controlled for it in theory (and this methodology will be subject to the same criticism of absolute willingness to pay). Fortunately, this is not the case. The absolute willingness to pay for a reduction of a 1 in 1000 risk varies from $45 to $18,261 based on Viscusi and Masterman’s estimates.[107] These show a wide range of variation driven largely by income. However, if we

control for income, these differences disappear. We do not have a perfect measure of ability to pay, and so will use gross national income per capita as the denominator. Using this method, if we calculate the percentage willingness to pay for a 1 in 1000 risk, the resulting values only vary from 17.06% of income to 17.32% of income.[108] To further justify the claim that these are not based on income, the highest value of a life is from the Democratic Republic of the Congo, one of the poorest countries in the world, while the lowest value is from Malawi, another poor country. High income countries such as the United States or Japan can be found at the middle of the distribution at 17.20%. See Appendix D for a full table of these values. Intuitively this seems much more reasonable: lives are valued very similarly with only small differences, likely driven by preferences. There may be a lingering concern that this does not value all lives exactly equally.

While the benefits of risk reduction are not perfectly identical, this may be due to differences of levels of risk tolerance or other ethically relevant factors. As argued by Sunstein, we do not want to subject individuals to bear the costs of risk prevention when those costs outweigh the benefits.[109] While this does give rise to conclusions such as we might rather invest in a program that would statistically save 66 lives of a group that is more risk averse than 65 lives of a group that is less risk averse, this seems to be much more a feature of the ethically relevant factor of their risk tolerance, than underlying income inequality. Therefore, while this method does not value lives exactly identically, it is much better than willingness to pay, and differences can be chalked up to legitimate differences in risk tolerance, which might in fact merit justifiable differences in policy. Justice and Equitability The final premise in need of defense is the one related to justice and equity. Premise 11 claims that all else being equal, a method that controls for ethically irrelevant factors is more just and equitable. There are two ways that we can defend this premise, in terms of the impact on overall utility, and through an appeal to John Rawls’s veil of ignorance (more on this below). First, in terms of overall utility, ethically irrelevant features by definition fail to impact overall utility. Therefore, including them in an analysis inaccurately tips the scales toward those features. In the case of the inclusion of wealth and income in absolute willingness to pay, this biases the conclusions in the favor of the rich at the expense of the poor. It does so by making policies that benefit the wealthy appear to provide the world with more value and policies that benefit the poor appear to provide the world with less benefit. However, this is not the only ethically irrelevant feature that should be excluded. Others such as familial relations to the decision-maker must be excluded as well to get an accurate measurement of utility (these are not included in absolute willingness to pay, but serve to demonstrate why such a provision is required). You might imagine a methodology that weights the benefits or costs of anyone related to the decision-maker, or is from the same ethnic group as the decision-maker twice as much. These disparities are ethically irrelevant and including them inaccurately represents the total utility from a decision, leading to decisions that fail to maximize utility. Absolute willingness to pay is not importantly different from such

discriminatory policies. The only difference is that it is biased in favor of the rich instead of a particular ethnic group or family. Including ethically irrelevant factors embeds bias into our methodology, whether that is bias for the relatives of a policymaker or bias for the rich. To be clear, such a policy is not merely unequitable, it is also unequal. The distinction between equity and equality dates back to the discussion of Milo the wrestler in Aristotle’s Nicomachean Ethics.[110] Equity is concerned about giving people what they need (someone who desires risk reduction more should get it). Equality is concerned about giving people the same thing (everyone should get the same amount of risk reduction). Absolute willingness to pay is the opposite of equitable, giving more to the rich than the poor (by assigning the value of life not based on preferences but based on income). By including the ethically irrelevant feature of wealth it gives more benefit to the rich and less to the poor. However, absolute willingness to pay is also unequal. Even if you were merely concerned with treating everyone equally (not with giving individuals what they needed), absolute willingness to pay falls short. It privileges those who already have more and disadvantages those who have less. It places greater value (not the same) on the lives of the rich. By assigning a greater value of life to those that have more, the methodology is deeply unequal and unequitable. Percentage willingness to pay gives people what they need because it focuses solely on risk preference not on income or wealth. By assigning the value of a statistical life on the basis of real, ethically relevant factors like

preferences, percentage willingness to pay treats people equitably. Rawls and Statistical Lives Another way to frame this is in terms of justice. In his famous work A Theory of Justice, John Rawls argues for stepping behind the veil of ignorance when determining which polices to enact.[111] In other words, we should make polices imagining that we do not know what our demographics will be, whether we will be young or old, rich or poor, male or female, Black or White. This framework is helpful when determining which features of a population are ethically relevant. In addition, it demonstrates that ethically irrelevant features of a population are unjust to take into account when determining which policies to enact. Therefore, this theory provides support both for the claim that wealth and income are ethically

irrelevant (P3), as well as the claim that ethically irrelevant features should not be used in cost-benefit analysis (P11). To elicit this intuition, imagine for a moment that you are behind the veil of ignorance. You do not know what country you will live in, or what your preference for mortality risk will be (whether you are someone that is willing to take risks for monetary benefit or someone that would pay anything to avoid a risk). Now imagine that you are deciding which criteria to use to assess a policy that impacts the mortality risk of certain populations. It seems like preferences of the population make sense to include: a particularly risk-averse population may deserve a policy intervention more than a risk-seeking one that would value it less. As you don’t know if you are risk averse or not, it makes sense to support preferences as a just way to determine policies. If you were to be risk averse, you would want there to be a lower threshold for enacting the policy because you value risk reduction highly. If you were to be risk seeking, you would not be willing to pay as much to have a risk-reduction policy enacted. In either case you would support risk preference would be included in such an analysis. However, what of wealth? Should the wealth of a community be used in determining whether that community deserves a reduction in mortality risk? You do not know whether you will be rich or poor, and this will not impact your feelings about risk. If you were wealthy, you might prefer that wealth be taken into account, as this would give your preferences more weight. However, if you were poor, you would not want the wealthy’s preferences arbitrarily given more weight, as their wealth does not impact their risk tolerance or their ability to experience happiness. Therefore, you would likely conclude that wealth is an ethically irrelevant feature of a population and that policy decisions should not be made on the basis of it. In fact, I would expect that you would think choosing to value the lives of those with more money greater is unjust to those with less. When you don’t know if you will be rich or poor, wealth is not something that most people would think should feature into a society’s decision about the value of your life. There are two important features of this argument and the argument from the previous chapter to note. First, they only work if all other features of these methods are equal. If it is impossible to actually implement a percentage-willingness-to-pay analysis, then we might still prefer absolute

willingness to pay. Additionally, they are both comparative, it is possible that there are more accurate or equitable ways to measure benefits and costs than percentage willingness to pay. I am in no way convinced that percentage willingness to pay is the best solution for controlling for wealth (I suspect there may be better ones). However, percentage willingness to pay is an improvement on the current system of using absolute willingness to pay in terms of both accuracy and equity. The following chapter will show that this method is as effective to conduct real cost-benefit analyses as absolute willingness to pay. In other words, all other features of the methods are equal.

 

12. UNEQUALLAND (AN EXAMPLE)   In this chapter, I will tell you several stories about a place called Unequalland to showcase exactly how percentage willingness to pay could be operationalized. This land is not dissimilar to many countries that we see today that have yawning gaps between the rich and the poor. The demographics are simplified to showcase how these principles could be applied, but overall, there is nothing in principle that prevents these examples from fitting real countries. The final standard that percentage willingness to pay must meet is the ability for it to actually be used as a tool for policy makers to decide between different allocations of scare resources. To do this, we first need a definition of what it would mean to get a percentage of value as opposed to a dollar value of value. Then we will use several detailed examples to show that, not only can we construct values of a statistical life using percentage willingness to pay, we can conduct complete cost-benefit analyses to support data-driven, unbiased decision-making. The Meaning of Percentages One concern with shifting from dollars to percentages is that these will no longer be easily comprehensible by individuals either completing surveys or viewing the data presented. However, I think there is a strong case that these are in fact much more intuitive than specific dollar values. In absolute cost-benefit analysis, the final values are generally in dollars; a policy would create a net present benefit of $10,000 or a net present cost of $5,000. Since percentage willingness to pay is calculated in percentages instead of dollars, our units need to change. The units may be somewhat confusing, but one might think of it this way, a value of 100% means that one person just got the happiness of doubling their wealth. A value of -100% means that one person just became completely broke. Larger values mean that more people had these same benefits or costs. A benefit of 500% is equivalent to five people doubling their wealth or one person quintupling their wealth. A cost of -350% is equivalent to three people losing everything and one person losing half of their wealth. We could also

imagine this as 350 people losing 1% of their wealth, but because this is intuitively harder to grasp, I will focus on generally normalizing these to 100% (treating 500% as five people doubling their wealth, and -350% as three people losing everything and one person losing half of their wealth). If you want to understand how much happiness a 100% total gain provides, just imagine that you woke up tomorrow and had twice as much money in the bank as you do today, or that your salary had doubled overnight. That is the amount of happiness for each 100% increase. To aggregate benefits and costs, instead of adding up the absolute change in the values of willingness to pay, you sum the percentage willingness to pay of available resources for each individual or demographic group. If this value is positive, the policy is justified as increasing overall utility, if negative it decreases overall utility. This method retains the majority of the framework of cost-benefit analysis with only the slight change of the value to be maximized. So, if our policy only impacts two people, one that would receive a net loss of 5% of their wealth and one that would receive a net gain of 100%, we would claim such a policy would have a net benefit of 95%, or nearly doubling someone’s wealth. While we will not specifically cover the process of discounting future benefits here, such calculations could easily be made by discounting the dollar benefits first, then dividing by total wealth. That said there are a range of issues with the very concept of discounting (is saving a life 100 years from now really worth next to nothing?) that will need to await another text, though many of these issues have been addressed in the literature on environmental ethics.[112] It may be the case that a truly ethical cost-benefit analysis does not discount future lives saved, or does so in a drastically different way. Either way, the subject of discounting will be set aside for now in interest of supporting the central claim around the operationalizability of percentage willingness to pay. The following examples serve to demonstrate both the way in which this method can be viably used, but they also show that we can retain valuing statistical lives and disaggregating statistical life estimates by subpopulations while avoiding the conclusion that a single American is worth more than over 100 Gambians. They also showcase that the problem at the heart of this unethical conclusion is the use of absolute willingness to pay instead of percentage willingness to pay.

Millionaire’s Peak For our first example, imagine a small town of 1,000 people in Unequalland called Millionaire’s Peak. In Millionaire’s Peak there are exactly two types of people, people with $1,000,000 of wealth (the millionaires), and people with $10,000 of wealth (the working class). The millionaires make up 10% of the population while the working class makes up 90%. The city council is considering whether to implement a one-time tax to fund a community park. The tax would cost everyone 1% of their wealth this year. The city council has done a contingent valuation study and found that everyone in the community would be willing to pay $500 for such a park on average. Using absolute willingness to pay, we find that each millionaire has a cost of $9,500 for such a policy (they are taxed $10,000 but get $500 of benefit). Members of the working class each experience a benefit of $400 from such a policy (they pay $100 in taxes but receive $500 of benefit). Since there are 1,000 people in the society (100 millionaires and 900 working class people), the total net cost would be $590,000,[113] and it would only become more negative if Millionaire’s Peak were larger (assuming the class proportions remained constant). Despite 90% of the population receiving a sizable net benefit from this policy, absolute willingness to pay finds it unethical because absolute willingness to pay values the suffering of the rich more than the happiness of the poor. However, using percentage willingness to pay, we find a net cost of only 0.95% for millionaires ($9,500/$1,000,000), but a benefit of 4% for each of the members of the working class ($400/$10,000). With

 

Absolute WTP overvalues the costs and benefits to the rich while undervaluing the costs and benefits to the poor, further entrenching systemic inequalities.

100 millionaires and 900 working class people, the total net benefit would be 3505%[114] (or the happiness equivalent to about 35 people doubling their wealth). While this policy does not maximize absolute willingness to pay, it maximizes percentage willingness to pay. Using absolute willingness to pay maximizes the benefits to an already privileged minority, while using percentage willingness to pay maximizes happiness and utility in the whole society and measures benefits more accurately.   The Life of a Millionaire Before applying these to a scenario with statistical lives, let’s discuss how we might arrive at disaggregated values of statistical lives for Millionaire’s Peak, both in absolute terms and percentage terms. Imagine that we conduct a wage risk study disaggregated by wealth. We find that to accept a job with a 1 in 10,000 risk of death, millionaires need to be paid an additional $17,000. Using this we calculate that they have the value of a statistical life of $170 million. This means that a 1 in 10,000 risk of death leads to a 1.7% cost, or millionaires are willing to pay 1.7% of their wealth to avoid a 1 in 10,000 risk of death ($17,000/$1,000,000=1.7%). We also survey the working class and find that they need to be paid an additional $170 to accept the same risk. As we found with the similar country valuations shown in Appendix D, in percentage terms this is almost identical to the millionaire’s valuation: the working class would be willing to pay 1.7% of their wealth to avoid a 1 in 10,000 risk of death ($170/$10,000 = 1.7%). Also surveying the working class we discover that they are fairly consistent in percentage terms, being willing to pay $1,700 (or 17% of their wealth) to avoid a 1 in 1,000 risk. With this information in hand, let us examine another policy under consideration in Millionaire’s Peak. Millionaire’s Peak, like many cities, is starkly divided by income. All of the millionaires live in one part of town, and all of the working class live in another part of town. The coal plant located in the upper-class part of town was recently damaged by an earthquake. The city council is considering whether to rebuild in the same location or move a coal plant from the upper-class part of town to the working-class part of town.[115] Coal plants are harmful to the air quality of communities around them and increase mortality risk. Moving the plant would reduce the annual risk of death for the millionaires by 1 in 10,000,

but increases the annual risk of death for the working class by 1 in 1,000 (the working-class live in apartment buildings close together that have worse ventilation and so are more impacted by the pollution). Using the absolute willingness to pay, we can calculate that the millionaires are willing to pay $17,000 each to avoid the 1 in 10,000 risk, while the working class is willing to pay $1,700 to prevent the 1 in 1,000 risk. In a community of 1,000, moving the coal plant to the poorer part of town means the total benefit to the 100 millionaires would be $1,700,000[116] while the total cost to the working class would be $1,530,000.[117] Based on this analysis, this policy should be enacted, as the benefits outweigh the risks. If this risk was annual, after 100 years we would be condoning a policy that would likely kill 90 working class people to save the life of one millionaire. This is not as extreme a conclusion to our original dilemma, but it is still quite concerning because it values the lives of the millionaires more simply because they are rich. If we use the percentage willingness to pay instead of the absolute willingness to pay to calculate benefits, we can alleviate this concern because the math requires we always implement the policy that saves the most lives if they have equal monetary costs and percentage risk preferences are the same. In this case, we find that everyone is willing to pay 17% of their wealth to prevent a 1 in 1,000 risk and 1.7% of their wealth to prevent a 1 in 10,000 risk. Implementing this policy gains the 100 millionaires 170% benefit,[118] but costs the working class 15,300%,[119] for a total net cost of 15,130%, equivalent to 151 members of the society going completely broke. Once again, using percentage willingness to pay in place of absolute willingness to pay allows us to make choices that maximize happiness and are more intuitive than polices that maximize absolute willingness to pay, and condone counterintuitive choices that prioritize the lives of the rich few over the lives of the many poor folks.   Government and Corporate Costs and Benefits Millionaire’s Peak offers a simple example of how percentage willingness to pay could be used and how it could impact decision making. This, however does not address some more complicated aspects of applying percentage willingness to pay to cost-benefit analysis such as cost borne by the government and costs borne by corporations (both in terms of profits

and job losses). Rarely do governments implement polices that are free and have no impact on the bottom lines of companies. The following example of Bell Curve Valley provides a more complex and detailed explanation of how percentage willingness to pay can handle such situation.   Before diving into the example, here is a brief explanation of one method to using percentage willingness to pay in these situations. As with the initial defense of percentage willingness to pay itself, I am not advocating for the claim that is the only way or the best way to value these impacts. Instead, I am suggesting that it appears to be a much better method than absolute willingness to pay. The point of the following method and example is not to definitively establish the only way to measure costs and benefits on governments and corporations while controlling for wealth, but rather to short circuit the objection that percentage willingness to pay cannot be operationalized, or cannot measure costs to firms and governments. First, government costs may pose a challenge since there is no clear denominator to divide the cost by. Simply counting the government as an individual with a certain amount of wealth will drastically underestimate the costs borne by the public when the government incurs costs. In order to address this, we need to find the people who pay the price when a government incurs a cost. The answer is simple: the taxpayer. Most taxpayers are not concerned about government expenses out of a belief that these expenses are inherently bad, but rather they are concerned because they will eventually be paying that bill. In order to determine how a given expense impacts people, we need to see how that expense is funded and how much people pay in taxes to fund that project. For example, take the small Unequalland town of Dirtsville. If the Dirtsville city council funded a road improvement by leveeing a tax of $100 on each of its citizens, the cost of that project is the decrease in wealth for all citizens. If all 100 citizens have $10,000 in assets, they each bear a 1% cost, for a total cost of 100%. If the road does not provide 100% in benefits to those citizens, it is not justified by percentage cost-benefit analysis. This methodology recognizes that governments themselves do not have utility or happiness, and so cannot truly be said to be bearing a burden. The people who pay taxes that fund that government do have utility, and it is their

utility with which we should be concerned when we concern ourselves with government costs. One might be concerned that governments (particularly federal governments) often incur costs without a specific program to fund them. There are many possible ways to value such costs. One would be to distribute the cost over the existing tax brackets (proportional to the amount that they contribute to government revenue). I will not dwell on the exact method for every situation because this will likely depend on the exact policy and how it is being financed, beyond saying that there are ways that this could be done. Corporate costs and benefits can take multiple forms, but the same principle observed above will apply: what matters is not the money; what matters is the people. When a corporation bears a cost from a policy, we can judge its true percentage cost in two ways: in terms of the decrease in wealth of the shareholders and owners of that corporation and in terms of the individuals who may gain or lose jobs or see their salaries decline. Once again, we measure these costs against the actual wealth of these individuals. A full argument defending this methodology and looking briefly at questions of the rights of corporations can be found in Chapter 13. For example, imagine the government of Unequalland implements some new regulations that impacts a particular company, BanksCorp. We can measure the percentage impact of that policy with some basic information about its stockholders and workers. Say the policy led to BanksCorp’s share price dropping $10, and the corporation laying off 10 workers making $100,000 on average. If we found that the corporation’s 100 stockholders had $1,000,000 in assets and owned 150 shares each, we could compute this cost as $1,500 or 0.15% for each stockholder, or 15% in total. The cost of the layoffs could be computed in terms of losing 100% income for all 10 workers, or a 1,000% cost (though this would mix income and wealth) or calculate how long individuals in such jobs are generally out of the job market and use those lost wages as the numerator with their wealth as the denominator. We could then calculate the total benefit of these regulations to determine if the benefits are worth the costs. Note that the more detailed information we have about demographic groups in terms of wealth/income, the more accurate our estimates can be, though any percentage estimates will better track utility than absolute estimates.

  Bell Curve Valley In order to see how all of these calculations come together, take the following example where firms and governments are involved as well. This example is purposefully quite complicated, both to show how percentage willingness to pay is able to address a wide range of situations (government expenditures, stock impacts, layoffs), but also to showcase the differences between how absolute and percentage willingness to pay treat costs for the wealthy, like stock market fluctuations, and costs for the poor like layoffs. Appendix E has a full list of all of the values in tables for easy reference. Imagine a city in Unequalland called Bell Curve Valley. Bell Curve Valley has a population of 100,000. It has three main neighborhoods: a wealthy suburb, a downtown, and an industrial area. The main industry in Bell Curve Valley is a chemical plant “Bell Chemicals” located in the industrial area of town. One of the processes that Bell Chemicals uses currently produces toxins that are dangerous for humans. These toxins make it so that anyone living in the industrial district has a 1 in 1,000 annual chance of dying. The city council is currently considering a law that would ban the process that Bell Chemicals uses that creates the toxin. Passage of this law would have several impacts. People living in the industrial district would no longer have a 1 in 1,000 risk of dying from toxic exposure. Bell Chemicals would lay off 1,000 low-income employees, who each make $10,000 a year and would likely not find work again for a year (therefore losing out on $10,000 of wealth). The share price for Bell Chemicals would drop $100. In order to enforce this regulation, the law would create a new regulatory body that would be funded through increased property taxes on homeowners by 10% for one year. To avoid questions of discounting we look only at the impacts of this over one year. The city council has conducted a contingent valuation study of the population to determine their willingness to pay for a reduction in a 1 in 1,000 mortality risk. They found that the residents would generally sacrifice around 17% of their wealth (similar to the results in Appendix D) in order to eliminate such a risk ($170,000 for people with $1,000,000; $17,000 for people with $100,000; etc.). Before examining the specific numbers, what do you feel are ethically the most significant impacts of this policy? Is

losing your job or dying worse than your stock portfolio dipping or your taxes going up slightly? I think most people would say the reduction in mortality and the job losses are ethically more important than the stock market and the taxes, but you might have a different intuition. As we will see, percentage willingness to pay takes this ethical intuition into account, while absolute willingness to pay confuses the stock market with the wider economy of the labor market and treats the lives of the poor as less important than the pocketbooks of the rich. In order to fully see the impact of this policy we will look at 10 residents of Bell Curve Valley, each of whom is representative of a segment of the population in how they are impacted. First there are four types of residents that live in the suburbs, which we will represent with Alice, Brahma, Carlos and DeShawn. They are distinguished only by their wealth and whether or not they own stock in Bell Chemicals. None are impacted by the toxins, so receive no benefit but they will all bear some cost of the policy. Alice lives in the suburbs (and so is unaffected by the toxins) and owns 500 shares of Bell Chemicals. If the policy was passed, her shares in the company would decline in value by $100 each. She has an overall wealth of around $1,000,000. She owns her own home and would pay an additional $10,000 in new property taxes due to this policy. Overall, she would incur $60,000 of costs if this policy was passed, or 6% of her wealth. Alice is representative of 1,000 residents of Bell Curve Valley, who would all bear similar costs. Brahma is also a wealthy suburban homeowner, but he does not have any stock in Bell Chemicals. He has $1,000,000 in wealth, and would pay $10,000 in additional taxes but otherwise would see no direct impacts of the policy. Therefore, Brahma bears $10,000 in net costs, or 1% of his wealth. Brahma’s situation is representative of the remaining wealthy population of Bell Curve Valley: around 19,000 residents. Carlos is a middle-class resident of the suburbs with around $100,000 of wealth. He owns a home and 10 shares of Bell Chemicals. If the law was passed, he would bear $1,000 in additional taxes and his shares would decline by $100 each. He would not see any other impacts of the policy, and therefore would bear $2,000 in costs or 2% of his total wealth. Carlos’s situation is representative of 4,000 residents of Bell Curve Valley.

DeShawn is also a middle-class resident of the suburbs. He owns his own home, but has no shares in Bell Chemicals. His total wealth is $100,000. If the law was passed, he would bear $1,000 in additional taxes this year, but would see no other effects. Therefore, he would bear $1,000 in costs or 1% of his income. DeShawn’s situation represents the situation of the remaining residents of the suburbs: 26,000 people. Now we move on to the residents of downtown. They make up just under a third of Bell Curve Valley’s population, but are fairly homogenous. They are all represented here by Emily. Emily is middle class with a wealth of $100,000. She does not own stock in Bell Chemicals. She rents an apartment and so will not see her taxes increase. Overall, her life is completely unaffected by the passage of the law, meaning her net benefit is $0, or 0% of her wealth. Her situation is representative of 30,000 residents of Bell Curve Valley. This may seem odd, but in many cost-benefit analyses, a subset of residents is completely unaffected. Finally, we move to the residents of the industrial area of town, containing only one-fifth of the population, represented here by Farrad, Gin, Hopi, Ingrid, and Juan. These residents all have a 1 in 1,000 risk of death from the toxins if the law is not passed and so this will be where all of the benefits are located. They are separated into five categories by their wealth and relationship to Bell Chemicals. Farrad is a middle-class homeowner living in the industrial area of town with no stock in Bell Chemicals. He has $100,000 in wealth. He would pay $1,000 in taxes if the policy passed, but would also receive benefits totaling $17,000 based on our contingent valuation study from a reduction in mortality risk. Overall, Farrad will receive a net benefit of $16,000, or 16% of his income. There are 5,000 people in Bell Curve Valley who are in a similar situation to Farrad. Gin represents the last of the middle-class residents of Bell Curve Valley. She is identical to Farrad but rents instead of owning a home, and therefore does not pay the property tax. She has $100,000, and would receive $17,000 in benefit from the reduction in mortality risk. She therefore receives $17,000 of benefit or 17% of her wealth. Gin’s situation is representative of 5,000 residents of Bell Curve Valley. Hopi is one of the low-income workers who would be laid off if the law was enacted. He has $10,000 in wealth and we will value his being laid off

as a complete loss of that wealth (a cost of $10,000). He owns a small apartment and would pay an additional $100 in property taxes. He owns one share of Bell Chemicals stock, meaning he bears an additional $100 cost. However, because he lives in the industrial area, he also gains a $1,700 (17% of his wealth) benefit from the 1 in 1,000 reduction in mortality. Overall, he bears a net cost of $8,500 or 85% of his wealth. He is representative of all 1,000 workers who will be laid off if the new regulation is implemented. Ingrid is also a low-income apartment owner in the industrial district. She does not work for or own stock in Bell Chemicals and so will not be laid off or see her stock value drop if the policy is enacted. As a homeowner with a wealth of $10,000, she will pay the additional $100 in property taxes. As she lives in the industrial district she will receive the 1 in 1,000 reduction in morality risk for a benefit of $1,700 or 17% of her wealth. Overall, her benefit from this policy would be $1,600, or 16% of her wealth. Ingrid’s situation is representative of 4,000 members of the community. Juan is our final example. He represents the final 5,000 residents of Bell Curve Valley. He rents a home in the industrial district. He is therefore not impacted by the tax but will receive a reduction in mortality risk. He has $10,000 in wealth and would get $1,700 in benefit from the mortality reduction, or 17% of his wealth. No other elements of the policy impact him. Overall, his benefit is the full $1,700, or 17% of his wealth.   Bell Valley Winners and Losers In order to see which policy is recommended by absolute and percentage cost-benefit analysis, we need to sum the benefits and costs of all the residents of Bell Curve Valley. In order to get a sense of how these benefits accrue by wealth, I will first sum across wealth levels, then for the society as a whole. Wealthy residents (with $1,000,000 in assets) bear the brunt of the costs for this policy in absolute terms but not in percentage terms. People like Alice bear an absolute cost of $60,000 each, or $60,000,000 collectively, while people like Brahma bear an absolute cost of $10,000 each, or $190,000,000 collectively. Overall, the wealthy residents bear $250,000,000 in costs. In percentage terms these costs are much smaller. People like Alice bear a cost of only 6% of their wealth, for a collective

cost of 6,000%. People like Brahma each incur a 1% cost, or a cost of 19,000% collectively. Overall, the 20,000 wealthy members of Bell Curve Valley bear a percentage cost of 25,000% equivalent to 250 of them losing everything. If you think that the most important impacts of this policy are raising taxes for millionaires and driving the stock of a polluting company down, then you might think absolute willingness to pay captures the impacts well. Middle class individuals have a more mixed picture: Carlos and DeShawn face net costs, with people like Carlos facing a net cost of $2,000 each, or $8,000,000 collectively, and people like DeShawn facing a net cost of $1,000, or $26,000,000 collectively. In percentage terms, people like Carlos face a 2% cost individually but an 8,000% cost collectively. People like DeShawn face a 1% cost each and a 26,000% cost collectively. It is notable that despite including 10,000 more people, this group’s absolute costs are valued at nearly an order of magnitude less than the last group ($32,000,000 vs. $250,000,000 in cost) but are just slightly more in percentage terms (32,000% vs 25,000% in cost). In part this is because a percentage tax has a larger absolute effect on the wealthy but an equal percentage impact. If you think that people’s views shouldn’t count more because they have more money, then percentage willingness to pay better captures your intuition. Other middle-class folks like Gin and Farrad receive net benefits. People like Gin receive $17,000 in benefits each, or $85,000,000 collectively. People like Farrad receive $16,000 in net benefits, or $80,000,000 collectively. In percentage terms these benefits are valued more (when compared to the costs for the wealthiest) since they represent a higher share of wealth. People like Gin get 17% benefit each, or an 85,000% benefit collectively, while people like Farrad get a 16% benefit, or 80,000% collectively. Including Emily (who receives neither benefit nor cost) DeShawn, and Carlos, the net benefit to the middle class is $131,000,000 or 131,000%, equivalent to 1,310 people getting a raise that doubles their income. For the lower class, the results are also mixed. People like Hopi bears a net cost despite the health benefits due to the job losses, losing $8,500 per person, or $8,500,0000 collectively. This translates into substantial losses in percentage terms at 85% each, or 85,000% collectively, the greatest

percentage loss of any group, despite being one of the smallest groups. If your intuition was that job losses were a more important ill to capture than tax increases or stock declines, then percentage willingness to pay does the best job of giving these impacts the weight they deserve. Both Ingrid and Juan experience substantial benefit from these policies. People like Ingrid have $1,600, or 16% per person, in net benefits for a collective benefit of $6,400,000, or 64,000%. People like Juan get a $1,700, or 17% benefit for a collective benefit of $8,500,000, or 85,000%. In both calculations, the benefits for people like Juan cancel out the cost borne by people like Hopi and the lower class has a net benefit of $6,400,000 or 64,000%, equivalent to 640 people getting a raise that doubles their income. Since Juan and Hopi have the same income, their percentage and absolute calculations are the same. Before adding up these final totals to see the fate of the regulation, there are two important comparisons we should make. First, look at Alice and Hopi. Both of

these groups are the same size, and both of their costs are driven by the impact of the policy on Bell Chemicals. However, Alice’s impact is driven by the drop in share price (being a wealthy individual and owning a lot of shares), while Hopi’s is driven by job loss. Which action causes more real harm in the world, a company’s share price declining or people losing their jobs because of that decline? My intuition is that the job losses have more real impact since Alice may not sell those shares until they recover and therefore never “realize the loss,” but Hopi will be directly and immediately impacted by losing his job, possibly for years to come. We would hope that our cost-benefit tool would recognize this distinction. Percentage willingness to pay does. Alice loses only 6% of her wealth due to the policy, while Hopi loses 85% of his. Absolute willingness to pay does not. Alice’s costs are $60,000, while Hopi’s are only $8,500. Second, compare Gin and Juan. They are in identical situations; the only difference is their wealth. They both receive the same benefit, a reduction in mortality risk, and bear no cost. Why then are Gin’s benefits valued at $17,000 each but Juan’s are only valued at $1,700? Because according to absolute cost-benefit analysis, the lives of the poor are simply worth less. Percentage willingness to pay does not make such an inaccurate and immoral error. It classifies both as receiving a 17% benefit because wealth is irrelevant to which life you should save. Overall, as we might expect, absolute cost-benefit analysis suggests that we not implement the policy because the costs are too high (a net present value of negative $112,600,000). Note that these costs are primarily driven by the costs to the few, privileged wealthy individuals, despite such a policy saving 20 poor and middle-class lives. It is also worthy of note that if only one rich life was saved by such a policy, then suddenly it would be worth it, as a rich life is worth $170,000,000, the same as 100 poor lives. Percentage cost-benefit analysis, on the other hand, would suggest we should implement such a policy because the benefits outweigh the costs at 170,000% net benefits, equivalent to 1,700 people doubling their income. Percentage willingness to pay does not claim that rich lives are worth more than poor lives, so if we saved 20 rich lives instead of 20 middle class and poor lives, the final benefits would be the same. The group that is shown to bear the largest cost is not the wealthy stockholders whose portfolio

declines a few percentage points, but the people who lose their entire income when they lose their jobs. Percentage Impact Using percent willingness to pay isolates the actual preferences from the individual’s station in life. It does not make the assumption that individuals with more wealth are capable of greater happiness or utility than poor individuals. As demonstrated, we do not need to be aware of the individual wealth or preferences of any individual; having just the broad demographic groups will suffice. This example demonstrates that we can account for costs to companies and the government as well as benefits to individuals. That is because the government is made up of taxpayers, and companies are made up of shareholders. Governments and companies do not have happiness or utility, people do. By focusing on percentages instead of absolute benefits, we maximize utility. Changing to a percentage measure of willingness to pay is a small and simple change to the current methods of cost-benefit analysis. It is something that is clearly measurable and would take only a small amount of information in addition to absolute willingness to pay, namely ability to pay. As Sunstein noted with the disaggregated value of a statistical life, it is unlikely that we can get individual estimates for this, but we can get estimates for demographic and regional groups. This basic change will not only solve the disturbing problem of differential cross-country values of a statistical life, but it will provide us with a more accurate while still useful indicator of overall utility in a society. Such a shift in practice would lead to the recommendation of more equitable policies because it is a more accurate representation of actual utility and because it does not include the irrelevant factor of wealth. Summer’s memo raised a concern not merely for the valuation of statistical lives in developing countries but for the basic principles of economics that give more worth to the wealthy than the poor. This method would equalize the playing field and only use ethically relevant factors, such as preferences to determine policies. Using absolute willingness to pay, it makes sense to dump toxic waste in developing countries because their lives are worth less. Using percentage willingness to pay, such gross disparities disappear without sacrificing accuracy or preference.

 

13. THE LIE AT THE HEART OF ECONOMICS   The concerns we have explored here represent more than a minor issue relegated to the valuation of life. This concern about absolute willingness to pay represents something much deeper. It is a failing at the heart of economics to accurately measure welfare. This failing systematically prioritizes the needs of the wealthy over the needs of the poor. It is baked into the basic assumptions of economics, that the needs of those who can pay more are worth more. Economics is not merely the dismal science, it is the immoral science because it uses mathematics to hide a deeply biased and political assumption resting at its foundation: that poor people don’t matter, only rich people do. However, unlike many philosophers who would throw the economics

department out with the bathwater, I think it can be reformed, though it may take some effort. In this chapter I make the case that the problems exposed by the differential valuing of lives are just the tip of the iceberg but that a small change in methodology can remedy this grave and systematic discrimination against the poor that pervades the field of economics. There is a strong theoretical case of how our current economic theories have inaccurately measured the benefits to the rich at the expense of the poor due to their use of absolute instead of percentage measures that I outline here. The Surplus is a Lie Let me be perfectly clear. My qualm is with the implicit or explicit claims of some economists that consumer or producer surplus is something that is a useful measure of benefit, welfare, or something that a government should strive to maximize. In fact, surplus is a fabricated measure that serves only to obscure the costs and benefits to the poor while overstating the costs and benefits to the rich. I am not arguing that economics fails to measure surplus correctly, but rather that surplus is not something we should care about measuring or maximizing, at least not from the public perspective. To be clear, I am not claiming that individual firms should stop attempting to maximize profit and producer surplus through technological innovation and

enhancements in productivity and efficiency. Firms are not concerned with overall welfare, nor do they claim to be. Rather, I am arguing that public institutions, such as governments, should not care about surplus because it fails to capture anything of benefit. They should care instead about real public welfare, which can be better measured than the blunt and biased tool of absolute willingness to pay currently allows. I also want to make clear that I am arguing for accuracy in our measurements, not a particular political agenda. I am not saying that benefits that accrue to the poor should be valued more than benefits that accrue to the rich. I am advocating for benefits being normalized to a more accurate description of human experience, not arbitrary currency. I am claiming that the wealthy do not have a greater capacity for happiness and suffering simply because they are wealthy. The proponents of absolute willingness to pay as a measure of welfare would have you believe that for every dollar you make, you gain a physiological ability to feel more pain and more pleasure. I am arguing that all people’s pain and happiness should count the same, regardless of how many dollars they have in their bank account. This methodology will correct the inaccurate and immoral skewing of policies to favor the rich, but it will not create policies that are biased toward the poor, just polices that treat all people as equals. In order to understand this argument, we must return briefly to the basics of economics: supply and demand curves. For the uninitiated, the yaxis of the graph represents the price of goods, while the x-axis represents the quantity exchanged in a market. The supply curve represents the amount of goods that firms in the market are able to produce as a given price. Generally, the higher the price, the higher the quantity of goods firms can produce, so this curve is generally upward sloping. The demand curve represents the consumer willingness to pay for the goods. Consumers have different amounts of money to spend on goods and different preferences of which goods they want to spend money on, and so are willing to pay different amounts for a good. Generally, the lower the price, the more consumers can afford the particular good, or would be willing to buy it, so this curve generally slopes downward. Where these two curves meet is the equilibrium, where the firm could not make more profit by selling more units, and no more consumers can join the market without lowering the

price. This is usually where the price is set, and the corresponding amount of goods is purchased. Once the market has arrived at an equilibrium, economists use this graph to measure the consumer and producer surplus. Producer surplus is the area above the supply curve (that curve representing the cost of creating the goods) and below the price. In other words, the profit. Consumer surplus is the area under the demand curve and above the price. In other words, the absolute amount that consumers would be willing to pay above the price that they actually pay. Economists

According to economists, surplus is a good measure of benefit. However, this is flawed for two reasons. Consumer surplus is measured in terms of absolute WTP. Those at the left of the graph have higher WTP (and are often wealthier) and so are counted as receiving more benefit (despite this difference being largely driven by wealth not psychological benefit). Producer surplus is profit, often captured by only one individual, yet their benefit is often larger than all the consumer benefits combined. Owning a company does not make one capable of more happiness.

argue that we should attempt to maximize total surplus because it is at least a proxy for welfare or happiness. In this chapter I make the case that, not only are both surpluses meaningless to a public organization attempting to maximize welfare, they create the appearance of a justification for policies that favor the rich even though no such justification exists. I am concerned both with the supposed measurement of consumer absolute willingness to pay, a deeply problematic concept, as we have seen throughout this book, as well as the assertion that producer surplus is ever something that policymakers should concern themselves with. The concerns arising from producer surplus are more straightforward, so I tackle these first. However, the concerns arising from conceptions of consumer surplus are much more pernicious as I show.   Corporations Can’t Be Happy In philosophy, moral status or moral standing means that something has its own moral interests; it is the kind of thing that can be harmed or wronged. [120] Many debates in ethics surround questions of whether fetuses or animals have moral status (or if they only have derivative moral status, i.e., they only have status because someone with moral status cares about them). A number of philosophers have argued that businesses and corporations are not considered the kinds of things that have moral status on their own (in the sense of moral consideration, not in the sense of moral responsibility, i.e., few contend that corporations have rights, but many claim that corporations have responsibilities).[121] We might care about them because they have impacts on people and people have moral status, but on their own, it seems intuitive to claim that businesses don’t have moral status. They can’t think or feel, they don’t have rationality or experience pain. If a business stopped existing, but everyone that benefitted from that business kept receiving those benefits (wages, profit, goods, etc.), I doubt we would say a wrong has been done. When one business merges with another, few think that an entity with moral status has been killed. The only reason a business matters is because of the people that it impacts. Even if corporations were people with rights, at best they would have equal rights to any other person, not more rights just because they have more money, meaning once again percentage willingness to pay is more justified.

With this in mind, the concern with measuring “producer surplus” should become clear. A business running a profit does not translate directly into someone being happy, and two different businesses making the same amount of profit certainly does not translate to the same amount of good or utility in the world. The problem with the current conception of producer surplus is that all firm profits are treated the same, when they in fact have very different impacts on the welfare of the world. To understand this, imagine two department stores: Mallmart and Costgo. Mallmart is owned entirely by one man, John. John gets all of the profits from Mallmart’s sales, and must pay when Mallmart is running at a loss. If Mallmart makes $100,000,000,000 in net profit a given year, all of that profit goes to John. Now take another department store, Costgo, which is publicly traded and owned largely by its employees. All employees of Costgo get stock options, and while some individuals have more stock and therefore gain a greater share of the profits, for the most part the profits are well-distributed across Costgo’s 200,000 employees and 50,000 private shareholders. When Costgo makes $100,000,000,000, that money is shared across all of its employees and shareholders in the form of dividends, benefiting 250,000 people with an average of $400,000 in wealth. According to our current economic models, this producer surplus is identical. The same amount of benefit or happiness is created by Mallmart as it is by Costgo. However, as we have seen, this is clearly not the case. John does not have the ability to be made happier than the total happiness of 250,000 people simply because he has more money. Having a bigger bank account does not impact your biological ability to feel happiness or sadness. Producer surplus is therefore a poor measure of benefit to society that is deeply biased toward the rich, treating them as if they have greater capacity for happiness and suffering. This is not to say the companies should not try to maximize their profits. It is merely to say that companies whose shares are more evenly distributed across the population create more actual societal benefit than companies that have a sole owner that reaps all of the profits even when they have the same “producer surplus.” One way to resolve this in our calculations is to focus on the benefit to individuals instead of companies, as shown in the example from Chapter 12. Percentage willingness to pay treats everyone as if they have the same capacity for happiness or suffering regardless of their wealth. It makes it so

that John is not treated as being able to experience as much joy as 250,000 people combined. It focuses governments on the things with moral status: people. Percentage willingness to pay can not only make it so we do not value some lives hundreds of times more than others, it can make it so that we accurately value producer surplus instead of treating people with more money as if they had more capacity for joy. A more accurate measure of producer surplus is one that focuses on the people who actually receive the benefit, such as percentage willingness to pay. As shown in Chapter 12, using such a calculation does not eliminate the impact of producer surplus but instead makes it more accurately measure overall welfare.   Consumers Are King The other half of total surplus is defined as consumer surplus. In this calculation, the bias toward the rich is even more apparent. In order to see this, take this simple example. Methuselah is selling lemonade at his lemonade stand for $2 each. Bert, Lavinia, Hester, and Adam each want to buy some lemonade. Bert is willing to pay $10 for a cup. Lavinia is willing to pay $6 for a cup. Hester is very thirsty and would be willing to pay more, but only has $2 on her right now and so she is willing to pay $2 because willingness to pay is constrained by ability to pay. Adam is willing to pay $1.50. At the price of $2, Methuselah sells three cups of lemonade (Adam is not willing to pay $2 and so does not buy). The consumer surplus in this case is the difference between the amount that each consumer would be willing to pay and how much they actually pay. Bert gets $8 of consumer surplus, Lavinia gets $4 of consumer surplus, and Hester gets no consumer surplus

because, according to economists, she is indifferent between having her $2 and having the lemonade. There are some deeply counterintuitive implications of this situation. It seems strange to claim that Bert enjoys his lemonade twice as much as Lavinia, simply because he is willing to pay more for it. As we have discussed, having more money does not make one capable of greater happiness. Similarly, Hester is not in fact indifferent about having the lemonade and having her money. She simply is constrained by her ability to pay. She is very thirsty and gets the most actual pleasure from drinking the

lemonade. However, according to absolute willingness to pay, she is indifferent between having the lemonade and having the two dollars. This is clearly false. Once again absolute willingness to pay fails to accurately represent the world. Specifically, it systematically undervalues the benefits to those with low ability to pay, and overvalues the benefits to those with high ability to pay. To be clear, if Methuselah had only one cup of lemonade, he should sell to Bert at $10, as this will make him the most profit. As a producer his job is to maximize his own profits, and does not need to sacrifice those simply because Hester is very thirsty. It is not the job of a firm to do the right thing (though perhaps it should be), but it is the job of a government. My point is the claim that Bert gains more utility than Lavinia simply because he has more money is inaccurate. Claiming that Hester is indifferent between having $2 and having a cup of lemonade is false. A government that is choosing whether or not to regulate the lemonade market should focus on measuring reality, not the artificial construct of “consumer surplus” that is more related to wealth than actual benefit. Absolute willingness to pay is not merely inaccurate, but just as with statistical lives, it values the happiness of wealthy people more than the happiness of the poor, simply because they have money. To understand this, imagine the following scenario. Imagine two products sold in the country of Mount Diamond (a small mountainous country with 1,000 very wealthy ($20,000,000 annual income) residents who use it as a skiing vacation spot and 2,000,000 very poor residents ($2,000 annual income) who work in the resorts) that are bought by very different consumers. Both products are being sold at cost in a completely competitive market, so there is no producer surplus. One product is luxury jewelry made with an exotic metal found in the area sold at $100,000 per piece to 1,000 consumers every year who would willingly pay $200,000 for each piece. This market has a total yearly consumer surplus of $100,000,000 (1,000 sales with a consumer benefit of $100,000 each). Another product is a cheap space heater used only by low-income families (the wealthy residents have central air). These are sold at $100 each to people who are very budget constrained and could pay at most $110 for these devices (though, like Hester, if they had more money, they would be willing to pay more because

they deeply value being able to stay warm in the cold winters of Mount Diamond). Every year these space heaters are sold to 1,000,000 low-income families (the space heaters are low quality and so need to be replaced often), for a total annual surplus of $10,000,000 ($10 for each of 1,000,000 families). The government of Mount Diamond has a problem. The manufacturing processes that create these products create gasses in the atmosphere that are harmless alone, but when put together they create a deadly poison that is killing many people. The government has calculated the potential loss of life is astronomical and they must ban one of the products (either product on its own is harmless; only the combination of them is deadly). However, they want to use a fair method to determine which product to ban based on which creates more value for the economy. According to absolute measures of consumer surplus, the jewelry creates more value because people are getting it for much less than they would be willing to pay for it, and therefore the government should ban the space heaters and let the people freeze. However, this is deeply contrary to our basic intuitions. Something that benefits 1,000,000 people should not be banned simply because the people that consume it are poor. The wealthy are not budget constrained and therefore will almost always have a much higher willingness to pay than the price in a competitive market. The poor are budget constrained and therefore get a much higher real benefit from goods than their absolute willingness to pay would suggest. It is also concerning that 1,000 people could somehow get more happiness from a non-essential item than 1,000,000 people get from something that is essential. One way to address this, is to normalize the benefit to the person by using willingness to pay. In doing so we find that the wealthy each get 0.5% benefit from their jewelry ($100,000 WTP above cost/$20,000,000 annual income) for a total of 500% benefit. The poor similarly get a 0.5% benefit each ($10 WTP above cost/$2,000 annual income) for a total of 500,000% benefit. Percentage willingness to pay can’t perfectly capture the fact that one item is a luxury and the other an essential, but it can normalize the benefit to the individual, so that more people will generally mean more benefit and the more budget constrained someone is, the more impactful their choices with that budget.

This will almost always be the case for wealthy consumers who are almost never constrained by their income. To understand why, let’s imagine another society, call it Leadpipeville, where the only source of safe drinking water was bottled water. Everyone needs to buy water to survive, and so everyone is willing to pay almost everything that they have to get water (as they would die without it). If water is priced in a way such that most people can access it, the extremely wealthy, who would be willing to pay a huge sum to stay alive, will always have a huge consumer surplus because the necessities of life will always be priced well below their means. Since everyone needs to live, the rich will always get more absolute surplus from necessities that are priced just at the point to make it so the poor can barely afford them. Clearly everyone receives the same psychological benefit and welfare benefit from being able to drink clean water, meaning governments should not focus on maximizing absolute willingness to pay. The way to resolve this, as with the issue of lives, is to use percentage willingness to pay. This controls for wealth in the equation to make it so that no individual’s preferences count for more because of their wealth, and it allows for people who are barely able to afford something to not be discounted as being indifferent between having the money and having the good. To be clear, I am not arguing for market regulation that changes who receives what goods, or prioritizes goods going to poor over the rich. Instead, I am advocating for a more accurate system for governments to understand the impact of their regulations and policies on the utility of their citizens. Surplus may be an effective tool for firms to understand how to set prices, but it is not something that in any way accurately measures benefit to society, and therefore should be expunged from the process of costbenefit analysis.  

 

14. THE VALUE OF BLACK LIVES   The hypocrisy of the economist who proclaims out of one side of their mouth that Black lives matter, while out of the other side argues that if you really look at the numbers they don’t matter nearly as much as white lives, should be self-evident. Even if the United States only explicitly values statistical lives differently by nation not race, the fact that the lives of SubSaharan Africans are worth much less than the lives of white Europeans should raise concerns. The implications of this for international development have been thoroughly expressed in the rest of this text. This chapter will rather focus instead on how absolute willingness to pay implicitly devalues the lives of Black people in the United States and contributes to racial disparities in policing. It demonstrates that absolute willingness to

pay creates ethical dilemmas even when not explicitly valuing life. Redline Town and Greenyard Hills Imagine two communities, Redline Town and Greenyard Hills. They are both suburbs of a large metropolitan area. There are 5,000 people in each community, each with a home worth $400,000. The residents of each community have similar incomes ($80,000 per household), but different levels of wealth. The residents of Redline Town are overwhelmingly renters lacking the income or net worth to afford to buy a home, and spend on average 25% of their income on rent (around $1,666 per month). The properties are owned by five wealthy landlords who are considering this new policing strategy. The owners of Redline Town don’t live in the community but live in the city so they have “standing” for the cost-benefit analysis. The residents of Greenyard Hills, on the other hand, all own their own homes. Now imagine that both of these communities are separately considering adopting the same new policing policy. This policy allows police to enforce quality of life violations (often called broken windows policing)[122] by issuing tickets for misdemeanors. These include things like having a broken window on your home, graffiti on your wall, litter or trash in your yard, etc.

Imagine that a rigorous study finds that this new policing policy has substantial impact on property values (people want to live in an area that looks cleaner, safer, and has fewer quality of life violations). The study also finds that while the policy does have the impact of reducing crime in a neighborhood, it also has the impact of increasing the risk of a violent confrontation between a member of the community and a police officer due to the requirements of police to issue tickets for even very small violations. Specifically, this study finds that property values in the community will increase by 10% each year if it is enacted. Individuals in either community would be willing to pay $1,000 per year to get the benefits of reduced crime, but would need to be paid $5,000 per year to accept the increased risk of violent interactions with the police. For the residents of Greenyard Hills, the results are easy to calculate. They each receive a benefit of $40,000 per year in increased property values, a benefit of $1,000 per year in crime reduction, but a cost of $5,000 per year in increased risk of violent interactions with the police. This leads to a net benefit of $36,000 per person, or $180,000,000 in total net benefit. Using absolute willingness to pay the policy should be adopted. The calculations for Redline Town are different. Each resident incurs a benefit of $1,000 per year in crime reduction, and a cost of $5,000 per year in increased risk of violent interactions with the police, for a net cost of $4,000 per person, or a cost of $20,000,000. This does not take into account the fact that if property values increase, residents will likely bear the costs of increasing rents. Based on this, it might seem that a cost-benefit analysis should find that such a policy would not be welcome in Redline Town. However, there is another party to consider, the five landlords, who each have a net worth of $400,000,000 (owning 1,000 homes at $400,000 each). They will each see an annual gain of $40,000,000, for a total net benefit to the landlords (who do not live in the community) of $200,000,000. If we subtract the costs for the residents, the net present value of the policy is identical to that in Greenyard Hills, $180,000,000. Such a cost-benefit analysis would lead us to enact the policy in both communities, despite the fact that the vast majority of Redline Town residents would bear a cost. In other words, traditional cost-benefit analysis would saddle the residents of Redline Town with a policing policy for which they bear the entire cost, in the form of more violent confrontations with the police and see very little of

the benefits. This bad deal might lead such residents to rightfully resent such a policy, further exacerbating confrontations with police. It is not hard to take a step from this example to the possibility that the residents of Redline Town happen to be of a minority racial group, and lack the wealth to purchase a home due to historic exclusion from housing programs making homeownership more accessible. This is not in any way to say that such considerations

 

Focusing on enforcing quality-of-life violations disproportionately harms poor renters who are harmed by increased property values (in the form of higher rent) and increased negative interactions with the police.

are the only cause behind racial disparities in policing, or even to claim that these are among the largest causes. Factors such as conscious and unconscious bias of officers,[123] correlations between poverty and crime,[124] distrust between minority communities and the police,[125] militarization of police forces,[126] lack of social work training for police,[127] lack of accountability for officers that break the law,[128] and many more, may also contribute. Rather, this is simply to show that the way that economists calculate value and benefit is one of those contributing factors. The remainder of this chapter will show first that percentage willingness to pay would perform better in this situation, but also that the above thought experiment is anything but a thought experiment.   Percentages Make Lives Matter Percentage willingness to pay can help to resolve not merely issues of the value of a statistical life, but concerns around the racial disparities in policing highlighted here. To claim that this methodological change would completely address the systemic issues present in these confrontations would be naïve. However, this framework of valuing percentages instead of dollars is a piece of the puzzle in both understanding why this disparity exists and one thing economists can do easily to help resolve it. Returning to our example, each resident of Greenyard Hills values the policy at 45% of their annual income ($36,000/$80,000) for a net total benefit of 225,000% (45% multiplied by the 5,000 residents, or the equivalent of 2,250 getting raises that doubled their salary). That community would still experience a net benefit from enacting the policy. However, each Redline Town resident bears a net cost of 5% of their annual income ($4,000/$80,000) for a net total cost of 25,000% cost (or the equivalent of 250 people becoming completely destitute). Imagine that the landlords each make $10,000,000 annually (they make $20,000,000 in rent, but spend half of that in property upkeep and management). They would each get an annual increase of $40,000,000, for a benefit of 400% each, or 2,000% total benefit for all five. This means that the policy would cost residents of Redline Town a total of 23,000%, or the equivalent of 230 people losing everything. Based on this analysis the policy should be enacted in Greenyard Hills but not in Redline Town.

Before moving away from this example, there are a couple of things to note. First, this likely underestimates the costs to residents of Redline Town because rising property values lead to rising rents, meaning that they will bear a monetary cost as well as the risk of confrontations with the police. Additionally, there are interactions effects between these incentives. Because rising property values lead to increased rents, renter communities have a financial incentive to make their community appear less appealing to external buyers that might price them out of a home (or at the very least engage in less upkeep of their homes since they are not reaping the rewards from that investment, the landlord is). This means they are incentivized to engage in the very activities that broken windows policing instructs police to ticket them for (further increasing the likelihood that they would engage in violent confrontations with the police). Third, because the residents are renting, they have fewer resources to start with despite having the same income because they are spending a portion of their income on rent, which the homeowners are not spending (which is not factored into this analysis). This means that, in a percentage sense, they actually value the decrease in crime and risk of confrontations with the police more (because they have a smaller share of resources after rent), despite stating they have the same absolute valuation and having the same income. Finally, it is important to note that this would not necessarily apply to all renter communities. Some communities may value the reduction in crime more and the risk of being in a violent confrontation with

police less. They might therefore receive a net benefit even if they do not receive the benefits of increased property values (or in fact receive a cost in the form of higher rents). However, it seems likely that a community that has a negative history with police, or one that was more likely to be victims of police violence would place a higher value on fewer violent confrontations with the police than a community that did not have such a history. More than a Thought Experiment The remainder of this chapter demonstrates that this is not merely a thought experiment but that these dynamics do play out across the United States and contribute to (though certainly do not fully account for) racial disparities in policing. It will also show that the use of absolute willingness to pay instead of percentage willingness to pay disproportionately impacts Black communities and other communities of color that have historically been excluded from methods of generating intergenerational wealth (such as homeownership). The argument is divided into three sections. The first shows that Broken Windows Policing has a cost for renters and a benefit for owners, the second shows that this cost is more likely to fall on poor, Black communities, while the benefit is more likely to accrue to wealthy, white communities. The final section shows how absolute willingness to pay contributes to this problem,

and how percentage willingness to pay could help alleviate it. Section 1: Broken Windows Policing is a cost to renters, but not nonresident owners 1. Broken Windows Policing (BWP) increases property values 2. BWP leads to more violent confrontations with police 3. Other positive outcomes of BWP (e.g., lower crime) are spurious. 4. Increased property values are a benefit to owners but a cost for renters (because they lead to higher rents. 5. Violent confrontations with police are a cost for residents, but not nonresident owners of rental properties. 6. Therefore, BWP leads to a net cost for renters and a net benefit to nonresident owners. Section 2: Broken windows policing costs poor communities of color, while benefitting wealthy white communities

7. Renter communities have less wealth on average and are more likely to be communities of color. 8. Owners of rental properties have more wealth on average and are more likely to be white. 9. Therefore, BWP is more likely to be a net cost to poor communities of color (who are more often renters), and a net benefit to wealthy white communities (who are more often owners). Section 3: Percentage willingness to pay values these outcomes more justly and more accurately. 10. Absolute willingness to pay as a decision framework overemphasizes benefits to wealthy communities and underemphasizes the costs to poor communities compared to percentage willingness to pay. 11. Therefore, absolute willingness to pay underestimates the cost of BWP to poor Black communities of violent confrontations with the police, while overestimating the benefits of higher property values to white owners, compared to percentage willingness to pay. 12. Therefore, percentage willingness to pay would more accurately value the costs and benefits of programs such as BWP, and lead to fewer such programs being adopted in neighborhoods where renters would bear a net cost (thereby reducing violent confrontations with the police). The Impacts of Broken Windows Policing There is a great deal of debate as to the impact of BWP. For our purposes, three impacts stand out: the impact of BWP on property values, the impact of BWP on violent confrontations with the police, and the impacts of BWP on other outcomes such as crime rates. To defend the first six premises of the argument above, I will argue that BWP increases both property values and violent interactions with the police, while the evidence on the impact on serious crime is mixed at best. With respect to premise 1, the conceptual link between broken windows policing and property values should be clear. If you punish property crime and quality of life violations more severely, they would be expected to decrease. A reduction in these activities leads to a neighborhood being more desirable to live in, increasing its property value. While there is debate as to whether BWP actually succeeds at its promise of decreasing major crimes by targeting small ones, there is evidence that BWP does succeed at

reducing those small quality of life crimes and improving property values, even if that never translates into a reduction in major crimes.[129] For example, Beck (2020) finds “a clear, consistent, and positive relationship” between BWP policing policies and property values. Even if BWP fails to have an empirical effect on property values, one benefit of it is its perceived impact on property values, which will show up as a benefit for owners in cost-benefit analyses based on contingent valuation even if that benefit does not actually take place in the market, so long as they think that there will be a benefit. Second, we must defend the second premise, the claim that BWP leads to an increase in violent interactions with the police. Intuitively this also makes sense. Police being required to give out more tickets to low-level offenders would increase the range of the population that they interact with. Additionally, these crimes are often a consequence of poverty instead of criminal intent (lacking the funds to fix a broken window, or the private home to avoid public urination), meaning that people can feel that they do not have other alternatives and may lash out at an officer giving them a ticket that could drive them further into poverty. The evidence for this can be found in the many Black men who have been killed by police in response to quality-of-life violations, whether that is Eric Garner selling cigarettes on a street corner, or George Floyd being accused of paying with a counterfeit bill. Had police not been required to enforce these quality-oflife violations so stringently in line with BWP, these confrontations might have been avoided.[130] For example, Joscha Legewie and Jeffrey Fagan (2019) find that broken windows policies lead to increased contact with the police for Black adolescents, as well as lower test scores for Black boys.[131] It should be clear that BWP has at least some adverse effect on poor communities often in the forms of violent confrontations between police and citizens. The costs accrue particularly to poor non-owners. Finally, there is the question of premise 3, whether the theory behind BWP actually holds water. Does BWP actually lead to decreases in major crimes? If it did, there might be a case for the benefits to residents outweighing the costs. While a range of studies have been conducted on this, few have found a significant relationship, and the relationships found have been mixed at best.[132] A common explanation of the apparent success of BWP in spite of limited evidence for this claim is that the adoption of

broken windows policing in the United States coincided with a general decline in crime, even in communities that did not adopt BWP policies.[133] In other words, what initially appeared to be evidence for BWP might have been simply a reversion to the mean. It would not be surprising that many poor residents (who would in theory get the benefits of lower crime) did not buy into the theories of BWP as much as the non-resident owners. This might be because the true impact of BWP was on property values (felt by the owners, but not the local renters), not actually crime rates. If BWP has a positive impact on property values, a mixed or negligible impact on other crimes, and a negative impact on poor people living in the community through additional violent confrontations with the police, then it should be clear that the costs of BWP will be borne by the poor, and the benefits will be reaped by the rich. The remaining three premises follow intuitively from these initial claims. Property values increasing are beneficial to owners but not to renters (premise 4). Non-resident owners see their assets increase and could conceivably sell or rent out the units for a higher price. Renters, on the other hand, face a cost of increasing property values, both in terms of potentially higher rent, but also higher costs of living as local businesses are forced to raise prices to pay their own higher rents. Clearly, non-resident owners do not face the costs of increased violent confrontations with the police in these areas because they are not present in the impacted areas, and even if they were, they are likely wealthy enough to avoid the quality-oflife violations that disproportionately impact the poor (premise 5).[134] Based on these considerations we can conclude there is strong support for premise 6: BWP appears to have a net cost to renters and a net benefit to nonresident owners (though conducting an actual full cost-benefit study of BWP would be a fascinating avenue for future study and could further illuminate these tradeoffs). Race and Wealth If race were not correlated with wealth and homeownership, this discussion might end here, or focus on how absolute willingness to pay once again overstates the needs of the wealthy and understates the needs of the poor. However, in the United States, this is not the case. Wealth and race are correlated, so a policy that disproportionately harms the poor and benefits the wealthy will generally disproportionately benefit white families at the

expense of families of color. To defend the next three premises, we will look briefly at the racial wealth gap in America. To support the first premise, we must show that there are three features that are correlated: race, rental status, and wealth. The racial wealth gap in the United States is well documented.[135] Around 2016, the median Black family had one-tenth of the wealth ($17,150) of the median white family ($171,000) and less wealth than the median white family living at the poverty line ($18,000).[136] The median Black family at the poverty line has a wealth of nearly zero.[137] There is a clear correlation between race and wealth, meaning that policies (such as using absolute willingness to pay) that necessarily benefit the wealthy, will disproportionately benefit white families over Black families. A full discussion of the systemic injustices and government polices that have led to this wealth gap are beyond the scope of this book. However, it is important to note that at least one influencing factor is the lower rates of homeownership among Black families (which both makes wealth harder to build, and is driven by both lower starting wealth and historic policies excluding Black families from buying homes).[138] There is a strong link between homeownership and wealth accumulation in the long term.[139] There is also a link between homeownership and race. In 2017, 42% of Black families owned their home, while over 70% of White families did.[140] These consistent links between rates of homeownership, wealth, and race mean that a policy (such as BWP) that disadvantages poor renters and privileges wealthy property owners is more likely to disadvantage Black communities while advantaging wealthy white communities. Absolute Willingness to Matter So what? It is unlikely that a cost-benefit analysis is required to show that policies like BWP disproportionately harm Black communities and benefit white ones. The problem here is that even if all issues of legal discrimination and individual bias were resolved, the problems of police disproportionately having violent confrontations with Black people would continue because our systems understate the costs to poor communities in the form of confrontations with the police, and overstate the benefits to wealthy landlords in the form of property value increases. The problem is that, according to absolute willingness to pay, there may still be a net benefit to these polices because the benefits to the few property owners are

so large, and the renters are so poor that the costs that they bear are explicitly ignored. This example drives to the very heart of the problem with absolute willingness to pay. It privileges the desires of the wealthy few over the desires of the poor many, simply because they are poor. In America, this means that absolute willingness to pay often disadvantages the needs of Black communities, which are often poorer, in favor of wealthier whiter communities. In the case of BWP, this is not merely an issue of money, but an issue of life. The risk to life of a violent confrontation with police is undervalued because poor people are confined in their ability to pay for a reduction in risk and so are ignored in the face of larger, solely monetary benefits to the wealthy few. The failure of absolute willingness to pay is that it is explicitly an unequal framework: those with more money get a bigger vote in the policies we adopt. And too often communities of color have less money and therefore have their costs and benefits ignored by the system because of a simple accounting calculation. Percentage willingness to pay would not come close to solving all of the problems of policing in America, but it would be a step in the right direction by controlling for wealth, and therefore removing the systemic bias against poorer, and often non-white communities from the calculation. There are many other hurdles that must be removed, but the clear overvaluing of the desires of the rich few and the undervaluing of the lives of the poor many is an important one.  

 

15. COVID-19: YOUR MONEY OR YOUR LIFE   In March of 2020, early in the covid-19 pandemic, Dan Patrick, the Lieutenant Governor of Texas made a comment seeming to advocate sacrificing the nation’s grandparents to preserve economic growth: “No one reached out to me and said, ‘as a senior citizen, are you willing to take a chance on your survival in exchange for keeping the America that all America loves for your children and grandchildren?’ And if that’s the exchange, I’m all in.”[141] While Patrick was pilloried on social media, with many users comparing his comments to dystopian worlds such as Logan’s Run and Soylent Green and accused of being a sociopath,[142] his comments strike at the heart of one of the central ethical issues of the pandemic: how much economic growth is a society willing to sacrifice to save a life? Or, put another way, how many lives is a society willing to sacrifice to save a percentage point of economic growth? The covid-19 pandemic has raised many different ethical questions as it has evolved and changed, and will likely raise many more before it is done. This chapter will examine three questions that have bearing on how welfare is measured and how lives are valued, each from a different chapter of the pandemic. First, the chapter examines the question that Lieutenant Governor Patrick was responding to: is shutting down the economy worth it to save a few lives? This question was particularly salient in the early days of the pandemic, but has appeared on and off throughout. Second, the chapter looks at questions of who should have first access to vaccines, a question that was relevant for developed countries in early 2021, but continues to be relevant for many developing countries. Finally, the chapter covers the question of what to do when individuals have preferences that are actively harmful to others due unjustified beliefs based on disinformation, in the form of vaccine refusal and vaccine mandates. In each case the recommendations of absolute willingness to pay is found wanting when compared with measuring the same scenarios using percentage willingness to pay.

VSL In Vogue Throughout the pandemic, economists have crafted a range of cost-benefit analyses of various polices from economy-wide lockdowns, to social distancing, to virtual learning.[143] These studies weigh the costs of each policy against the lives that are expected to be lost if such a policy is not adopted. As might be expected, minimally invasive policies such as social distancing have more benefit than cost, while policies such as virtual learning, when modeled with potential impacts on a student’s lifetime of learning have more cost than benefit, though these vary by the individual study and the parameters used. Any individual study faces issues as the virus changes, and as we learn more about the long-term impacts of various policies (both positive and negative), the guiding principle seems sound. The more extreme the disease, the more extreme the prevention measures people should be willing to accommodate. One way to think of this is to imagine several different possible diseases. One, call it Rarity Fever, is very contagious, but is completely harmless to most people it infects: it kills only one in a billion people. While we might want to research a cure for Rarity Fever, few people would be willing to change their lives even in a small way to avoid such a small risk (only seven or eight people in the whole world would die from it). Now imagine another, Apocalyptococcus, which is highly infectious and kills 99% of individuals who contract it. If you heard about an outbreak of Apocalyptococcus in your area you would likely do anything in your power to avoid contracting it: stay inside, grow your own food, never leave your home. Society would be more worried about whether the human race would survive at all than if children had lost a few years of schooling. The point is that there are some risks that we are willing to bear a very high societal cost to avoid, and other risks that we are completely unwilling to bear any societal cost to avoid. Therefore, when faced with a disease that is somewhere between Rarity Fever and Apocalyptococcus, where a society might be willing to take some measures to mitigate it but not any measures at all, cost-benefit analysis can be a useful tool for aggregating preferences. covid-19 was such a disease, which was harmful enough to raise questions of what social action should be taken, but not so harmful as to make it clear that any and all measures must be taken.

Incommensurability (Reprise) As economists were publishing these studies, some who may be sympathetic to the views of Anderson and Raz from Chapter 6 argued that we should not be comparing lives to money. Essentially, this line of thought makes the case that any disease should be treated like Apocalyptococcus. Lives cannot be compared to money. It is never valid to say that we should choose saving a dollar over saving a life. Intuitively there is some power to this argument. If you saw a baby and a priceless vase falling from a window and chose to catch the vase, most people would think you were immoral, regardless of how much the vase was worth. Similarly, it is easy to paint those that want workers to stay in jobs while a deadly virus roams the streets as greedy business owners who care more about putting another dollar in their pockets than saving lives. While this seems quite intuitive, there are reasons we might worry that this, like the original arguments about incommensurability, are oversimplifications. While some analyses are comparing the billionth dollar in the pocket of a mogul to the life of worker, not all dollars lost from the economy come from the stock market. Some are coming out of the pockets of people who don’t have a dollar to lose. Ethical economists do not care about the economy for the economy’s sake, but rather care about it because it does have a real impact on people’s lives. Someone that can’t afford food, a home, or medical care is experiencing real suffering that a society should care about, they might even die. If we care about lives lost to covid-19, we should care about lives lost to unaffordable medical care for those that got laid off due to a pandemic shutdown. A more nuanced version of this argument can be found in Stephen John and Emma Curran’s paper “Costa, cancer and coronavirus: contractualism as a guide to the ethics of lockdown” (2021). They argue that there are some monetary costs that can be compared to mortality risk (e.g., children’s education) and other costs that cannot (e.g., individual’s enjoyment of a soccer game, or the enjoyment of a cup of coffee). Specifically, they make the case that if a cost-benefit analysis was run that determined a particular country should lock down for a disease but at the last moment a coffee company presented an analysis that showed that the cost-benefit analysis had not taken into account the degree to which people would miss going out to get their morning coffee, (which would just barely put the analysis in

favor of not locking down) that the government should ignore the numbers because lives are more important than coffee. John and Curran go on to make the case that individual policymakers should first decide whose benefits and costs matter, ignoring ones that the particular policymaker judges are irrelevant, and only counting the costs and benefits that are judged as relevant to that particular politician. The profoundly concerning and deeply discriminatory consequences of such a proposal should be obvious to anyone who has read Chapter 6, taken an introductory class in political science, or ever spoken to a politician. The entire point of cost-benefit analysis is that it removes the potential for individual bias and the

 

If given the opportunity to simply ignore the costs or benefits to any particular group, politicians will ignore the needs of those that did not vote for them.  This committee member would likely ignore the needs of his sheep constituents if given the chance.

ability for politicians to discount and ignore the concerns of marginalized groups. A proposal that allows politicians to exclude costs to groups that are not part of their constituency or benefits that do not fit their world view is dangerous and would completely defeat the purpose of cost-benefit analysis. Under such a model, it is easy to imagine a politician who does not get support from a particular racial group ignoring the concerns of that group as “ethically irrelevant,” possibly because the politician is bigoted, possibly because they lack an appreciation for the values of another culture they do not understand. It is easy to see a politician choosing to exclude the benefits to emissions reduction because they receive money from the fossil fuel industry, or ignore the costs to consumers of not passing digital privacy protections because they receive substantial donations from technology companies. Regardless, this proposal, as with the other often short-sighted and naïve proposals from philosophers, would reduce cost-benefit analysis to nothing but a messaging tool that any given politician could alter to conclude whatever they wanted by excluding or including the right benefits or costs. However, setting aside the substantial and troubling consequences of such a policy, there are also several reasons that we can reject it on a purely philosophical basis. First, return to the example of Rarity Fever. John and Curran make the case that “Frustration at missing your morning coffee, unlike loss of basic schooling, is not morally significant enough to enter the conversation when costs like death are on the table.”[144] Imagine that we discovered that Rarity Fever could be prevented if everyone in the world stopped drinking coffee forever. According to John and Curran, even if it is only a one in a billion chance that you will die from Rarity Fever, we can’t let concerns about drinking coffee to even enter the conversation. If that is not convincing, substitute your favorite food for coffee, and consider no one in the world ever having it again to prevent the one in one billion chance that someone will die. What if everyone needed to eat a carb-free diet? Only protein shakes? At what point do the preferences of other people matter to prevent a one in one billion chance of death? There seem to be a number of tasty foods that we are aware lead to a range of deadly health conditions, which are much more likely than Rarity Fever but John and Curran are not advocating banning them, despite the fact we are comparing the pleasure gained from your favorite snack with death. It seems to me that

the problem is not the fact that we are comparing coffee and lives, but the value each are given. If you overvalue coffee, it is certainly a problem, but claiming that everyone never drinking coffee again is more important than preventing a 1 in 1,000,000,000 risk is quite the bullet to bite. One might even argue that these small

concerns that John and Curran find ethically irrelevant are the very things that make life worth living. John and Curran might respond to this with the example they use, which they take from T. M. Scanlon. This is based on the thought experiment where we are asked to imagine that a technician has gotten their hand trapped in the machinery that was broadcasting the World Cup. The producers need to decide whether to shut off the broadcast for 15 minutes (giving frustration to one billion viewers) or the technician will lose their hand. John and Curran argue that there are levels of harm that cost-benefit analysis cannot capture, no amount of aggregation of frustration will amount to a hand. Even if on billion people were made very unhappy, the person’s hand should be saved. There are many issues with this example. A key problem, as mentioned in Chapter 4, is that cost-benefit analysis is not looking at harm to a specific individual, but rather collective mortality risk. These are the philosophers that Cameron is engaging with when she argues that philosophers do not understand what is meant by a statistical life. A better comparison would be to imagine that the organization that broadcasts the World Cup has discovered that watching soccer is very exciting and will cause a fatal heart attack in 0.0000001% of the population that watches (in fact, there is surprisingly extensive research that shows these risks are very real, and much higher than one in one billion[145]), but if they play the game at 10% speed, this will not occur because the game is much slower and therefore less exciting. It is merely an inconvenience to the viewers to watch the game very slowly, and it would statistically save someone’s life. Perhaps someone would bite the bullet and claim that all sporting events should be played at 10% speed based on this evidence, but I think the intuition here is not as clear-cut as Scanlon’s example. Particularly given the fact that this is a real concern and that few are advocating for slowing down sports. It seems that frustration can be compared with

statistical lives, though there may be an argument that our current valuation of these is incorrect. The mistake that those who claim lives cannot be compared to money are making is not realizing that money is simply the units that economists like to measure welfare in. However, they do expose an important problem for measuring welfare with absolute willingness to pay. Dollars in the pocket of a billionaire are worth less welfare than dollars in the pocket of someone barely surviving. The problem is not that we are choosing to catch a priceless vase instead of a baby, but that we are treating a vase that is about to be donated to fund a lifesaving charity (which will save thousands of lives) and a vase that has the same monetary value but will just sit in the mansion of a billionaire as the same. Absolute willingness to pay cannot take such a difference into account. Percentage willingness to pay can.   Percentages of the Pandemic This debate showcases that the real problem with valuing lives is not that we are monetizing lives at all, but that we are failing to accurately map monetary losses to the real benefits that accrue to people because absolute willingness to pay treats a billionaire losing $100 million and one million people in poverty losing their last $100 as the same. The billionaire won’t starve or die because they can’t afford a medical treatment, but the people in poverty might die. We should compare lives lost to the pandemic to the people who will truly suffer from losing an income, but we should not compare those to the billionaire’s losses—or at least we should take into account the fact that it is a billionaire bearing that cost. And we need to do so in a way that cannot be manipulated by politicians who may have conscious or unconscious biases against particular groups. The clear way to do this is by using percentage willingness to pay. As we have noted previously, percentage willingness to pay does a much better job at tracking real welfare than absolute willingness to pay. This means that it will do a much better job of responding to the concern that money is incommensurable to lives because it is much clearer how the units track with real welfare. A loss of 100% is equivalent to someone losing absolutely everything, something that has a common emotional feeling to everyone, in a way that losing $1,000 does not. Such an analysis would also accurately track when policies are increasing inequality. If we are choosing

between two virus-mitigation methods that save lives, one of which creates costs for the rich few, and another that creates the same costs for the poor many, we should choose the former over the latter, even if absolute willingness to pay might say otherwise. We can only wonder whether we would have seen the same “K shaped” recovery and increases to inequality following the pandemic if the cost-benefit analyses used had used the more accurate percentage willingness to pay methodology instead of absolute willingness to pay to help policymakers determine the best responses.   Vaccines are Scarce Resources Around a year into the pandemic, there was a turning point: the creation of vaccines that could save lives and limit the spread of the virus. In the first months following these discoveries, many questions were raised about who would get first access to these vaccines. For those first months, vaccines were truly scarce resources (and in much of the developing world they still are). Economics often claims to be the study of the allocation of scare resources so it would seem to be the field to look to. How would economists choose to distribute these scarce resources, whose primary benefit is to reduce mortality risk, the exact kind of thing that the value of a statistical life was created to measure? The simple answer for the advocate of absolute willingness to pay is the people who should get vaccines first are those with the most money, or at least the highest willingness to pay. The advocate of absolute willingness to pay might use this situation to raise the following objection to the advocate for percentage willingness to pay. While using percentages may help us understand what policies a government should undertake, when dealing with real goods, percentages create suboptimal and inefficient markets that do not maximize surplus either for firms or for consumers. The percentage willingness to pay framework leads to deadweight loss in a system: benefits that are not captured by anyone but could be. Since it does not maximize overall benefit, percentage willingness to pay is inefficient and, in a consequentialist framework, immoral. For these economists, if we have a finite number of vaccines, those with the greatest absolute willingness to pay are the ones who receive the greatest benefit from the vaccines, and so they should be the ones to get them. They also happen to be the ones who

will give firms the highest profit, so on both counts this is the best system because it gives both consumers and producers the greatest surplus.   Tinyworld and Vaccine Equity To identify the flaws in this reasoning, we can look at a simplified example of the world (see Appendix F for tables of the costs and benefits associated with this example).[146] Imagine that there are only 15 people in the world (call it Tinyworld), all of whom are in danger of getting sick. They have only four vaccines, which completely eliminate the dangers of the disease. In Tinyworld there are also four people who are particularly vulnerable to the disease (these four are 10 times more likely to die from it) and would get an outsize benefit from vaccination. One would initially think that the method for distributing these vaccines is clear: give them to the people who most need it, but following the dogma of the absolute willingness to pay economists, that is the wrong way to think of things. For them, first we need to understand how much money everyone has, and give the vaccines to those with the most money (or at least those who are willing to give up the most money to be vaccinated). In Tinyworld there are four rich people who make $1 million each year, five middle-class people who make $50,000 per year and six poor people who make $10,000 each year. Of the four vulnerable people, one is rich, one is middle class, and two are poor. The rich vulnerable person is willing to pay $100,000 (10% of their income) for the vaccine, while the three rich non-vulnerable people are willing to pay $10,000, (1% of their income). The middle-class vulnerable person is willing to pay $5,000 (10% of their income), while the non-vulnerable middle-class people are willing to pay $500 (1% of their income). The poor vulnerable people are willing to pay $1,000 each (10% of their income), and the poor non-vulnerable people are willing to pay $100 (1% of their income). This example is set up to mimic the empirical evidence we saw with the value of a statistical life: the actual risk tolerance for people as measured by percentage willingness to pay is dependent on their real risk, not income, but absolute willingness to pay is dependent on income. The vaccines are produced by two companies each entirely owned by one of the non-vulnerable rich people. The firms can make the vaccines at a cost of $100 per vaccine, but due to a lack of materials and production

capacity, etc., are each only able to make two this year. While a firm that only produces two vaccines a year may seem odd, this is similar in percentage terms of the entire population to the very limited availability of vaccines early in 2021, and which still exists in many developing countries. The firm owners capture all of the profit from the vaccines that their firms sell. In such a world, who should get the vaccines? According to the advocate for absolute willingness to pay, the price would be set at $10,000 given the limited quantity of only four vaccines (if either firm tried to set the price higher to capture the surplus from the one individual willing to pay more, the other firm would undercut them, until competition brought the price down to $10,000). The four individuals with the highest willingness to pay will get the vaccines (the four rich people). Three will get no consumer surplus from this (having paid exactly what they were willing to pay for it) and one will get $90,000 in surplus (having paid $90,000 less than they were willing to pay). The owners of the firms will each make $19,800 in profit (the amount paid ($20,000 each), minus the cost to produce the vaccines ($200)). The total surplus in the system would be $129,600. The three poor and middle-class vulnerable people would likely die before the firm can ramp up production enough to provide them with a vaccine, or another firm can enter the market to add supply and push down the price, but at least the firms turned a healthy profit. For the economist, the world where the vulnerable people get the vaccine is at best sub-optimal, and at worst outright immoral because it does not maximize absolute surplus. However, for the rest of us, it seems the most vulnerable do receive more benefit, even if they cannot pay for it. To resolve this, we could imagine a world where the government restricts vaccine access to only vulnerable people. In such a world, the price of a vaccine would drop to $1,000 each. The rich vulnerable person would get $99,000 of benefit (as they are paying much less than they would be willing to pay), the middle-class person would get $4,000 in benefit, and the two poor vulnerable people would get no benefits because they are paying exactly what they are willing to pay. The owner of the firms would make $1,800 in profits ($2,000 paid for the two vaccines minus the $200 cost of production). This would lead to a total surplus of $106,600, meaning a loss of overall benefit of $23,000. According to absolute willingness to pay, the

lives of a few poor people are nothing in the face of a few thousand dollars of profit for the rich because they weren’t willing to pay enough to save their own lives.   Absolute Gets It Wrong It should be clear that, despite the protestations of economists, the second world is more preferable. The health benefits to the vulnerable individuals are underestimated by absolute willingness to pay, while the health benefits to the rich nonvulnerable are overestimated. Some economists (such as Sunstein) might try to brush this under the rug by claiming we should have a separate welfare assessment. However, the problem is that assessing the costs and benefits to society is exactly what absolute willingness to pay is trying to do, and it is failing at it. Even setting aside percentage willingness to pay, the outcomes of each scenario are clear. If the government does not restrict who gets access to vaccines, more people will be dead, and the firms will make slightly more money. However, if the government does intervene, fewer people will risk death, and the rich vulnerable person will get more surplus. The only people who lose out are the business owners who were taking advantage of limited supply to overcharge for vaccines (they still turn a profit, just a smaller one). It seems that the second scenario is preferable, and this is in fact what many governments did. Note that this is a philosophical claim, not an economic one. I am not making the case that there is some market failure that leads to a market equilibrium that is less than the most economically efficient result. Nor am I making the claim that there are some other individuals in society who are impacted by an externality not captured in the market. Rather, the proponent of percentage willingness to pay is making the case that the economically socially optimal equilibrium as measured by absolute willingness to pay is neither socially optimal nor moral. Government intervention in this case leads to a more moral outcome, so any system that finds otherwise must be inaccurate and immoral. The reasons for this are clear: absolute willingness to pay inaccurately overstates the benefits to the wealthy while understating the benefits to the poor.   Percentage Willingness to Pay Gets It Right

A government that measured benefits using percentage willingness to pay would get this problem right. The scenario that is identified as socially optimal by such a system would in fact be the morally right course of action because it is not skewed by wealth. If we distributed vaccines by percentage willingness to pay, the four individuals who are willing to pay 10% of their income should get the three vaccines and the other 11 people should wait. This also maximizes the health outcomes for the population. Looking at the market solution with no government intervention where only the rich people get the vaccines, we see that one individual (the wealthy vulnerable person) gets 9% benefit (10% benefit from the vaccine -1% cost); two individuals (the wealthy owners of the firms) gets no net benefits from the vaccine (1% benefit, -1% cost) but each get a 1.98% benefit from the firm’s profits; and the last rich individual gets no benefit (1% benefit, -1% cost). This provides a total percentage surplus of 12.96% to society. If the government does restrict vaccines to vulnerable people, then the rich vulnerable person gets a 9.9% benefit (10% benefit, -0.1% cost); the middle-class vulnerable person gets an 8% benefit (10% benefit, 2% cost); and the two poor people each get no net benefit (10% benefit, 10% cost). The two firm owners each get 0.18% benefit (0.2% benefit, 0.02% cost of production). This gives a total benefit of 18.26%. Percentage willingness to pay is therefore a better model because it accurately measures the scenario that has more societal benefits as more beneficial. Note that even this is not the maximally beneficial situation. A government that subsidized

vaccines for poorer people using taxes from the wealthy would be measured as even more societally beneficial under this framework because we would capture the 10% willingness to pay from each of the poor people, at a 0.1% cost to the millionaires.   Supply-biased Economics The advocate of absolute willingness to pay might argue that firms will not have an incentive to produce if the prices they can sell at are too low. If the government were to set price controls or require higher income individuals to stay out of the market, then the firms could not turn a profit. Particularly when it comes to vaccines, given the cost of research and development, if firms cannot recuperate their research costs, they will not have any incentive to develop the vaccines in the first place. The overall benefit of having vaccines at all outweighs some amount of higher costs for society that allow firms to recuperate their initial investment. While in principle this may be the case, the details matter substantially. In Tinyworld, the firms had very low costs, which means they would not have been materially harmed so long as the prices were above their costs. Even at the lower $1,000 price, the firms were still making a 900% return on their investment. If firm costs were higher, say at $5,000 per vaccine, the government simply restricting vaccines to vulnerable people would be insufficient to achieve the socially optimal outcome. Rather, the government would also need to subsidize these vaccines. This does not mean that we should give vaccines only to the wealthy, just that we need to be careful with the government tools used to influence the market. Additionally, there is a case that in a competitive market with high demand, limited supply is generally a temporary phenomenon. As we saw with vaccines in the developed world, while there was initially a need to restrict access to individuals that were more vulnerable, there is no longer. The challenge now is largely that vulnerable individuals in poor countries that are unable to similarly subsidize the vaccine are unable to access vaccination. Lacking a global government that values all lives equally, there were no restrictions placed on young, healthy Americans getting the vaccine before poor vulnerable Senegalese. The traditional economist, following the doctrine of absolute willingness to pay, finds nothing wrong with this, any more than moving the toxic waste to poor

countries. It is simply that people in those countries are not willing to pay the market price for covid-19 risk reduction. People in wealthy countries are willing to pay much more to have extra vaccines around going to waste than people in poor countries are to pay to have any at all. Even if you accept that there are political pressures that pressure leaders of wealthy countries to choose to vaccinate their own populations first, it should be clear that such a choice is immoral for the same reason that the rich nonvulnerable people in Tinyworld getting the vaccine first was immoral: it

 

Absolute Willingness to pay overemphasizes benefits and costs to suppliers, since it treats profits directly as benefits, even though, for sole proprietorships, profits are only held by a single person. When the product that the company is producing cand save millions of lives, it seems strange to hold one person’s income over the lives of millions of people, just because those people are poor.

increases suffering, and does not maximize benefits as measured by the more accurate percentage willingness to pay. Traditional economics is not truly a means to find the most societally beneficial outcome, but rather a tool to make firms high profits and justify giving benefits to the rich at the expense of the poor.  

Losing a Tennis Match to an Idea In early 2022, the reigning champion of the Australian Open, Novak Djokovic, lost his title, not to a better player and not because he did not want to defend it. He lost it to misinformation. Specifically, Djokovic had his visa revoked by the Australian government over his refusal to be vaccinated against covid-19, his anti-vaccine rhetoric, and breaking of quarantine rules, which included attending large charity events days after testing positive for the virus.[147] This should not be surprising given Djokovic’s other questionable beliefs, such as the claim that people can remove toxic materials from water and food with their minds, and apparent support for politicians who deny the Bosnian genocide.[148] However, it does raise several questions: What should economists do when people are bad at assessing the benefits that a policy would give them? How can we measure benefits to a population that believes that these benefits don’t exist and would pay money to avoid those same benefits? The remainder of this chapter will look at this third phase of the pandemic, where the cure is available for most in the developed world, but too many people don’t want it. If such irrational beliefs were relegated to tennis stars, there would not be a substantial problem for measuring risk tolerance, as they are few and far between, and rarely captured in surveys. Unfortunately, misinformation about vaccines has become widespread in many countries despite the clear benefits of vaccination. This contradiction has been highlighted by online communities that have been created to highlight individuals who denied the benefits of vaccines or masks and subsequently died from the virus, such as the Reddit community r/HermanCainAward.[149] The persistence of these beliefs poses a philosophical problem for economists: How to value benefits that recipients do not believe are benefits. While this problem poses issues for anyone trying to accurately count benefits, it is particularly problematic for the current paradigm of measuring

mortality risk because of its reliance on averages and its reliance on absolute willingness to pay. Averages pose a problem because this is a case where there is a real, substantial difference in people’s willingness to pay for risk reduction that is not based on income (some view vaccines as a cost, while others view it as a benefit). This means that the average measure of economic benefits is diluted by those that think this benefit is really a cost. Disaggregating benefits will show that some individuals value vaccination highly, whereas others would pay a great deal to avoid it. Absolute willingness to pay is problematic because even with disaggregation, a small group of misinformed rich people could outweigh the accurate beliefs of a much larger group of poor people, but even percentage willingness to pay will struggle to accurately capture benefit in this situation.   Agonizing Averages Again As discussed in Chapter 7, current practice of measuring mortality risk requires averaging across demographic groups. This obscures the uncomfortable reality that, when using absolute willingness to pay, rich people’s lives are worth much more than the lives of poor people. We looked at some benefits of disaggregating risk preferences in Chapter 7, but the example posed by vaccine resistance is quite stark. If half of a population views vaccines as a benefit and the other half views them as a cost of the same amount, these preferences will average out and the apparent benefit of risk reduction will be zero. This is problematic for several reasons. First, people do not actually value their lives at $0, so using vaccines as a way to gauge overall mortality risk reduction may be misleading, as there is a subset of the population that is biased against vaccines in a way that they are not biased against other risk reductions. Second, it fails to capture the true preferences of either community of people, viewing the vaccines neither as a cost or a benefit, when no one in the community has that view, their true views are simply hidden by averages. To understand the impacts of this, imagine a country called Conspiracyland. This country that has recently been hard-hit by a dangerous virus that is killing hundreds of people each day. The government of Conspiracyland is currently debating whether or not to develop a vaccine

for this illness. The majority of the population (70%) supports the development of this vaccine and would willingly pay $100 each for a dose. However, 30% of the population are conspiracy theorists who believe that the vaccine will cause everyone who takes it to mutate into pig monsters and destroy the town (unfortunately not far from various conspiracy theories that claim covid-19 vaccines will alter your DNA).[150] These conspiracy theorists are each willing to pay $250 to prevent the country from developing a vaccine because they really do not want to be taken over by pig-people. This means that on average residents of the town would need to be paid $5 to even let the town research such vaccines, but clearly this is an inaccurate picture of the desires of the town. Most people in the community want the vaccine. But a vocal minority does not want it. Because this vocal minority sees vaccines as a very high cost on them, the overall calculus of the community is to not research the vaccine. Note that if the results of such a survey were never disaggregated by preferences, decision-makers would not see that most people actually wanted the vaccine to be developed, only that on average individuals viewed vaccine development negatively. This is partially due to absolute willingness to pay, thereby skewing the preferences of a minority, and partially due to averages obscuring differences in preferences. If the government were instead provided with the disaggregated data, they might realize that the minority’s concerns were irrational and develop the vaccine anyway—or subsidize anti–pig-monster fences for conspiracy theorists. Averages mask real disparities in risk preferences, as well as factual misunderstandings.   Plutocracy from Pluto These issues are exacerbated by the use of absolute willingness to pay. Imagine a very unequal community, New Unstable City, where 100,000 people have only $1,000 in wealth, and one person has $1,000,000,000. The government of New Unstable City is deciding whether or not to mandate a vaccine against a highly contagious and deadly disease. The 100,000 poor people are all strongly in favor of such a policy and would willingly pay $300 for a mandate to be put into effect because they are aware that if anyone is not vaccinated the virus may mutate and evade the vaccine. Unfortunately, the sole rich person in the city believes that the vaccine was

developed by aliens from Pluto and that it contains microchips that will try to control his mind (once again, not far from real covid-19 conspiracy theories).[151] Because he is so afraid of this, he is willing to pay $50,000,000 to avoid this from happening. Using absolute willingness to pay, the rich conspiracy theorist’s views prevail despite being the only one in the community to have those views: the policy is judged to have a social cost of $20,000,000 (benefits: 300 x 100,000 = $30,000,000; costs: $50,000,000). Such a farce would not occur if we used percentage willingness to pay to measure benefits. The benefits to the majority of the population (30% x 100,000 = 3,000,000%) grossly outweigh the benefits to the single rich person (5%). Politicians may debate over the best way to achieve full vaccination, whether that is mandates or incentives, but an economic system that privileges the views of a single rich conspiracy theorist simply because he has money is clearly morally bankrupt.   Imperfect Solutions While allowing for disaggregation and percentage willingness to pay provide a more moral and accurate picture of the costs and the benefits of a policy, they are at best imperfect solutions to this problem because they do still count the conspiracy theorist’s views as real costs, even when they are based on falsehoods. If a majority of people thought that vaccines would turn us into pig-monsters, no amount of disaggregation would show that their self-assessments of preferences were wrong. Another way to think of this is that surveys of consumer preferences may be effective at measuring the values of an individual consumer but fail at measuring whether or not a policy will provide benefit because consumers are bad at assessing which policies will actually provide those benefits. Two people might both have the goal of living a long life, but one might do so by consuming horse dewormer (yes, another real covid-19 conspiracy), while the other might get vaccinated and wear a mask.[152] Those two individuals may actually value their lives the same amount, and so be willing to pay a similar amount (or percentage) to avoid the risk of covid-19. However, if presented with the question of “What would you pay for a shot of horse dewormer?” or “What would you pay for a vaccine?” they will provide different answers because they have different views about

the facts despite having the same views on the value question. Should economists count the benefits of the vaccine for the conspiracy theorist who does not acknowledge it? Should they price in the benefits of horse dewormer even if there is no scientific evidence behind it because enough people believe it is beneficial? Traditional economic logic seems to state that for the conspiracy theorist we must value the dewormer and not the vaccine, though this is clearly problematic. To clarify this, imagine that you are the dictator of a country that is about to experience an outbreak of Apocalyptococcus (the deadly disease mentioned earlier that kills 99% of people who are infected). Unfortunately, everyone in the country (other than you) thinks that a truly effective vaccine will kill them and Apocalyptococcus is not dangerous at all. You, not bound by pesky democracy or cost-benefit analyses, can force everyone to get vaccinated (imagine that you have a military that will do what you tell them even if they disagree with you, or perhaps a force of robots to enforce your will). You know that doing so will save everyone’s life (but they are all very unhappy about it), but not doing so will lead to 99% of the population dying (you know that other policies like public information about the vaccine or incentives will fail). What is the right action to take? Should you give everyone the vaccine even though they don’t want it? This is a problem for another underlying assumption of cost-benefit analysis that individuals are the best judges of their own welfare. If you are justified in doing so, then why are we measuring individual’s preferences in the first place? Wouldn’t it be better to measure actual impact? If not, should governments give into conspiracy theories if they make people happy? This paradox is very interesting, but a full treatment of it will need to await a future text. For now, it should be clear that these issues can be at least mitigated by disaggregated percentage willingness to pay by preventing a minority of wealthy conspiracy theorists from skewing averages.  

         

 

16. OBJECTIONS AND RESPONSES     Throughout this text we have considered many potential concerns or objections to the central argument presented here. This penultimate chapter will expound upon four such objections that may appear to either philosophers or economists and preempt challenges by responding to them. If this book is asking economists to be critical of the very foundations of their discipline, while asking philosophers to accept that in fact there are some situations in which we can value lives, the least it can do is try to offer these claims a strong defense. The first objection comes from the economist’s perspective, which focuses on the claim that this is not how VSLs are actually used and that no one would ever really compare two projects where drastically different values of statistical lives were used, so the initial concern is moot. The second comes from economists who claim that we do not need to worry about measuring percentage willingness to pay because if the Kaldor-Hicks criterion is correctly applied, there will not be consistent winners and losers over time. The third objection comes from the philosopher’s camp, and essentially argues that cost-benefit analysis as a whole is a nonsensical, defunct, unscientific doctrine that needs much more extensive revision, due to concerns about the Kaldor-Hicks criterion and the weakness of the underlying data that informs VSL calculations. The final objection is more a politician’s concern than either an economist’s or philosopher’s concern. This objection claims that using percentage willingness to pay will bias our public policies toward poverty reduction and take something that should be an unbiased measurement exercise and bake in a political view that we should be supporting the poor more than the rich. All of these concerns are unfounded, but this chapter will attempt to give them a full treatment and then show why they are unjustified.   Objection 1: Really, That’s Not What VSL Means! While writing this, I cannot help but hear the chorus on economists shouting the title of Chapter 4 at the page. Economists have a great deal of institutional power in the bureaucracy of public policy and international

development, which often have entire departments full of them. This power generally allows them to ignore the concerns of philosophers (who are rarely if ever found in such institutions). It therefore is difficult to bring a challenge to economists on philosophical grounds without being dismissed out of hand. However, I am hopeful that if this objection is given full treatment, economists may be interested in reforming their views. One of the key responses to this objection that was discussed above is that regardless of what you call a VSL or what units you use to delineate it, it makes a material difference in policies that lead to pressure on decisionmakers to accept polices that save fewer lives of rich people than those that save more lives of poor people. The economist might respond that in practice no organization is actually comparing policies where different values for a statistical life are being used. These values are for use when assessing the impacts of a project in a given country but not for comparing projects across countries. Therefore, so long as the estimates are averaged across the population used, no moral dilemma arises. No one is actually comparing saving two French lives to saving 100 Ethiopian lives because we are never using cost-benefit analyses to assess the comparative advantages of polices that save rich lives and policies that save poor lives when those lives are valued differently (Sunstein’s arguments notwithstanding). There are three responses that we might offer to such an argument. First and foremost, one could offer a real situation where cost-benefits analyses are used to judge between projects in different countries where differential values of lives are used. Second, one could show that even if no one is actually doing this, it is only to avoid public relations disasters, as the foundations of economics prescribe such an approach. Third, one could argue that concerns regarding the value of a statistical life are merely symptoms of the greater problem of using absolute willingness to pay to value benefits, which would not disappear even if economists never valued a statistical life again.   Response 1.1: Yes, Economists Do Use It That Way The simplest way to respond to the claim that X does not exist is to provide an example of X. In this case, our example comes in the form of a U.S. government foreign development agency called the Millennium Challenge

Corporation or MCC. MCC was founded in 2004 with a model focused on giving aid based on underlying principles of good governance and costeffective investments.[153] Given MCC’s focus on transparent and accessible data, their methods are available for the world to see.[154] Unfortunately, an institution that is focused on hewing closely to economic principles instead of political ones will inevitably run headlong into the underlying flaws of economics itself that more political institutions have papered over with averages and obfuscation. MCC’s process involves conducting a cost-benefit analysis of any proposed projects. These cost-benefit analyses are then used to calculate an economic rate of return or ERR, which is essentially the internal rate of return for the projects benefits (i.e., the expected percentage return on the investment).[155] In order for MCC to invest in a project, the ERR must exceed 10%. In other words, projects in different countries are weighed against each other based on the net benefits they create to determine if they will receive funding.[156] In order to calculate benefits of individual projects, MCC does use differentially valued statistical lives based on the income of the country they are working in. In fact, they explicitly note the connection between the value of a statistical life and the income in a country, actively using the income of a country to calculate how much a life in that country is worth. [157] One could easily conceive of a situation where MCC was deciding between two health and sanitation projects in different countries, one that saved fewer rich lives, but those lives were rich enough to put it over the 10% ERR threshold, and another that saved more poor lives, but because of this use of absolute willingness to pay to value statistical lives, those lives were not worth enough to get them over the hurdle. While this distinction is couched in enough technical jargon to avoid public criticism, it is no different from the initial thought experiment that compared two projects, one that saved more poor lives and one that saved fewer rich lives. MCC does use this very calculation and in such a situation would easily arrive at the conclusion proscribed by the principles of economics: poor lives are simply worth less, and are not as worth saving as rich lives. Note that this is not to be particularly critical of MCC over other development institutions. In fact, it is a testament to MCC’s transparency that they are willing to “say the quiet part out loud,” when it is likely that

similar calculations are being done in other development organizations, just behind closed doors. That said, such an example should clearly dismiss the claims of economists that no one is actually making decisions of whether to save a few rich lives or many poor ones on the basis of disparate valuations of statistical lives. Many institutions may be making such a calculation, but at least one is doing so sufficiently transparently to respond to this objection.   Response 1.2: The Foundations of Economics Even if they are not convinced by that response, economists are faced with a dilemma. Either they think that valuing statistical lives differently based on absolute willingness to pay is immoral or they do not. If they think it is immoral, then they should be concerned that the fundamental tenants of economics imply an immoral conclusion, regardless of whether that conclusion is actually in use or not. If your fundamental beliefs imply that you should commit genocide, but you personally decide to go against them, it does not mean that your fundamental beliefs are therefore excused. Instead, it means that you may be moral, but your belief system is not. Similarly, if economists acknowledge that making decisions due to differential valuing of lives based on absolute willingness to pay is immoral and that the underlying principles of economics imply such a conclusion, that is sufficient to repudiate those fundamental claims, even if the economists fail to act on the dictates of their discipline. On the other hand, the economist may claim that rich people’s lives are in fact more valuable. If given the choice between saving the life of one billionaire or a hundred paupers, we should always pick the billionaire. When put in such stark terms, I doubt that an economist would defend such a position, despite the fact that this is what the underlying precepts of economics imply. However, if they did, there seem to be a multitude of philosophical arguments that could be arrayed against them. From the consequentialist standpoint, killing more people is clearly doing more harm. While the deontologist might make a distinction between killing one or letting 100 die, it is unlikely that if faced with letting one die or letting 100 die they would choose the one. And I doubt that a virtuous person would choose to save one life instead of 100. From the philosophical point of

view, there are few that would support such a seemingly immoral conclusion. Finally, the economist might attempt to avoid this dilemma altogether, claiming that even in theory there is never a situation where someone would be making a decision between saving 1 life and saving 100. Rather these questions should be about risk reduction. Statistical lives are asking the question of if we should give the billionaire a vaccine or 100 poor people a vaccine. As we have shown, this is merely an obfuscation of the issue. Once the numbers are rounded up, the result is the same. If you have a vaccine that reduces a 1 in 100 chance of death to 0 that you can either give to 100 billionaires or 10,000 poor people, the economist still thinks you should give it to the billionaires who have a higher willingness to pay and therefore receive greater benefit from the marginal reduction in mortality risk. However, statistically that vaccine would save the lives of one billionaire or 100 paupers, meaning this is essentially the exact same case as before, just with a layer of statistics. It is a tall order for the economist to make the case that that layer of statistics justifies killing the poor many to save the rich one. Some economists have defended the objection by claiming that this should not be the purpose of cost-benefit analysis. Sunstein argues that welfare should be assessed separately from a cost-benefit analysis and that these two may be in conflict, particularly when dealing with statistical lives. [158] Sunstein admits that welfare is primary and that cost-benefit analysis can fail to perfectly capture measures of welfare, but he argues that this is not what cost-benefit analysis was designed to do. The response to this is quite simple: if welfare is the underlying thing we are trying to measure, and we have a better measure of welfare (percentage willingness to pay), why not use that and abandon absolute willingness to pay? By considering welfare separately, we weaken the power of cost-benefit analysis to fully capture the goods received by individuals, since there is always a general feeling that it fails to accurately capture the thing we really care about. Further, there is little to no cost of controlling for wealth in our calculations and therefore better capturing welfare, so why not do it? Why use a bad tool when there is a better one available?   Response 1.3: Lives Are Not the Only Problem

While the issue of differential valuing of statistical lives is the most morally obvious and egregious example of the failings of economics, it is far from the only one. Even if economists were to never value a life again, the underlying measurement problem of failing to accurately value individual preferences would remain. The concerns raised in section III of this book are far from isolated to the value of a statistical life. The economist continues to be committed to the claim that

 

Even if economists swear never to value a statistical life again, or can answer the central question of this book around disparate valuations of lives, this has not resolved the root cause of this objection: the fact that the preferences of the wealthy are given more weight than the preferences of the poor, simply because they are wealthy.

rich people’s preferences count for more simply because they are rich. If the economist attempts to justify their project with reference to the underlying consequentialist principles of maximizing happiness and minimizing pain that should be underpinning the practice, it leaves them committed to ridiculous claims like rich people have an inherent ability to feel greater pain and happiness than poor people, or that someone with no money at all can experience no pain (something anyone that has had absolutely no money at all for any real period of time can easily dispute). The concerns around VSL are simply a symptom of the underlying problem: that absolute willingness to pay is an immoral and inaccurate measure of benefit and cost. To make the force of the argument clear, the point is that any costbenefit analysis that uses willingness to pay to value benefits or costs is both inaccurate and inherently immoral. Value of a statistical life is simply an easy way to see this inherent immorality. This is another reason why the objection that “no one uses VSL that way” is utterly toothless because even if no one ever used VSL, so long as cost-benefit analyses were conducted using absolute willingness to pay, the methodology would be unethical and deeply biased toward the preferences of the wealthy.   Objection 2: Disparities Come Out in the Wash Another economic objection to this process can be found in Boardman et al. This objection makes the case that because in the long term there are no consistent “winners” or “losers” in cost-benefit analysis, the differences in actual benefit will be moot, since some policies will benefit the poor and create more utility than their cost-benefit analysis might suggest, and other policies will benefit the rich and create less utility than their cost-benefit analysis would suggest. Initially Boardman et al. frame out one of the central problems for cost-benefit analysis that we have been addressing here:   “As an illustration, consider a policy that gives $10 of benefits to a person with high wealth and inflicts $9 of cost on a person with low wealth. If the low-wealth person’s marginal utility of money is higher than that of the high-wealth person, then it is possible that the utility

loss of the low-wealth person could outweigh the utility gain of the high wealth person.” [159]   In other words, someone with $100,000 in wealth gets a 0.01% real benefit from $10 in benefit, while someone with $10 in wealth bears a real cost of 90% from a $9 cost. Before continuing with Boardman et al.’s argument, we should note the arguably unnecessary conditional added in Boardman et al. around “If the low-wealth person’s marginal utility of money is higher.” To translate the economic speech this means that a poor person gains more utility from gaining a single dollar than a rich person. As we have argued, this difference in marginal utility of wealth is not merely a matter of preference, but psychological fact. Someone living below the poverty lines gets more happiness from $100 than a billionaire. People cannot feel more simply because they have more money. Rich people do not feel happiness more intensely because they are rich. In the terms of Boardman et al., this means that rich people always have smaller marginal utilities of wealth, so this conditional is superfluous. Boardman et al. go on to argue that these differences can be ignored because some policies will help the poor more and understate benefit, while others will help the rich more and overstate benefit:   “Policies with positive net benefits that concentrate costs on lowwealth groups may not increase aggregate utility; moreover, policies with negative net benefits that concentrate benefits on low-wealth groups may not decrease aggregate utility. However, if the potential Pareto principle [Kaldor-Hicks Criterion] is consistently applied and adopted, then polices do not produce consistent losers or winners. Consequently, the overall effects of these policies taken together will tend to make everyone better



off. Hence, concerns about reductions in aggregate utility would be unfounded.”[160]   Essentially, this is making the case that, as long as we assure that there is no bias in the individuals that benefit from policies and the individuals

that bear a cost, the results will cancel each other out. The onus is on the policymaker to ensure that benefits accrue to different groups, not on costbenefit analysis to accurately measure benefits. The lower utility gained by policies that benefit the rich will be canceled out by the higher utility gained by policies that benefit the poor. This matters for percentage willingness to pay because if we don’t need a better measure of willingness to pay since the differences will come out in the wash, there is no need to stop using absolute willingness to pay. Any individual disparities in benefits will be outweighed by the overall balance that expects there to be no overall winners or losers.   Response 2.1: Why Trust When You Can Measure? The clearest initial response to this concern is an epistemic one: why should we take your word for it that these disparities will balance out? When we have a relatively easy method of measuring utility, why would we assume that clear and acknowledged disparities and inaccuracies will simply disappear with the process of averaging and aggregation? Proponents of cost-benefit analysis generally are strong advocates of ensuring more accurate measures of all of the components in their models, why give up here? Why stop at the last most important step of the math problem and say “it probably works out overall in the end” without doing the work to calculate it? Boardman et al. admit that there are ways of measuring utility. At the very least should we not test the assumption that the benefits to utility, the thing that we actually care about maximizing, would balance out in the end? The point is that if the proponents of cost-benefit analysis are able to fund large contingent valuation studies, a simple division problem at the end of their calculations cannot be too costly in terms of time, to outweigh the benefit of actually capturing utility, or at the very least something much closer to it. Boardman et al. might advocate for another method for controlling for wealth or calculating utility, but there seems to be no reason not to perform such a calculation and claiming that such calculations are irrelevant without a shred of evidence strains credulity, particularly when, as we will see in the next response, we have very good reasons for thinking that, in fact, policies based on absolute willingness to pay will never get around to implementing policies for the poor. This is not because of a

failing to consistently apply the Kaldor-Hicks criterion, but rather something inherent to cost-benefit analysis itself. Furthermore, this means that there are some policies that are actively creating harm in the world that have a positive net benefit but a negative utility cost. Such policies are causing more harm than good, specifically more harm to poor people, but Boardman would have us allow such polices because economists are too lazy to do the last step of the calculation? Such a position either speaks to a deep bias in favor of the rich, a complete unwillingness to admit that the current system is inaccurate and immoral, or both.   Response 2.2: Net Cost Policies Won’t Happen Cost-benefit analysis is not merely a tool to weigh policy options against each other, but one to weigh a particular policy against the status quo and determine if it will create net benefits. Boardman et al.’s argument might have some weight if we expected that all policies that provided anyone any benefit would be implemented, but that is not the case. Cost-benefit analysis recommends against implementing policies that have a net cost, even if they have a net benefit for utility. The second response therefore is to claim that, if cost-benefit analysis using absolute willingness to pay is applied correctly, there will be consistent losers in a utility sense: the poor, as policies benefitting them will appear to have a net absolute cost, even when they have net benefit in terms of utility. With this objection, Boardman et al. are essentially asking us to imagine two possible policies. One that has high net benefit as measured by absolute willingness to pay and low overall utility gain, and another with low net benefit as measured by absolute willingness to pay and high overall utility gain. According to Boardman et al., because Kaldor-Hicks expects that both policies will be enacted, the difference between the utility and the absolute willingness to pay does not matter. There are several problems with this scenario. First, policymakers often only have the resources to enact one policy on any given opportunity, and each time cost-benefit analysis dictates that we should enact the one with greater benefit as measured by absolute willingness to pay. Therefore, the policies that benefit the wealthy will consistently be enacted, even if, in theory, we could enact a policy for the poor. Second, any policies that have a negative willingness to pay, but a

positive utility (e.g. polices that disproportionately help the poor but disadvantage the rich) will be considered unacceptable, and so not enacted. Because of this hard line at zero, the claim made by Boardman et al. that there will be sufficient policies enacted that understate the overall benefit gained (because they benefit the poor) to balance out those policies that overstate the benefit gained (because they benefit the rich) is clearly false. One way to think of this is to realize that, as Boardman et al. admits, absolute willingness to pay is an imperfect and biased indicator of benefit compared to percentage willingness to pay, or some other indicator that controls for wealth when measuring benefit. To see

 

Economists admit that absolute WTP is biased against the poor, and fails to support polices that benefit the poor.  Continuing to use it is like continuing to use a methodology that counts people of particular races as partial people.  It is biased and unethical.

the full scope of this distinction, imagine you are a county executive deciding how to measure the potential benefits of where to build a hospital, either in the small town of Illsboro or the larger town of Metrosick. The first method counts the total number of people that benefit from the project, and counts everyone equally. The second method counts everyone, but White people count triple and Black people only count as half a person. The two methods point to placing the hospital in different towns: the first method points to placing it in the larger, but more diverse town of Metrosick, while the second method points to placing the hospital in the smaller and whiter town of Illsboro.[161] Method 2 is not importantly different from absolute willingness to pay. It understates the real benefit to one group of people (poor people in the case of absolute willingness to pay) and overstates the benefit to another group of people (rich people in the case of absolute willingness to pay). Boardman et al. acknowledge that the corollary of method 2 (absolute willingness to pay) is deeply flawed, inaccurate, and biased yet somehow think it is a more appropriate measure of benefits and should be used to place the hospital in Illsboro because another project will hopefully come around to benefit Metrosick and balance out the bias. Clearly this is deeply problematic. There is no reason to knowingly introduce unnecessary bias into a measurement method when there are easy ways to control for that bias, particularly when there is no guarantee that such a bias will eventually be reconciled. It is not hard to simply count up the number of people that would benefit from the hospital in the same way it is not hard to control for wealth to prevent cost-benefit analysis from consistently being biased against the poor.   Response 2.3: Percentages Ensure Distribution Boardman et al. make the case that the problem here lies with the policymakers who choose projects that consistently advantage one group over another. For them, it is not the business of cost-benefit analysis to avoid systemic biases that may consistently advantage one group over another, though they admit such a system is preferable. Once again, percentage willingness to pay has a solution. If a certain group is consistently advantaged by policies, and gains wealth, the percentage benefit of doing another project for that group will decline over time,

assuming their absolute willingness to pay for such benefits remains constant. Therefore, if ever a set of policies disproportionately advantages one group over another, the system will be self-correcting. The comparative advantages to benefitting the group that has not received investment before will steadily increase. To see this, imagine a very small town called Equihoma City. In Equihoma City there are two demographic groups. Let’s call them farmers and ranchers. The town is about made up of 60 farmers and 40 ranchers. Each month the town’s council meets to decide what project to invest in. Because the interests of the farmers and ranchers are often at odds, investing in a project that benefits one group often disadvantages the other. Imagine that everyone has equal wealth to start off with (say $1,000), and the town council is deciding between two projects. One that gives all the farmers a benefit of $100, while all the ranchers bear a cost of $50, and one that gives all the ranchers a benefit of $100, while all the farmers bear a cost of $50. Upon conducting a cost-benefit analysis, the council finds that the project that benefits the farmers has a higher net benefit, simply because there are more of them (farming project: total benefit = $6,000; total cost = $2,000; net benefit = $4,000; ranching project: total benefit = $4,000; total cost = $3,000; net benefits = $1,000), so the committee chooses to invest in the farming project because it has a higher net benefit. Note that this would be the same conclusion as if percentage willingness to pay were used since everyone has the same wealth at this point (the farming project would provide 400% benefit, while the ranching project would provide 100% benefit). Come the next month, the town council is once again deciding between a ranching project and a farming project, each with the same benefits. However, now the farmers each have $1,100, and the ranchers each have $950 because of last month’s investment. The calculations using absolute willingness to pay will be the same. The council might ignore the higher absolute benefits of the farming project and invest in the ranching project, but there is no guarantee they will, and cost-benefit analysis would generally lead them to the project with the higher net benefits. However, percentage willingness to pay will provide different calculations this month. Now, the net benefits of the farming project are smaller because the farmers are wealthier, (leading to only a 335% benefit instead of last month’s

400%), while the net benefits of the ranching project are bigger, up to 148% from 100%, because they bore a cost last month and are poorer now. The council may still invest in the farming project, but by the fourth month of such an investment, the percentage benefits of the ranching project will outweigh the benefits of the farming project (the farming project’s net benefits when the farmers have $1,300 in wealth and the ranchers have $850 in wealth is only 226%, while the ranching project’s net benefits are 239%). Percentage willingness to pay forces there to be no winners and losers on average instead of ignoring the more accurate measurements and hoping that things work out for the best as Boardman et al. advocate. Absolute wiliness to pay crosses its fingers and hopes policymakers will go against its recommendations to

balance things out. Percentage willingness to pay ensures that the recommendation will consistently balance out benefits. For full tables of how these two policies (relying on percentage willingness to pay vs relying on absolute willingness to pay) will impact the overall benefit of these two communities, consult Appendix G, which shows how benefits will accrue using each decision rule for the first year.   Objection 3: One Reform to CBA is Insufficient Having settled the economist’s concerns, we can turn back to the philosopher. Specifically, the philosopher might be unsatisfied that using percentage willingness to pay is sufficient to resolve the underlying issues with cost-benefit analysis. There are other, deeply concerning elements of cost-benefit analysis that might lead us to think that more needs to be done to make this process moral. This objection is more nuanced than the earlier objection that we should completely discard cost-benefit analysis, as it claims cost-benefit analysis requires further modifications, not the complete elimination of the practice. Therefore, this objection avoids the earlier response that decisions made without cost-benefit analysis are more prone to bias and will be less transparent than those using cost-benefit analysis. That is because this objector is focused on further reforming cost-benefit analysis, not dropping it completely.   This objection might take the form of objections to specific portions of the cost-benefit analysis process, such as the Kaldor-Hicks criterion (the criterion that claims that so long as overall benefits are increased, some individuals may be made worse off). One may be concerned that KaldorHicks allows for unequally distributed benefits regardless of whether absolute or percentage willingness to pay is used. If policies always benefit one group and never benefit another group, it may not matter how those benefits are measured. Inequality will increase, and policymakers will use the justification of cost-benefit analysis as a shield to hide behind policies that systematically benefit one group and disadvantage another just because the overall benefit is being increased. Note that this is an objection coming from the other direction of the last objection. It focuses on the failings of the Kaldor-Hicks criterion regardless of the measurement method used. The proponent of such an objection might argue that we should hold polices to

the higher standard of Pareto Efficiency, where a policy must not only maximize benefit, but not create any net harm to an individual.   Response 3.1: Percentages Fix Kaldor-Hicks One of the many advantages of using percentage willingness to pay over absolute willingness to pay is that doing so will significantly lessen concerns about the use of the Kaldor-Hicks criterion, particularly with respect to the distribution of benefits. Using percentage willingness to pay will actually lessen the calculated benefits from policies that consistently benefit one group and disadvantage another group, as we saw with the Equihoma City example above. It will not fully solve the problem, but it will require much bigger benefits of a program that provides benefits to the rich, and much smaller costs to a program that charges the poor. To illustrate this, take the following example. There is a community of 100 people, each with only $100 to their name. We are looking at implementing a policy that will tax everyone $1, then provide a benefit of $200 to one person, call that person Rene. Such a policy would provide a net absolute benefit of $100 (1 dollar of cost to everyone, and $200 of benefit to one person ($200 - $100 = $100), and a net percentage benefit of 100% (a cost of 1% of their wealth to everyone, and a benefit of 200% to one person). This policy seems beneficial to the community as a whole, as if such a policy were implemented 99 more times once for each member of the community, they would all end with $200 instead of $100 (their initial $100, minus $100 in taxes, plus $200 in benefit). However, one central critique of the Kaldor-Hicks criterion is that often the benefits of a policy continue to accrue to a single group of people, and many people bear the cost of policies while never seeing the benefits of these policies. If we implemented that policy 100 times, but listed Rene as the beneficiary every time, he would end with $20,000 and everyone else would have nothing. Absolute willingness to pay is indifferent between Rene receiving the benefit and someone else receiving it. Using absolute willingness to pay, this would be justified, in fact the two final scenarios (everyone with $200 and Rene with $20,000 and everyone else with $0) would be equivalent because the total net benefits are the same. Yet, a similar problem will not arise with percentage willingness to pay. Beyond solving the issues of valuing the preferences of the rich more,

percentage willingness to pay can actually significantly reduce the issues raised by the Kaldor-Hicks criterion as well, without resorting to using Pareto Efficiency. Returning to our community of 100 people, once the policy has been implemented once, the analysis for percentage willingness to pay will change. After one implementation of the policy, Rene will have $299 (100 initial wealth, plus $200 in benefits, minus $1 in taxes) and everyone else will have $99. If the policy were to benefit Rene again, his wealth would rise by $199, a 67% benefit ($199/$299 = 0.67). Everyone else would bear a cost of just over 1% ($1/$99 = 0.0101), for a total cost of 100%. Based on percentage willingness to pay, this policy would no longer be justified if it continued to give benefits only to Rene because its net benefits would be negative. However, if the policy were to provide benefits to a different beneficiary, call them Fitrat, on the second iteration, the net benefits would be positive. Rene would bear a cost of 0.3% ($1/$299), Fitrat would

receive a net benefit of 201% ($199/$99), and everyone else would bear a cost of just over 1% ($1/$99). This would lead to a total net benefit of 102% for the community. Using percentage willingness to pay not only makes the system more accurate and just, it also ensures that the KaldorHicks criterion is not abused to systematically only advantage certain individuals or groups of people. It should be noted that this will not ensure that the poor will receive benefits in place of the rich. If the second iteration of the policy provided $1,000 for Rene at the cost of $1 for everyone, it would have a net percentage benefit: Rene receives a benefit of 334% ($999/$299 = 334%) while everyone else still bears the cost of just over 1% for a total cost of 100% and a net benefit of 234%. However, if given the choice between benefitting Rene and benefitting someone else with such a policy, the choice would be clear, as anyone else will benefit much more from a policy that gives $1,000 in a percentage sense than Rene will ($999/$99 = 1,009%). This means that such a policy would only be implemented if Rene was the only person that could benefit from it, since the next best option, of given the benefit to someone else, would give more benefits. So, even if it does not resolve all the problems with Kaldor-Hicks, it substantially reduces them.       Response 3.2 Pareto Efficiency Isn’t Better: If this objection proposes to merely reform cost-benefit analysis instead of eliminating it, then some replacement must be supplied for a decision rule instead of Kaldor-Hicks. One option is the Pareto Efficiency criterion, which only allows for policies that provide net benefit to individuals but does not allow anyone to bear any net cost. There are philosophical reasons that one might be interested such a policy, either from concerns about the potential for systems to systematically advantage some and disadvantage others with the Kaldor-Hicks criterion, as well as concerns that polices should “do no harm.” However, there are several issues with such a proposal. First, Pareto Efficiency with absolute willingness to pay is insufficient to solve the central problem of this book (as noted above saving lives can be framed

exclusively as benefits), and it is insufficient to solve the concerns around equal distribution with Kaldor-Hicks. You might have a series of policies that all provide only benefits and no costs, but if they provide benefit to the same person every time, the same inequality noted with the Kaldor-Hick criterion will emerge. The poor will not get poorer, but the rich will get richer. This does not solve the distribution problem, which is commonly the central objection to Kaldor-Hicks. However, percentage willingness to pay does solve this issue as explained in responses 2.3 and 3.1 above. Additionally, as noted when Pareto Efficiency was introduced, this does not resolve the central issue of the book, that lives may be valued differently because the value of a statistical life is generally used to value mortality reduction benefits, and need not be valuing costs to lead to disparities. One might instead advocate for Pareto Efficiency and percentages with the goal of avoiding both the problems with absolute willingness to pay and the ethical concerns that Kaldor-Hicks fails to “do no harm.” While this may be an internally consistent position, it leads to some extreme conclusions about the role of government and public policy. For example, if individuals cannot bear any net cost, they cannot be taxed unless they receive a direct benefit from that tax. Even if 99.9% of the population views a particular policy as providing them a benefit, if one person sees it as a cost, that policy would not be justified. One might think back to the issues with conspiracy theories and misinformation to see the true concerns raised by this. Such a decision rule would make it nearly impossible to pass any policy, as there is almost no situation where there is not at least one person who would pay $1 to have a policy not enacted and therefore bear a cost from it. While governing by consensus may work for very small groups of people, for an entire country there will almost always be at least someone who does not like a particular policy and therefore would bear a cost if it were implemented.   Response 3.3: More Reform Might Not Be Bad The final response to this line of reasoning is to show that it is not actually an issue for the advocate of percentage willingness to pay at all. There is no reason anyone should object to real reforms to the process of cost-benefit analysis that will actually improve it, whether those focus on the KaldorHicks criterion of some other element of it. So long as the objector does not

go so far as to argue for the elimination of cost-benefit analysis, there is no need to object to new changes, and anyone who objects to the entirely of cost-benefit analysis will face the responses outlined in Chapter 6. In this way, general reforms to other components of cost-benefit analysis do not lead to reasons to either ignore the reforms of percentage willingness to pay or completely get rid of cost-benefit analysis. Even direct attacks on the specific method of using percentages instead of some other method fall flat because the case here is not that percentages are the best method, but rather that we should control for wealth when measuring willingness to pay. Percentages are merely one method of controlling for ability to pay. The enemy of absolute willingness to pay might easily admit that percentages were not the perfect solution to controlling for ability to pay in such measurements. The force of the argument is behind the need to enact such controls, not the specific mathematical process of making them. Unless the objection is focused on specifically the need to control for ability to pay, or is making the case that all methods of controlling for willingness to pay have some issue, there is no real objection to the central force of the argument presented here.   Objection 4: Baking-In Poverty Reduction A final objection comes neither from the economist nor the philosopher, but rather from the politician, who might object that such a framework “bakesin” poverty reduction and a focus on the interests of people over the interests of businesses. A given government or individual person may have different policy preferences. They might think that poverty reduction is important, or they might think that climate change, inflation, a strong military, civil rights, immigration, or any other number of policy positions are important. According to this objection, using percentage willingness to pay requires any policy we are implementing to focus disproportionately on the poor when in fact we may need to focus on the rich to achieve other goals. We might need to focus on giving tax cuts to the rich in order to create jobs, and percentage willingness to pay may understate these benefits. Additionally, this method seems to give preference to benefits that accrue to publicly traded companies over privately owned companies since there are many more people who benefit from the public stock of a

company increasing and may have any range of income or wealth levels, whereas a privately held company will likely be owned by very few people who are generally comparatively well off. This might therefore create incentives for the government to support publicly traded companies more than privately owned companies since the benefits to public companies accrue to more people. An extreme version of this concern is that this methodology requires any public program to prioritize the poor, to prioritize the worst off, and that is a bad thing. Some might make the claim that giving benefits to the poor over the rich is a political view. Some think that the rich deserve to gain more than the poor because the rich have worked hard to get where they are, whereas the poor simply live off of charity and do not deserve more. Such a viewpoint does not think that inequality is a symptom of unequal systems, but rather a necessary consequence of the unwillingness of some individuals to work hard. Under such a view, the rich should be prioritized, since they are the hard-workers in a society, the job-creators. While the poor are simply lazy, unwilling to work, and do not have their benefits overestimated, as that will only incentivize them to remain poor. To see the force of this objection, imagine two people, Anne Frances and Gipp. They both make the same amount of money, but work in different industries, Anne is a lawyer, Gipp is an accountant. Anne spends all her money frivolously on luxury goods, fancy alcohol, and expensive dinners. Gipp diligently saves most of his paycheck and lives on a very small portion of his income. Then, a recession hits, both Gipp and Anne Frances are about to lose their jobs if their firms don’t get government bailouts. The government is considering whether to bail out Gipp’s accounting agency or Anne’s law firm. Anne has only $10 to her name because she spends all of her money irresponsibly, so would benefit substantially in a percentage sense from a bailout. Gipp has twice his yearly salary saved ($200,000), so would benefit much less from a bailout in a percentage sense (imagine Gipp is willing to pay $20,000 to keep his job, and Anne is willing to pay $8). Using percentage willingness to pay, (assuming all the lawyers are like Anne, all accountants are like Gipp, and both firms have a similar number of employees), the government would bail out the lawyers because they were reckless and leave Gipp out to dry because he was responsible.

Not only would such a result be in some way unjust because it rewarded someone for acting frivolously and punished someone else for acting responsibly, it would incentivize others to act that way in the future. If people believe that the government is more likely to bail them out if they are destitute than if they save responsibly for a rainy day, no one will save for a rainy day, and the government will not be able to bail everyone out, leading to further economic collapse and therefore further suffering. From the point of view of this objection, percentage willingness to pay incentivizes poverty, and disincentivizes savings and building wealth, both of which are dangerous for the long-term stability of a society.   Response 4.1: Status Quo Bias The status quo has a policy goal that is much more baked-in, and switching to percentage willingness to pay does not create a new bias but instead corrects for the bias of the status quo. When a methodology calculates that the benefits of saving the life of a rich person, or reducing mortality risk for a rich person, are 10, 100, 1,000 times greater than saving the life of a poor person (or reducing the same mortality risk for a poor person) it is clear that the status quo has a deep bias toward overstating the benefits to the rich. The alternative of using percentage willingness to pay does not have a bias toward poverty reduction; it is returning to an equal balance of power. Certainly, it is more helpful to the poor than the current status quo, but that is only because the current status quo is so gratuitously biased against the poor. Percentage willingness to pay treats everyone equally: you can only bear a cost of up to 100% of your money. Absolute willingness to pay skews the scales toward the rich, allowing them to “feel more pain” than the poor. This solution does not bake in a bias toward the poor, it removes the already baked-in bias toward the rich.   Surely there are other priorities that one might be concerned with, but it is important to see that this methodology isn’t about poverty reduction but about correctly measuring the costs and benefits to the poor. Say you want to measure the impacts of inflation. Using absolute willingness to pay, you might think that a small increase in food prices would not have much of a real cost on the overall economy because the increase is small in absolute terms. Percentage willingness to pay shows that poor people have less to

start with, and so even small increases hit them harder. This is not about saying we should reduce poverty instead of tackling inflation, it is about correctly and accurately measuring the impact of inflation on everyone, rich and poor included. The same could be said of climate change. This methodology is not about choosing poverty reduction over reducing emissions. It’s about showing that the death of a poor person due to extreme weather is not hundreds of times less important than the death of a rich person. This methodology is about accuracy and justice in all analyses, not about prioritizing any one policy priority over another.   Response 4.2 Democracy vs Plutocracy We have a choice between valuing the benefits of everyone equally so that no one’s preferences can exceed 100%, or valuing the preferences of the wealthy much more. To put it simply, we have a choice between a democracy, where everyone gets a vote, and a plutocracy where only the wealthy get a vote. Absolute willingness to pay creates a plutocracy because the preferences of the wealthy can so outweigh the preferences of the poor. The life of a single wealthy person is judged to be so much more important than the lives of hundreds of poor people. The preferences of a rich person count for more, simply because they are able to pay more to get a policy enacted than a poor person is. Such a process is not a democratic process, it does not truly aggregate preferences of all people, it counts how much money you have, then gives more votes to people with more money. It is possible that the economist is actually trying to defend a plutocratic state, arguing that the rule of the wealthy is preferable to the rule of the poor unwashed masses, but I think this is unlikely. On its face, cost-benefit analysis proports to aggregate the costs and benefits to society of a policy, not just aggregate the preferences of the wealthy elite. So, either costbenefit analysis using absolute willingness to pay is a thinly veiled attempt to only consider the views of the wealthy and discount the views of the poor (while claiming to value the benefits and costs to everyone) or it is an inaccurate measure of true benefits and true costs to individuals. To fill the logical space, there are a range of concerns that most people would have with someone who responds by biting the bullet and claiming that plutocracy is a preferable system of government. There are any range of arguments in political philosophy that might be brought against such a

claim, from the argument that money does not give one the right to govern others, to the claim that such a government will cause more harm than a democracy since it will prioritize the needs of the privileged few over the poor many. One might conversely argue that such a system is contrary to the stated purpose of cost-benefit analysis, which attempts to measure benefits to everyone, without bias or preference to certain types of people. Even if a plutocracy is desirable, we should clearly state that the goal of cost-benefit analysis is to measure the costs and benefits to the wealthy. To truly see this, take the example offered in the objection comparing the benefits to a publicly traded company and a privately owned company. This harkens back to the example of Costgo and Mallmart from Chapter 13. The simply response is that businesses cannot feel. Businesses do not have moral standing. We care about businesses because of the benefits they can provide to people only, not because they are inherently valuable. Therefore, the comparison between one business and another is fallacious because businesses do not have feelings. Instead, we should compare the people who are actually impacted by those businesses: the workers, the owners, and the customers. Another way to think of this is that if a business closed down tomorrow and no one was harmed by it, all the employees got generous severance packages, the owners made a profit on the sale of its assets, and the customers found a better alternative source of the product, this would not be a bad thing. Businesses dying on their own is not bad. People are what matter. With respect to the specific question of whether this methodology would preference publicly traded companies over privately owned companies, the answer lies in the question: how many people does the company help? If a privately owned company benefits more people in a bigger way than a publicly traded one, it is more valuable. It is not only owners who benefit from a company but customers and employees too. A privately owned company that gives generous bonuses and creates real value for its consumers might very well create more benefit than a publicly traded company, whose stock is owned by comparatively few, and which pays its employees poorly and create sub-standard products. The point is not to prioritize one kind of firm over another but to prioritize people over firms.  

Response 4.3: CBAs Measure, They Don’t Judge The last three responses focus on the more extreme version of this objection, centered around the Gipp and Anne Frances thought experiment. This objection is particularly powerful because it attempts to make the case that wealth is in fact an ethically salient factor, essentially objecting to P3 from Chapter 9 that wealth is not ethically salient. In this example, Gipp’s wealth is due to his “good” actions of frugality, while Anne Frances’s poverty is due to her “bad” actions of spending frivolously. This response will focus on the question of whether this is the kind of ethically relevant factor that we should consider in a CBA. Response 4.4 will look at the empirical case that situations where differences in wealth are due to moral differences are few and far between and finally response 4.5 will examine the ethical claim that saving is good and spending is bad. The purpose of a cost-benefit analysis is to measure the costs and benefits to society, not to pass judgment on those receiving the benefits. Imagine that you are a prison warden at the Panopto State Penitentiary and you are deciding between refurbishing a wing for low-level offenders and refurbishing a wing for serial-killers and rapists. The purpose of the costbenefit analysis is to measure the benefit that will be felt by those individuals, not judge them for their crimes. It is ethically salient that the serial killers do not like their current cells and would like an improvement, while the low-level offenders are perfectly happy with their current accommodations, and would rather not be moved out to other wings for the time that improvements would take. But the moral history of the individuals in question cannot be taken into account by a cost-benefit analysis. Someone does not become less able to experience happiness because they have committed a crime, nor does someone who is more virtuous become more sensitive to costs.   Cost-benefit analyses have the purpose of measuring benefit, not determining if the benefit is deserved or not. From a utilitarian perspective, increasing happiness is the goal. It does not matter if the people whose happiness you are increasing did bad things in the past. Politicians may, and do, take past actions of individuals into account when setting policies, but a cost-benefit analysis is not the justice system. Its goal is to measure the benefits, not determine if those benefits are deserved. One could easily

imagine politicians who think all gay people are immoral, or all Christians are immoral claiming that cost-benefit analysis should not count if they accrue to whatever group is considered immoral. This clearly opens up the possibility for bias. Politicians may ultimately make decisions based on these biased beliefs, but they should not obscure those biases behind the façade of a supposedly objective cost-benefit analysis. Therefore, it should not matter to a cost-benefit analysis if Gipp is the most moral person in the world and Anne Frances is completely immoral. Requiring that cost-benefit analysis take into account the morality of the individuals receiving benefits would require comprehensive data on their lives and a sufficiently objective moral framework to judge their actions against, neither of which is easily obtainable. Not to mention the sheer impossibility of the government ethically gathering such data. Overall, we cannot use cost-benefit analysis to attempt to judge those who have been immoral, only to measure the benefits that may accrue to society overall. You might be able to capture some of the knock-on effects of such a policy and use those to determine action, but this would not prescribe a certain way to measure willingness to pay. For example, you might attempt to measure the impacts of incentivizing individuals to spend instead of save, and the potential future harm this would cause when considering whether to bail out the accounting firm or the law offices. It is possible that the costs of this incentivization would shift the cost-benefit analysis in the other direction, and considering all of the potential ramifications of a policy makes sense. However, this does not pose an issue for percentage willingness to pay, as such cost and benefits could be measuring using this framework just as easily as absolute willingness to pay.   Response 4.4: Wealth is Rarely Morally Relevant The example of Gipp and Anne Frances paints a very specific and unique picture, where Gipp was rich due to his own choices and Anne Frances was poor because of choices she made. Since they had the same salary, they could have made the choices of the other person and ended up in the other’s situation. One might argue that, even if practically we cannot measure the morality of everyone receiving benefits in a cost-benefit analysis, we can generally assume that the wealthy took the moral action of “saving” and the poor took the immoral action of “spending” (a spurious ethical claim we

will examine more in the last response). Even if you accept that saving is morally better than spending, and attempt to claim that we should take these past actions into account when using a cost-benefit analysis, if differences in wealth are generally due to moral actions, one might argue that we should give an advantage to the wealthy because they got their wealth by making more moral choices. Imagine two worlds: Meritoworld and Statictopia. In Meritoworld, everyone starts with nothing and is forced to make their own way in the world. Those who have wealth are those that succeed and act ethically, and those who have nothing are those that failed and were immoral. In Statictopia, everyone inherits their class status from their parents and retains that wealth and income level with few exceptions. It seems that few would claim that wealth is an ethically salient factor in Statictopia. Based on this, there are two questions to answer: whether the real world looks more like Statictopia or Meritoworld, and whether we would be justified in counting more benefits for the wealthy of Meritoworld. The answer to the first question is a fairly straightforward empirical one, though unfortunately the answer is that our world looks more like Statictopia. In the United States, two-thirds of class differences are attributable to family wealth.[162] And developing countries generally have even less intergenerational mobility than developed ones.[163] A much longer passage could be written on this, gathering evidence for the claim that the world is more static than meritocratic. However, one’s views on this may be more firmly built into a political ideology, and there is evidence that many people (particularly Americans) drastically overestimate economic mobility, so a full debate will need to await a future text.[164] However, even if you think that within your country there is economic mobility, it is much harder to argue that someone born into a family of subsistence farmers in a poor country in sub-Saharan Africa is poor because of the frivolous choices they made. For now, it seems that there is at least some evidence that the majority of inequality in the world is not driven by choices like the Gipp and Anne Frances example but by morally irrelevant features of the world, like your parents’ wealth or the country you were born in. Furthermore, even if the world was like Meritoworld, it is unclear why someone should be treated as more of a person for being more successful. Why should people who did not succeed be excluded from decision-making

processes? Why should we devalue the pain of those who might have done some immoral acts? In a pure meritocracy, there will be winners and losers. Some people may be poor because of their choices, but others will be poor because they lacked the skills to succeed. Should people who lack these skills be treated as subhuman, with their preferences and benefits ignored? What does the skill to make money have anything to do with your rights to be heard, or your ability to feel pain? The entire process smacks of dehumanization of those that are different from you. It seems immoral on its face to treat those who are weak or have a different moral code from you as less than human, as if their happiness or pain matters less than yours simply because you were more effective at gaining wealth. Even in an entirely meritocratic system, it is unclear that we should punish and devalue those who are struggling instead of lifting them up, helping them and treating them equally to everyone else.   Response 4.5: Saving is Not a Moral Good An implicit move in the Gipp and Anne Frances example is to assume that there is some morality attached to saving and spending, that Anne is a bad person for spending freely and Gipp is a good person for saving his money. In fact, some economists think that spending is quite good for an economy with weak demand.[165] And certainly, Anne Frances is assumedly making a great deal of happiness for herself by consuming her paycheck each week, while Gipp is quite possibly not as happy living on a shoestring budget. He might gain happiness from saving money and planning what to do with it, but it seems unlikely he is getting appreciably more happiness than Anne Frances. It is unclear why Anne Frances did something wrong deserving of punishment in stimulating the economy with her spending while Gipp did something good deserving of reward by saving. Maybe then the argument is not that saving is necessarily good but that Anne Frances should be punished so that she will save in the future. We are not punishing her because she did something wrong but rather because we want to influence her future behavior. However, this too does not stand up to scrutiny because it still assumes that saving is the goal. We are punishing Anne Frances with the goal of getting her to save more. But why do we want her to save more? Because saving is an inherently good activity? You might say that it is to help her avoid the future pain of a recession, but she

might simply be someone who is willing to accept a certain degree of risk with her actions. Who are we to say that her personal tolerance for risk is wrong? Instead of trying to discover what would lead one to think that this action is immoral, we can look at the opposite question. Why is it more moral to help Anne Frances than Gipp? The case here is clear, Anne Frances might suffer substantially without help, she might die. Gipp will be fine without government help. And if Gipp really valued his job more, he would have a higher willingness to pay to have his firm bailed out. Note that Gipp could have responded that he really did value this job and would be willing to pay $190,000 to keep his job. He would not actually need to pay that; this is just a measure of his preferences. If he is only willing to pay a small portion of his wealth to keep his job, it clearly does not matter that much to him, possibly because he has savings that he can live off of while he finds a new job. Gipp isn’t getting bailed out because he does not value that job much and Anne Frances values that job a lot (which we would expect, given that she has nothing without it). Overall, while a powerful objection, it seems that there are a range of reasons to think that percentage willingness to pay continues to be both a more accurate and a more just means of measuring benefit. The world is rarely meritocratic, cost-benefit analysis cannot measure how morally good the recipients are, and saving is not necessarily morally better than spending.  

           

17. FINAL THOUGHTS     How did we get to a point where these disciplines no longer communicate? Something is wrong with our underlying systems when such a clearly immoral conclusion—that it is better to save the life a of a single rich person over the lives of hundreds of poor people—goes unquestioned by the vast majority of economists and politicians, but the best philosophers can do is to write papers dismissing the entire process or the field of economics. Resolving this issue is not merely about starting to use percentage willingness to pay in place of absolute willingness to pay, but also about changing the fundamental way that the practice of economics, particularly in public policy, has become detached from its ethical underpinnings. Both philosophers and economists have a part to play in such a process. Economists must change their practice based on criticisms of their underlying principles on ethical grounds, and philosophers must propose solutions that can actually be implemented and operationalized in such a way as to resolve, not exacerbate, the underlying issue.   Let Philosophy In Economists need to allow philosophers to have a part in the economics departments of public organizations that have a duty to take ethical actions in the best interests of all. Philosophers have been critiquing cost-benefit analysis for decades but have been largely ignored.[166] The vast majority of the research staff at development organizations have a background in economics, to the complete exclusion of other fields.[167] This can lead such organizations toward groupthink and ignoring the important contributions of other disciplines. In my own experience of working for years in international development, for the Gates Foundation, the World Bank, and the U.S. government, organizations are focused on accomplishing their goals, and are deeply dismissive of any questions of whether those are the right goals to pursue. This focus on hypothetical imperatives at the expense of categorical

 

Philosophers might not want to burn the economics department down if economists took the concerns of philosophers into account.

ones, the desire to achieve a goal as efficiently as possible while maintaining a complete unwillingness to question that goal, brings to mind Hannah Arendt’s famous book Eichmann in Jerusalem: A Report on the Banality of Evil. Economists are in danger of repeating the mistakes of Eichmann, valuing certain lives more than others, when they only question how something can be done and exclude those who would ask whether it should be done in the first place. A philosopher of economics or applied ethics would make a fantastic addition to the economics department of any public organization. It is often perceived that philosophers might only deal with broader ethical questions while specific questions that engender broad disagreement are often left to politicians in public organizations. However, as seen here, there are ethical questions that have answers few would disagree with (it is not better to save one rich person’s life instead of the lives of 405 poor people), but require an intimate understanding of economic methodology to pinpoint. Such philosophers could assist in the development of improved economic methods of measuring benefits, as well as methods for embedding local autonomy into project development, or ensuring the epistemic justification for statistical claims made on shaky evidence. In developing such methodologies there are many ethically impactful choices that must be made that are often not apparent to the policymakers and may not be proscribed by politicians. Everyone should advocate for letting the study of ethics into international development organizations that make weighty ethical decisions every day.   Make Philosophy Practical Philosophers are, at least in part, to blame for the current state of affairs. If economists need to let philosophers into the room, philosophers need to acknowledge the practical impacts of their theories, the realistic constraints on the measurability of their proposed methods, and the political pressures that organizations face. We saw the impact of the first in the objections raised in Chapter 6. Advocating for a system that excludes data and experts will generally lead to reinforcing existing biases and power structures. People are rarely rational, so creating a system that will only work if decision-makers are perfectly rational and moral will inevitably fail. While data can be challenging to gather, and may imperfectly measure the

concepts we care about, it can also avoid psychological biases, and along with a sufficiently robust legal structure prevent decisions from being made, even by very biased decision-makers, in the face of evidence. Philosophers must also not make the perfect the enemy of the good. We may never be able to perfectly measure certain concepts like benefit. But that cannot lead to philosophers abdicating the task of trying to make a better measurement. As a philosophical skeptic,

 

Economists might be more likely to take philosophers seriously if philosophers acknowledge the real constraints of politics and measurement.

I suspend judgment on the truth of any given proposition. I don’t know if we can know anything. In this case, I don’t know if percentage willingness to pay is a good measure of benefit, or if any of the arguments I offer in this book are correct (see Appendix C for more). While I can suspend judgment, I cannot suspend action, and I do not give up on looking for knowledge. Philosophers in institutions of public policy must take the same tack with respect to measurement. We may never be able to measure something correctly, but that should not lead us to give up on ever measuring it, or attempting to find better ways to measure or understand these concepts. If we do, we are in danger of leaving important questions of what is right for a society to do in the hands of economists who are focused only on how to maximize benefits, and not on how those benefits are best defined. Percentage Willingness to Pay To wrap up, the solution proposed here for the immoral disparate valuation of statistical lives is to change from measuring benefit using absolute willingness to pay to measuring it using percentage willingness to pay, and then to not paper over real differences in preferences with averages. This will make the practice of cost-benefit analysis both more accurate and more just. When faced with the clearly impossibly immoral conclusion that rich countries have a moral duty to dump toxic waste on the shores of poor countries, there are simply three ways that we can respond. We can be an economist, who accepts this conclusion, arguing that the foundations of economics justify the claim that the rich simply have more capacity for happiness or suffering than the poor do. Such an economist favors institutionalizing a plutocracy, ruled by the rich regardless of the preferences of the poor. We can instead be a philosopher, who attempts to dismantle all empirical tools of measuring benefits in favor of potentially irrational processes easily prone to bias and reinforcing existing power structures. Such a philosopher naively thinks that a simple rational discussion between interested parties will resolve a disagreement and that those with power will not impose their will on those with less power. Or we can take a middle path, claiming that the system is not so flawed as to require that we throw the good out with the bad. We can instead embrace a methodology that makes the measurement of economic benefit both more accurate and more ethical. We can make a simple and small change to our methodology that avoids the ridiculous assumption that rich

people have a greater depth for suffering and pleasure than poor people. We can allow for ethically relevant differences in risk tolerance while avoiding irrelevant features like wealth. We can use empirical tools to measure mortality risk reduction in a way that does not lead to our concluding that some lives are worth more than others. We can find that among those with similar real risk tolerances all lives really are equal. I don’t know that this is the right path, but I am hopeful it is the one that we take.  





 

APPENDICES A

    Lawrence Summers Memo

B Viscusi and Masterman VSL Table C

Note on Skepticism

D Viscusi and Masterman VSL Calculations as Percentage Willingness To Pay E

Bell Curve Valley Costs and Benefits

F Tinyworld Costs and Benefits G  

Equihoma City Costs and Benefits

Pg 353 Pg 355 Pg 359 Pg 363 Pg 367 Pg 373 Pg 375

APPENDIX A: LAWRENCE SUMMERS MEMO     “DATE: December 12, 1991 TO: Distribution FR: Lawrence H. Summers Subject: GEP   'Dirty' Industries: Just between you and me, shouldn’t the World Bank be encouraging MORE migration of the dirty industries to the LDCs [Less Developed Countries]? I can think of three reasons: 1) The measurements of the costs of health impairing pollution depends on the foregone earnings from increased morbidity and mortality. From this point of view a given amount of health impairing pollution should be done in the country with the lowest cost, which will be the country with the lowest wages. I think the economic logic behind dumping a load of toxic waste in the lowest wage country is impeccable and we should face up to that.   2) The costs of pollution are likely to be non-linear as the initial increments of pollution probably have very low cost. I’ve always thought that under-populated countries in Africa are vastly UNDER-polluted, their air quality is probably vastly inefficiently low compared to Los Angeles or Mexico City. Only the lamentable facts that so much pollution is generated by non-tradable industries (transport, electrical generation) and that the unit transport costs of solid waste are so high prevent world welfare enhancing trade in air pollution and waste.   3) The demand for a clean environment for aesthetic and health reasons is likely to have very high income elasticity. The concern over an agent that causes a one in a million change in the odds of prostrate cancer is obviously going to be much higher in a country where people survive to get prostrate cancer than in a country where under 5 mortality is 200 per thousand. Also, much of the concern over industrial atmosphere discharge is about visibility impairing particulates. These discharges may have very little direct health

impact. Clearly trade in goods that embody aesthetic pollution concerns could be welfare enhancing. While production is mobile the consumption of pretty air is a non-tradable.   The problem with the arguments against all of these proposals for more pollution in LDCs (intrinsic rights to certain goods, moral reasons, social concerns, lack of adequate markets, etc.) could be turned around and used more or less effectively against every Bank proposal for liberalization.” [168]    

APPENDIX B: VISCUSI AND MASTERMAN VSL TABLE     VSL Country

USD

Afghanistan

105000

Albania

736000

Algeria

838000

Andorra

7444000

Angola

719000

Antigua and Barbuda

2283000

Argentina

2144000

Armenia

668000

Australia

10335000

Austria

8157000

Azerbaijan

1129000

Bahamas

3568000

Bahrain

3413000

Bangladesh

2496000

Belarus

1111000

Belgium

7613000

Belize

772000

Benin

145000 18261000

Bhutan

409000

Bolivia

516000

Bosnia and Herzegovina Botswana

                           

205000

Barbados

Bermuda

 

                 

803000 1111000

 

VSL Country

USD

Lebanon

1326000

Lesotho

220000

Liberia

65000

Lithuania

2570000

Luxembourg 13247000 Macao

11558000

Macedonia

884000

Madagascar

72000

Malawi

58000

Malaysia

1819000

Maldives

1196000

Mali

131000

Malta

4117000

Marshall Islands

821000

Mauritania

236000

Mauritius

1683000

Mexico

1671000

Micronesia

612000

Moldova

385000

Mongolia

666000

Montenegro

1242000

Morocco

521000

Mozambique

102000

Myanmar

200000

Brazil

1695000

Brunei

6539000

Bulgaria

1287000

Burkina Faso Burundi

110000 45000

Cabo Verde

564000

Cambodia

184000

Cameroon

227000

Canada C.A .Rep.

8179000 57000

Chad

151000

Chile

2426000

China

1364000

Colombia

1228000

Comoros

134000

D.R. Congo

71000

Congo, Rep.

437000

Costa Rica

1789000

Cote d'Ivoire

244000

Croatia

2185000

Cyprus

4471000

Czech Rep.

3121000

Denmark

10073000

Dominica

1170000

Dominican Rep.

                                                 

1074000

  Ecuador

1037000

Egypt

575000

El Salvador

678000

   

Namibia

893000

Nauru

2653000

Nepal

126000

Netherlands

8406000

New Zealand

6885000

Nicaragua

334000

Niger

67000

Nigeria

485000

Norway 16127000 Oman

2909000

Pakistan

248000

Palau

2095000

Panama

2044000

Papua New Guinea

385000

Paraguay

721000

Peru

1055000

Philippines

611000

Poland

2295000

Portugal

3532000

Puerto Rico

3324000

Qatar 14450000 Romania

1634000

Russia

1970000

Rwanda

120000

Samoa

676000

Sao Tome and Principe

303000

Saudi Arabia

4052000

Senegal

169000

2206000

 

Serbia

953000

Estonia

3159000

Seychelles

2539000

Eswatini

564000

Sierra Leone

107000

Ethiopia

102000

       

Singapore

8962000

Equatorial Guinea

Fiji Finland

831000 8009000

France

6975000

Gabon

1583000

Gambia, The

       

79000

Georgia

716000

Germany

7904000

Ghana

255000

Greece

3496000

         

Slovak Republic

3023000

Slovenia

3818000

Solomon Islands

330000

South Africa

1046000

South Sudan

136000

Spain

4908000

Sri Lanka

654000

St. Kitts and Nevis

2591000

St. Lucia

1265000

St. Vincent and the

Grenada Guatemala Guinea

1488000 618000 81000

GuineaBissau

102000

Guyana

704000

Haiti

139000

Honduras

392000

Hong Kong

7054000

Hungary

2233000

Iceland

8626000

India Indonesia

275000 592000

                     

Grenadines

1141000

Sudan

330000

Suriname

1610000

Sweden

9965000

Switzerland 14560000 Tajikistan

220000

Tanzania

158000

Thailand

984000

Timor-Leste

375000

Togo

93000

Tonga

736000

Trinidad and Tobago

3035000

Iran

1127000

Iraq

1001000

Ireland

9046000

Isle of Man

14674000

Israel

6154000

Italy

5645000

Jamaica

869000

Japan

6682000

Jordan

805000

Kazakhstan

1960000

Kenya

231000

Kiribati

583000

Korea, Rep.

4723000

Kosovo

683000

Kuwait

7252000

Kyrgyz Rep.

201000

Lao PDR

299000

Latvia

2577000

                                   

Tunisia

685000

Turkey

1712000

Turkmenistan

1270000

Tuvalu

1072000

Uganda

120000

Ukraine

454000

U.A.E.

7413000

United Kingdom

7465000

United States

9631000

Uruguay

2705000

Uzbekistan

372000

Vanuatu

545000

Vietnam

342000

West Bank and Gaza

532000

Yemen, Rep.

196000

Zambia

256000

Zimbabwe

148000

 

   

 

APPENDIX C: NOTE ON SKEPTICISM   Some who have followed my work online through the YouTube channel and website Carneades.org may be confused by what appears to be a specific position advanced here when I profess to be a philosophical skeptic. Even the central objection of the book— valuing rich lives more than poor lives—seems to presuppose many philosophical dogmas like moral realism. How can I remain a skeptic and present such arguments? The response is two-fold: the first reason relies on my particular brand of skepticism, the second is based on the difference between actions and beliefs. A full understanding of my version of skepticism is required to see why this book is not a betrayal of my lack of beliefs. A full defense and explanation of my position will need to await a future text (or can be found on my YouTube channel), but a short note should be sufficient to respond to this concern. I describe myself as an indirect skeptic, which is a variety of Pyrrhonian skepticism. In this context, I use “Pyrrhonian” skepticism to associate myself with the description of a skeptic as one who lacks beliefs, which is distinct from the academic skeptic who claims that knowledge is impossible. Following Sextus Empiricus, while the dogmatist claims that knowledge is possible, and the academic claims that knowledge is impossible, the Pyrrhonian skeptic affirms neither claim.[169] Unlike Sextus, the indirect skeptic does not actively seek a lack of beliefs but rather seeks belief systems that can stand up to scrutiny. In this pursuit, an indirect skeptic assumes various belief structures with the purpose of testing them. When those belief structures prove to lead to contradictions, or are unable to justify themselves, the skeptic steps out of these assumptions and reaffirms their lack of beliefs. You might think of this like a reductio ad absurdum, or an indirect logical proof, where assumptions are made with the goal of showing that those assumptions in turn lead to a contradiction. As the skeptic is unsure of their own reasoning or the law of the non-contradiction, even upon arriving at this conclusion we cannot affirm that the initial assumptions were false, but we can question them, doubt them, and be left in a state of ataraxia.

Using this framework, we can see how the arguments offered in this book are not out of line with indirect skepticism and are in fact an exemplification of it. I assume several premises that are generally assumed by those in public policy: that economics can represent the world in a way that helps us make decisions about the correct policies to make and that lives have similar values, and show that these lead to a contradiction. As the argument progresses through section II, we test each underlying assumption about the framework of economics and philosophy in turn to see which assumption is the problematic one that leads to the contradictory conclusion. Upon finding the offending assumption in section III, while I use strong language to defend this conclusion, I am not sure that I am right. I do not believe that this is the reason for the flaw. I might be mistaken, likely others will disagree with me and they might be right, but nor does it appear that the underlying assumption that absolute willingness to pay is an accurate or just measure of benefit can in any way be justified. Even given this, one might be concerned, as I do not merely offer a negative criticism in this book but offer a positive case for a particular methodology being more accurate and just than another. This certainly appears to be a philosophical position laden with assumptions and claims. However, this is largely as a rebuke of the economist’s claim that the status quo for cost-benefit analysis is not perfect, but it is the best possible system we can devise. I am not confident that using percentage willingness to pay is the best solution, and it is very possible that research could be done to better control for the relationship between wealth and benefit. However, it does seem that this is an improvement on the status quo, so there is a reason to doubt the claim that our current methods are the best. Another response to this critique can be found in the difference between actions and beliefs. The skeptic can suspend judgment on the truth of a given claim, and I do not believe that any claim presented in this book is certain knowledge, is justified, or is true. However, we cannot suspend action because not taking an action is itself an action. Public policy must be made even if it is imperfect, or insufficiently justified. We must make choices about the kinds of investments to make with public money with incomplete information. We must make decisions without knowledge. As a skeptic I don’t know that percentage willingness to pay is the best system. However, it seems to me that there are many persuasive arguments

outlining the failures of other systems. It may not be correct, but it seems less wrong than other systems. When it comes to belief, the skeptic can be a purist, but when it comes to action, we have no option to suspend action, so we must act even without confidence that our action is right or justified. I don’t know if I am a perfect skeptic. Perhaps I believe things that I do not realize that I believe. Perhaps I am deceived into feeling as if I have no beliefs when in fact I have many. Perhaps every claim I make in this book is wrong. I don’t know. But that does not mean I cannot try to find knowledge, or improve on imperfect decision rules for action, even if I don’t know if those decision rules have actually improved or not. A lack of knowledge is not an excuse for inaction, but it is good reason to suspend judgment.

APPENDIX D: VISCUSI AND MASTERMAN VSL CALCULATIONS AS PERCENTAGE WILLINGNESS TO PAY   These values are calculated by estimating the absolute value of reducing a 1 in 1000 risk for each country based on the Viscusi and Masterman calculations, then dividing these results by the country’s GNI per capita.   VSL   VSL % of Country

% of

GNI

Afghanistan

17.21%

Albania

17.20%

Algeria

17.21%

Andorra

17.20%

Angola

17.20%

Antigua and Barbuda

17.20%

Argentina

17.21%

Armenia

17.22%

Australia

17.20%

Austria

17.21%

Azerbaijan

17.21%

Bahamas

17.20%

Bahrain

17.20%

Bangladesh

17.23%

Barbados

17.20%

Belarus

17.20%

Belgium

17.20%

Belize

17.19%

Benin

17.26%

Bermuda

17.20%

Country

           

GNI

Lebanon 17.20% Lesotho 17.19% Liberia

17.11%

Lithuania 17.20% Luxembourg 17.20% Macao 17.20%

               

Macedonia 17.20% Madagascar 17.14% Malawi 17.06% Malaysia 17.21% Maldives 17.21% Mali 17.24% Malta 17.20% Marshall Islands 17.21%

           

Mauritania 17.23% Mauritius 17.21% Mexico 17.21% Micronesia 17.19% Moldova 17.19% Mongolia 17.21%

Bhutan

17.18%

Bolivia

17.20%

Bosnia and Herzegovina

17.19%

Botswana

17.20%

Brazil

17.21%

Brunei

17.20%

Bulgaria

17.21%

Burkina Faso

17.19%

Burundi

17.31%

Cabo Verde

17.20%

Cambodia

17.20%

Cameroon

17.20%

Canada

17.20%

C.A .Rep.

17.27%

Chad

17.16%

Chile

17.21%

China

17.20%

Colombia

17.20%

Comoros

17.18%

D.R. Congo

17.32%

Congo, Rep.

17.20%

Costa Rica

17.20%

Cote d'Ivoire

17.18%

Croatia

17.20%

Cyprus

17.20%

Czech Rep.

17.21%

Denmark

17.20%

Dominica

17.21%

Dominican

17.21%

     

Montenegro 17.20% Morocco 17.19% Mozambique 17.29%

         

Myanmar 17.24% Namibia 17.21% Nauru 17.20% Nepal 17.26% Netherlands 17.20%

                   

New Zealand 17.20% Nicaragua 17.22% Niger 17.18% Nigeria 17.20% Norway 17.20% Oman 17.20% Pakistan 17.22% Palau 17.20% Panama 17.21% Papua New Guinea 17.19%

         

Paraguay 17.21% Peru 17.21% Philippines 17.21% Poland 17.20% Portugal 17.20%

           

Puerto Rico 17.20% Qatar 17.20% Romania 17.20% Russia 17.21% Rwanda 17.14% Samoa 17.20%

Rep.

  Ecuador

17.20%

Egypt

17.22%

El Salvador

17.21%

Equatorial Guinea

17.21%

Estonia

17.21%

Eswatini

17.20%

Ethiopia

17.29%

Fiji

17.20%

Finland

17.21%

France

17.21%

Gabon

17.21%

Gambia, The

17.17%

Georgia

17.21%

Germany

17.21%

Ghana

17.23%

Greece

17.20%

Sao Tome and Principe 17.22%

     

Saudi Arabia 17.21% Senegal 17.24% Serbia 17.20%

       

Seychelles 17.20% Sierra Leone 17.26% Singapore 17.20% Slovak Republic 17.21%

   

Slovenia 17.21% Solomon Islands 17.19%

   

South Africa 17.20% South Sudan 17.22%

     

Spain 17.20% Sri Lanka 17.21% St. Kitts and Nevis 17.20%

   

St. Lucia 17.21% St. Vincent and the

Grenada

17.20%

Guatemala

17.21%

Guinea

17.23%

GuineaBissau

17.29%

Guyana

17.21%

Haiti

17.16%

Honduras

17.19%

Hong Kong

17.20%

Grenadines 17.21%

     

Sudan 17.19% Suriname 17.20% Sweden 17.20%

       

Switzerland 17.20% Tajikistan 17.19% Tanzania 17.17% Thailand 17.20%

Hungary

17.20%

Iceland

17.20%

India

17.19%

Indonesia

17.21%

Iran

17.21%

Iraq

17.20%

Ireland

17.20%

Isle of Man

17.20%

Israel

17.20%

Italy

17.21%

Jamaica

17.21%

Japan

17.20%

Jordan

17.20%

Kazakhstan

17.21%

Kenya

17.24%

Kiribati

17.20%

Korea, Rep.

17.21%

Kosovo

17.20%

Kuwait

17.21%

Kyrgyz Rep.

17.18%

Lao PDR

17.18%

Latvia

17.20%

       

       

Timor-Leste 17.20% Togo 17.22% Tonga 17.20% Trinidad and Tobago 17.21%

               

Tunisia 17.21% Turkey 17.21% Turkmenistan 17.21% Tuvalu 17.21% Uganda 17.14% Ukraine 17.20% U.A.E. 17.20% United Kingdom 17.20%

           

United States 17.20% Uruguay 17.21% Uzbekistan 17.22% Vanuatu 17.19% Vietnam 17.19% West Bank and Gaza 17.22%

       

Yemen, Rep. 17.19% Zambia 17.18% Zimbabwe 17.21%

 

 

APPENDIX E: BELL CURVE VALLEY

COSTS AND BENEFITS   Full Tables of the costs and benefits from the Bell Curve Valley Thought Experiment in Chapter 12.     Alice   Absolute Percentage Population Wealth

1,000 $1,000,000

Benefit From Risk

$0

0%

Cost From Taxes

$10,000

1%

Cost From Stocks

$50,000

5%

$0

0%

-$60,000

-6%

-$60,000,000

-6000%

Reduction

Cost From Layoffs Net Benefits/Costs Per Person Net Total Benefits/Costs

   

Brahma

  Population Wealth Benefit From Risk

Absolute

Percentage 19,000

$1,000,000 $0

0%

$10,000

1%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

Net Benefits/Costs

-$10,000

-1%

-$190,000,000

-19000%

Reduction Cost From Taxes

Per Person Net Total Benefits/Costs

 

       

Carlos

 

Absolute

Population

Percentage 4,000

Wealth

$100,000

Benefit From Risk Reduction

$0

0%

Cost From Taxes

$1,000

1%

Cost From Stocks

$1,000

1%

$0

0%

-$2,000

-2%

-$8,000,000

-8000%

Cost From Layoffs Net Benefits/Costs Per Person Net Total Benefits/Costs

   

DeShawn

 

Absolute

Population

Percentage 26,000

Wealth

$100,000

Benefit From Risk Reduction

$0

0%

$1,000

1%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

-$1,000

-1%

-$26,000,000

-26000%

Cost From Taxes

Net Benefits/Costs Per Person Net Total Benefits/Costs

   

Emily

  Population

Absolute

Percentage 30,000

Wealth

$100,000

Benefit From Risk Reduction

$0

0%

Cost From Taxes

$0

0%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

Person

$0

0%

Net Total Benefits/Costs

$0

0%

Net Benefits/Costs Per

   

Farrad

 

Absolute

Percentage

Population

5,000

Wealth

$100,000

Benefit From Risk Reduction

$17,000

17%

$1,000

1%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

$16,000

16%

$80,000,000

80000%

Cost From Taxes

Net Benefits/Costs Per Person Net Total Benefits/Costs

   

Gin

  Population Wealth

Absolute

Percentage 5,000

$100,000

Benefit From Risk Reduction

$17,000

17%

Cost From Taxes

$0

0%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

$17,000

17%

Net Benefits/Costs Per

Person Net Total Benefits/Costs

$85,000,000

85000%

   

Hopi

 

Absolute

Population

Percentage 1,000

Wealth

$10,000

Benefit From Risk Reduction

$1,700

17%

Cost From Taxes

$100

1%

Cost From Stocks

$100

1%

$10,000

100%

-$8,500

-85%

-$8,500,000

-85000%

Cost From Layoffs Net Benefits/Costs Per Person Net Total Benefits/Costs

         

Ingrid Absolute

Percentage

Population Wealth

4,000 $10,000

Benefit From Risk Reduction

$1,700

17%

$100

1%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

$1,600

16%

$6,400,000

64000%

Cost From Taxes

Net Benefits/Costs Per Person Net Total Benefits/Costs

     

Juan Absolute

Percentage

Population

5,000

Wealth

$10,000

Benefit From Risk Reduction

$1,700

17%

Cost From Taxes

$0

0%

Cost From Stocks

$0

0%

Cost From Layoffs

$0

0%

$1,700

17%

$8,500,000

85000%

Net Benefits/Costs Per Person Net Total Benefits/Costs

   

Total

 

Absolute

Population

Percentage 100,000

Total Benefits

$187,000,000

340,000%

Total Costs

$299,600,000

170,000%

Total Net Benefits

-$112,600,000

170,000%

          Demographics

 

Population Wealth

Alice

1,000

$1,000,000

Brahma

19,000

$1,000,000

Carlos

4,000

$100,000

DeShawn

26,000

$100,000

Farrad

5,000

$100,000

Emily

30,000

$100,000

Gin

5,000

$100,000

Hopi

1,000

$10,000

Ingrid

4,000

$10,000

Juan

5,000

$10,000

  Costs and Benefits Per Person

 

Absolute Absolute Percentage

Percentage

Benefits

Costs

Costs

Benefits

Alice

$0

$60,000

0%

6%

Brahma

$0

$10,000

0%

1%

Carlos

$0

$2,000

0%

2%

DeShawn

$0

$1,000

0%

1%

Farrad

$17,000

$1,000

17%

1%

Emily

$0

$0

0%

0%

Gin

$17,000

$0

17%

0%

Hopi

$1,700

$10,200

17%

102%

Ingrid

$1,700

$100

17%

1%

Juan

$1,700

$0

17%

0%

              Total Costs and Benefits

 

Absolute

Absolute

Percentage

Percentage

Benefits

Costs

Benefits

Costs

Alice

$0

$60,000,000

0%

6000%

Brahma

$0 $190,000,000

0%

19000%

Carlos

$0

$8,000,000

0%

8000%

DeShawn

$0

$26,000,000

0%

26000%

Farrad

$85,000,000

$5,000,000

85000%

5000%

Emily

$0

$0

0%

0%

Gin

$85,000,000

$0

85000%

0%

Hopi

$1,700,000

$10,200,000

17000%

102000%

Ingrid

$6,800,000

$400,000

68000%

4000%

Juan

$8,500,000

$0

85000%

0%

$187,000,000 $299,600,000

340000%

170000%

Total

     

APPENDIX F: TINYWORLD

COSTS AND BENEFITS   Full Tables of the costs and benefits from the Tinyworld Thought Experiment in Chapter 15. Using absolute willingness to pay, vaccines should be given to the rich, regardless of vulnerability. Using percentage willingness to pay vaccines should be given to those who are actually vulnerable, as shown by their percentage willingness to pay for the shots.   Absolute Willingness to Pay

 

Rich

Pop.

Inc.

WTP for

Net Total

Net Total

a vaccine

Benefit

Benefit

in free

w/gov.

market

Intervention

1

$1,000,000

$100,000

$90,000

$99,000

2

$1,000,000

$10,000

$39,600

$3,600

1

$1,000,000

$10,000

$0

$0

1

$50,000

$5,000

$0

$4,000

4

$50,000

$500

$0

$0

2

$10,000

$1,000

$0

$0

4

$10,000

$100

$0

$0

$129,600

$106,600

Vulnerable Rich Factory Owners Rich NonVulnerable NonOwner Middle Class Vulnerable Middle Class NonVulnerable Poor Vulnerable Poor NonVulnerable Total

 

15

 

 

    Percentage Willingness to Pay

 

Pop.

Inc.

WTP

Net

Net total

for a

total

benefit

vaccine

benefit

w/gov.

in free

Intervention

market Rich

1

$1,000,000

Vulnerable Rich

2

10%

9%

10%

1%

4%

0.4%

1%

0%

0%

10%

0%

8%

1%

0%

0%

10%

0%

0%

1%

0%

0%

13%

18%

$1,000,000

Factory Owners Rich Non-

1

$1,000,000

Vulnerable NonOwner Middle

1

$50,000

Class Vulnerable Middle

4

$50,000

Class NonVulnerable Poor

2

$10,000

Vulnerable Poor Non-

4

$10,000

Vulnerable Total



15

 

 

APPENDIX G: EQUIHOMA CITY

COSTS AND BENEFITS   Full tables of the costs and benefits from the Equihoma City thought experiment from Chapter 16. As these tables show, making decisions on the basis of absolute willingness to pay will create and exacerbate inequality. At the end of 12 months using and absolute CBA (first three tables), farmers have $2,100, while ranchers only have $450. However, using percentage willingness to pay (last three tables), both communities benefit alternatively from government policies.   Farmers Costs and Benefits Per Person, Absolute CBA

Assuming absolute willingness to pay is used to determine which policy is enacted, so the policy that benefits the farmers (policy 1) is always enacted. Mon.

Pop.

Wealth

Net

Net

Net %

Net %

Abs.

Abs.

Benefit

Benefit

Benefit

Benefit

Policy 1

Policy 2

Pol. 1

Policy 2

1

60

$1,000

$100

-$50

10%

-5%

2

60

$1,100

$100

-$50

9%

-5%

3

60

$1,200

$100

-$50

8%

-4%

4

60

$1,300

$100

-$50

8%

-4%

5

60

$1,400

$100

-$50

7%

-4%

6

60

$1,500

$100

-$50

7%

-3%

7

60

$1,600

$100

-$50

6%

-3%

8

60

$1,700

$100

-$50

6%

-3%

9

60

$1,800

$100

-$50

6%

-3%

10

60

$1,900

$100

-$50

5%

-3%

11

60

$2,000

$100

-$50

5%

-3%

12

60

$2,100

$100

-$50

5%

-2%

    Ranchers Costs and Benefits Per Person, Absolute CBA



Assuming absolute willingness to pay is used to determine which policy is enacted, so the policy that benefits the farmers (policy 1) is always enacted. Mon.

Net Abs. Benefit Pop.

Wealth

Pol. 1

Net

Net %

Net %

Abs.

Benefit

Benefit

Benefit

Policy 1

Policy 2

Policy 2

1

40

$1,000

-$50

$100

-5%

10%

2

40

$950

-$50

$100

-5%

11%

3

40

$900

-$50

$100

-6%

11%

4

40

$850

-$50

$100

-6%

12%

5

40

$800

-$50

$100

-6%

13%

6

40

$750

-$50

$100

-7%

13%

7

40

$700

-$50

$100

-7%

14%

8

40

$650

-$50

$100

-8%

15%

9

40

$600

-$50

$100

-8%

17%

10

40

$550

-$50

$100

-9%

18%

11

40

$500

-$50

$100

-10%

20%

12

40

$450

-$50

$100

-11%

22%

  Total Costs and Benefits, Full Population, Absolute CBA



Assuming absolute willingness to pay is used to determine which policy is enacted, so the policy that benefits the farmers (policy 1) is always enacted. Mon.

Net

Net

Net

Net

Policy

Absolute

Absolute

Percentage

Percentage

Chosen

Benefit

Benefit

Benefit

Benefit

Policy 1

Policy 2

Policy 1

Policy 2

1

$4,000

$1,000

400%

100%

Policy 1

2

$4,000

$1,000

335%

148%

Policy 1

3

$4,000

$1,000

278%

194%

Policy 1

4

$4,000

$1,000

226%

240%

Policy 1

5

$4,000

$1,000

179%

286%

Policy 1

6

$4,000

$1,000

133%

333%

Policy 1

7

$4,000

$1,000

89%

384%

Policy 1

8

$4,000

$1,000

45%

439%

Policy 1

9

$4,000

$1,000

0%

500%

Policy 1

10

$4,000

$1,000

-48%

569%

Policy 1

11

$4,000

$1,000

-100%

650%

Policy 1

12

$4,000

$1,000

-159%

746%

Policy 1

  Farmers Costs and Benefits Per Person, Percentage CBA



Assuming percentage willingness to pay is used to determine which policy is enacted, so sometimes policy 1 is enacted, sometimes policy 2 is. Mon.

Pop.

Wealth

Net

Net

Net %

Net %

Abs.

Abs.

Benefit

Benefit

Benefit

Benefit

Policy 1

Policy 2

Pol. 1

Policy 2

1

60

$1,000

$100

-$50

10%

-5%

2

60

$1,100

$100

-$50

9%

-5%

3

60

$1,200

$100

-$50

8%

-4%

4

60

$1,300

$100

-$50

8%

-4%

5

60

$1,250

$100

-$50

8%

-4%

6*

60

$1,350

$100

-$50

7%

-4%

7

60

$1,300

$100

-$50

8%

-4%

8

60

$1,400

$100

-$50

7%

-4%

9

60

$1,500

$100

-$50

7%

-3%

10

60

$1,450

$100

-$50

7%

-3%

11

60

$1,550

$100

-$50

6%

-3%

12

60

$1,500

$100

-$50

7%

-3%

  Ranchers Costs and Benefits Per Person, Percentage CBA



Assuming percentage willingness to pay is used to determine which policy is enacted, so sometimes policy 1 is enacted, sometimes policy 2 is.

Mon.

Pop.

Wealth

Net

Net

Net %

Net %

Abs.

Abs.

Benefit

Benefit

Benefit

Benefit

Policy 1

Policy 2

Pol. 1

Policy 2

1

60

$1,000

-$50

$100

-5%

10%

2

60

$950

-$50

$100

-5%

11%

3

60

$900

-$50

$100

-6%

11%

4

60

$850

-$50

$100

-6%

12%

5

60

$950

-$50

$100

-5%

11%

6*

60

$900

-$50

$100

-6%

11%

7

60

$1,000

-$50

$100

-5%

10%

8

60

$950

-$50

$100

-5%

11%

9

60

$900

-$50

$100

-6%

11%

10

60

$1,000

-$50

$100

-5%

10%

11

60

$950

-$50

$100

-5%

11%

12

60

$1,050

-$50

$100

-5%

10%

 

Total Costs and Benefits, Full Population, Percentage CBA

Assuming percentage willingness to pay is used to determine which policy is enacted, so sometimes policy 1 is enacted, sometimes policy 2 is. Mon.

Net

Net

Net

Net

Policy

Absolute

Absolute

Percentage

Percentage

Chosen

Benefit

Benefit

Benefit

Benefit

Policy 1

Policy 2

Policy 1

Policy 2

1

$4,000

$1,000

400%

100%

Policy 1

2

$4,000

$1,000

335%

148%

Policy 1

3

$4,000

$1,000

278%

194%

Policy 1

4

$4,000

$1,000

226%

240%

Policy 2

5

$4,000

$1,000

269%

181%

Policy 1

6*

7

Policy 2 $4,000

$1,000

222%

222%

*

$4,000

$1,000

262%

169%

Policy 1

8

$4,000

$1,000

218%

207%

Policy 1

9

$4,000

$1,000

178%

244%

Policy 2

10

$4,000

$1,000

214%

193%

Policy 1

11

$4,000

$1,000

177%

228%

Policy 2

12

$4,000

$1,000

210%

181%

Policy 1

  *Note that in month 6, using the percentage CBA, the percentage benefits are equal for each group are identical. In this example I choose policy 2 to be enacted, as the policymakers might be trying to conform to the KaldorHicks requirement that there are different winners and losers when possible, and the farmers have been the beneficiaries for 4 of the first 5 months.    

     

             

REFERENCES       Acemoglu, D., Chernozhukov, V., Werning, I., and Whinston, M. D. (2020). A multi-risk SIR model with optimally targeted lockdown. National Bureau of Economic Research, Working Paper 27102. https://www.nber.org/system/files/working_papers/w27102/revisions/ w27102.rev0.pdf?mod=article_inline Ackerman, F. & Heinzerling, L. (2004). Priceless. On Knowing the Price of Everything and the Value of Nothing. The New Press, New York, New York. Adams, J. (1996). Cost-benefit analysis: The problem, not the solution. The Ecologist, 26(1), 2–5. Al Jazeera (2021). Djokovic pictured with ex-commander of feared paramilitary unit. Al Jazeera. https://www.aljazeera.com/news/2021/9/21/djokovic-sparkscontroversy-after-meeting-drina-wolves-commander Alpert, G. P., MacDonald, J. M., and Dunham, R. G. (2005). Police suspicion and discretionary decision making during citizen stops. Criminology, 43(2), 407–434. Anderson, E. (1993). Value in Ethics and Economics. Cambridge, Massachusetts: Harvard University Press. Arendt, H. (1994). Eichmann in Jerusalem: a report on the banality of evil. New York, NY Penguin Books. Aristotle (1989). Nicomachean Ethics In A New Aristotle Reader Ackrill, J.L. (ed). Princeton University Press, Princeton, New Jersey. Ball, T. (1998). Green Political Philosophy. In Routledge Encyclopedia of Philosophy. Craig E. (ed). Vol. 4, 159–166 Barsky, R., Bound, J., Charles, K. K., and Lupton, J. P. (2002). Accounting for the black–white wealth gap: a nonparametric approach. Journal of the American statistical Association, 97(459), 663–673.

Baumhäkel, M., Kindermann, M., Kindermann, I., and Böhm, M. (2007). Soccer world championship: a challenge for the cardiologist. European heart journal, 28(2), 150–153. BBC (2022). Fulham fan dies after cardiac arrest during Blackpool game. BBC. https://www.bbc.com/sport/football/60185725 Beatty, A., Borkum, E., Leith, W., Henry, M., Berends, M. Null, C., and Ingwersen, N. (2020). MCC Indonesia Nutrition Project Impact Evaluation Final Report. Mathematica. https://www.mathematica.org/our-publications-andfindings/projects/indonesia-impact-evaluation-for-community-basedhealth-and-nutrition Beck, B. (2020). Policing Gentrification: Stops and Low-Level Arrests during Demographic Change and Real Estate Reinvestment. City & Community 19(1). https://journals.sagepub.com/doi/pdf/10.1111/cico.12473 Bill and Melinda Gates Foundation (2019) Homepage. Retrieved 26 March 2019 from https://www.gatesfoundation.org/ Blau, F. D. and Graham, J. W. (1990). “Black-White Differences in Wealth and Asset Composition,” Quarterly Journal of Economics 30, 321– 339 Boardman, A., Greenberg, D., Vining, A., and Weimer, D. (2014). CostBenefit Analysis Concepts and Practice. Essex, United Kingdom: Pearson Education Limited. Boyce J.K. (2001). From Natural Resources to Natural Assets. NEW SOLUTIONS: A Journal of Environmental and Occupational Health Policy. 11(3):267-288. doi:10.2190/5QPY-TPE0-JP5W-5FJE Bratton, W. J. and Kelling, G. L. (2015). Why We Need Broken Windows Policing. City Journal USA 2014(12). Cameron, T. A. (2010). Euthanizing the Value of a Statistical Life. Review of Environmental Economics and Policy, 4(2):161–178. Carmichael, F. and Goodman, J. (2020). Vaccine rumours debunked: Microchips, “altered DNA” and more. BBC. https://www.bbc.com/news/54893437

Chappell, T. and Crisp, R. (1998). Utilitarianism. In Routledge Encyclopedia of Philosophy. Craig, E. (ed). Charles, K. K. and Hurst, E. (2002). The Transition to Home Ownership and the Black-White Wealth Gap. The Review of Economics and Statistics 84 (2), 281–297. Chavas, J. P. (2017). On food security and the economic valuation of food. Food Policy, 69, 58-67. https://doi.org/10.1016/j.foodpol.2017.03.008 Choi, J. H. (2020). Breaking Down the Black-White Homeownership Gap. Urban Institute. https://www.urban.org/urban-wire/breaking-downblack-white-homeownership-gap Choi, J. H., McCargo, A., Neal, M., Goodman, L., and Young, C. (2019). Explaining the Black-White homeownership gap: A closer look at disparities across local markets. Urban Institute research report. Ciepley, D. (2017). Member corporations, property corporations, and constitutional rights. Law and Ethics of Human Rights, 11(1), 31–59 Cole, B. (2020). GOP Lieutenant Governor Faces Backlash for Saying Grandparents Don't Want to Sacrifice the Economy for Coronavirus Isolation. Newsweek. https://www.newsweek.com/gop-lieutenantgovernor-faces-backlash-saying-grandparents-dont-want-sacrificeeconomy-1493883 da Cunha, S. B. (2016). A review of quantitative risk assessment of onshore pipelines. Journal of Loss Prevention in the Process Industries, 44, 282-298. Daley, S. (2015). Speeding in Finland Can Cost a Fortune, if You Already Have One. New York Times. Retrieved 28 May 2019 from https://www.nytimes.com/2015/04/26/world/europe/speeding-infinland-can-cost-a-fortune-if-you-already-have-one.html Darity, W., Hamilton, D., Paul, M., Aja, A., Price, A., Moore, A., and Chiopris, C. (2018). What We Get Wrong About Closing the Racial Wealth Gap. Samuel DuBois Cook Center on Social Equity & Insight Center for Community Economic Development. http://narrowthegap.org/images/documents/Wealth-Gap—FINALCOMPLETE-REPORT.pdf Davidai, S. (2018). Why do Americans

believe in economic mobility? Economic inequality, external attributions of wealth and poverty, and the belief in economic mobility. Journal of Experimental Social Psychology, 79, 138–148. Davidai, S., and Gilovich, T. (2018). How should we think about Americans’ beliefs about economic mobility?. Judgment & Decision Making, 13(3). Delehanty, C., Mewhirter, J., Welch, R., and Wilks, J. (2017). Militarization and police violence: The case of the 1033 program. Research & politics, 4(2), 2053168017712885. Di, Z. X., Belsky, E., and Lui, X. (2007). Do homeowners achieve more household wealth in the long run? Journal of Housing Economics. 16(3–4), 274–290. Dreher, A. (2002). The development and implementation of IMF and World Bank conditionality. Available at SSRN 333960. Eck, J. E., and Maguire, E. (2000). Have changes in policing reduced violent crime? In A. Blumstein and J. Wallman (Eds.), The crime drop in America, 207–265. New York: Cambridge University Press. Edwards, S. B., and Harris, D. (2015). Black lives matter. ABDO. Empiricus, S. (1990). Outlines of Pyrrhonism. Prometheus Books. EPA (2018) Mortality Risk Valuation. United States Environmental Protection Agency. Retrieved 22 March 2019 from https://www.epa.gov/environmental-economics/mortality-riskvaluation Fletcher, T., and Hayes-Birchler, A. (2020, July 30). Comparing Measures of Internet Censorship: Analyzing the Tradeoffs between Expert Analysis and Remote Measurement. Data for Policy 2020. https://doi.org/10.5281/zenodo.3967398 Greenstrone, M. and Nigam, V. (2020). Does Social Distancing Matter? Becker Friedman Institute. https://bfi.uchicago.edu/wpcontent/uploads/BFI_WP_202026.pdf Guardian (2022) Key moments in Novak Djokovic’s Australian saga. The Guardian. https://www.theguardian.com/sport/2022/jan/16/key-moments-innovak-djokovic-australian-saga Harcourt, B. E., and Ludwig, L.

(2006). Broken windows: New evidence from New York City and a five-city social experiment. University of Chicago Law Review, 73, 271–320 Harcourt, B. (2001). Illusion of Order. The False Promise of Broken Windows Policing. Harvard University Press. Hartmann, C. (2017). ECOWAS and the Restoration of Democracy in The Gambia. Africa Spectrum, 52(1), 85–99. Headley, A. M. (2021). Accountability and police use of force: Interactive effects between minority representation and civilian review boards. Public Management Review, 1–23. Heller, S. B., Jacob, B. A., and Ludwig, J. (2011). Family income, neighborhood poverty, and crime. Making Crime Control Pay: CostEffective Alternatives of Incarceration, 419–459. Hernandez, D. (2017). Are “new” donors challenging World Bank conditionality? World Development, 96, 529–549. Hessen, R. (1978). In defense of the corporation. Stanford: Stanford Hoover Institution Press. Hinkle, J.C. and Weisbund, D. (2008). The irony of broken windows policing: A micro-place study of the relationship between disorder, focused police crackdowns and fear of crime. Journal of Criminal Justice 36(6), 503–512. https://www.sciencedirect.com/science/article/abs/pii/S004723520800 1128 Hooker, B. (2005) Consequentialism. In The Oxford Companion to Philosophy New Edition. Honderich, T. (ed.). Imran, M., Hosen, M., and Chowdhury, M. A. F. (2018). Does poverty lead to crime? Evidence from the United States of America. International Journal of Social Economics. IRS (2017) Internal Revenue Service Data Book. Publication 55B. Washington, DC Retrieved 22 March 2019 from https://www.irs.gov/pub/irs-soi/17databk.pdf

Jaworska, A. and Tannenbaum, J. (2021) “The Grounds of Moral Status,” The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.), https://plato.stanford.edu/archives/spr2021/entries/grounds-moralstatus/. Jensen, M. and Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3, 305–360 John, S. D. and Curran, E. J. (2021). Costa, cancer and coronavirus: contractualism as a guide to the ethics of lockdown. Journal of Medical Ethics. 0(1-8). doi: 10.1136/medethics-2020-107103. Johnson, V. B. (2019). KKK in the PD: white supremacist police and what to do about it. Lewis & Clark L. Rev., 23, 205. Katz, C. M., Webb, V. J., and Schaefer, D. R. (2001). An assessment of the impact of quality-of-life policing on crime and disorder. Justice Quarterly, (18), 291­–389. https://www.tandfonline.com/doi/abs/10.1080/07418820100095111 Kraus, M. W., and Tan, J. J. (2015). Americans overestimate social class mobility. Journal of Experimental Social Psychology, 58, 101–111. Kuflik, A. (1998). Moral Standing. In Routledge Encyclopedia of Philosophy. Craig, E. (ed). vol. 6, 550–555. Leeka, J., Schwartz, B. G., and Kloner, R. A. (2010). Sporting events affect spectators’ cardiovascular mortality: it is not just a game. The American journal of medicine, 123(11), 972–977. Legewie, J. and Fagan, J. (2019). Aggressive Policing and the Educational Performance of Minority Youth. American Sociological Review. 84(2): 220–247. doi:10.1177/0003122419826020 List, C., and Pettit, P. (2011). Group agency: The possibility, design, and status of corporate agents. Oxford: Oxford University Press. Loofbourow, L. (2021). The Unbelievable Grimness of HermanCainAward, the Subreddit That Catalogs Anti-Vaxxer COVID Deaths. Slate. https://slate.com/technology/2021/09/hermancainaward-subreddit-

antivaxxer-deaths-cataloged.html Manning, R. C. (1984). Corporate responsibility and corporate personhood. Journal of Business Ethics, 3(1), 77–84. Masera, F. (2021). Police safety, killings by the police, and the militarization of US law enforcement. Journal of Urban Economics, 124, 103365. Maturana, M. A., Glover, E. A., Raja, J., Dornbush, S. R., Alexander, J., Blount, C., Khouzam, N.R., Khouzam, A.R., and Khouzam, R. N. (2021). Are Die-Hard Football or Other Sports Fans at Risk of Cardiovascular Events?. Current Problems in Cardiology, 46(3), 100743. McIntosh, K., Moss, E., Nunn, R., and Shambaugh, J. (2020). Examining the Black-white wealth gap. Brookings Institute. https://www.brookings.edu/blog/up-front/2020/02/27/examining-theblack-white-wealth-gap/ Messner, S. F., Galea, S., Tardiff, K. J., Tracy, M., Bucciarelli, A., Piper, T. M., et al. (2007). Policing, drugs and the homicide decline in New York City in the 1990s. Criminology, 45, 385−414. Millennium Challenge Corporation (2022). About MCC. Millennium Challenge Corporation. https://www.mcc.gov/about Millennium Challenge Corporation (2021). Cost Benefit Analysis Guidelines. Millennium Challenge Corporation. https://www.mcc.gov/resources/doc/cost-benefit-analysisguidelines#annex-1-value-of-a-statistical-life, https://www.mcc.gov/resources/doc-pdf/cost-benefit-analysisguidelines Millennium Challenge Corporation (2020a) Economic Rates of Return. Millennium Challenge Corporation. https://www.mcc.gov/our-impact/err Millennium Challenge Corporation (2020b) Zambia: Lusaka Water Supply, Sanitation, and Drainage Project Economic Rate of Return. Millennium Challenge Corporation. https://www.mcc.gov/where-we-work/err/zambiacompact Millennium Challenge Corporation (2007) Mongolia Compact Economic Rate of Return Health Project. Millennium Challenge Corporation. https://www.mcc.gov/where-wework/err/mongolia-compact Mitnik, Bryant, and Grusky (2018). A

Very Uneven Playing Field: Economic Mobility in the United States. Stanford Center on Poverty and Inequality. https://web.stanford.edu/~pmitnik/Mitnik_Bryant_Grusky_2018_eco n_mob_wp.pdf Mokhiber, R. and Weissman, R. (1999). Memo Misfire: World Bank “Spoof” Memo on Toxic Waste Holds More Irony Than Laughs. San Francisco Bay Guardian. https://archive.globalpolicy.org/socecon/bwi-wto/sumers99.htm Morrow, W. J. and Shjarback, J. (2019). Police Worldviews, Unconscious Bias, and their Potential to Contribute to Racial and Ethnic Disparities in NYPD Stops for Reason of “Furtive Movement”. Journal of Ethnicity in Criminal Justice, 17(3): 1-32. 10.1080/15377938.2019.1636920. Moyo, D. (2009). Dead Aid. Farrar, Straus and Giroux. New York, NY. Narayan, A., Van der Weide, R., Cojocaru, A., Lakner, C., Redaelli, S., Mahler, D. G., Ramasubbaiah, R. G. N., and Thewissen, S. (2018). Fair Progress? Economic Mobility across Generations around the World. World Bank Group. https://www.worldbank.org/en/topic/poverty/publication/fairprogress-economic-mobility-across-generations-around-the-world New York Times (1992). Furor on Memo at World Bank. New York Times Archives. https://www.nytimes.com/1992/02/07/business/furoron-memo-at-world-bank.html Novak, K., Hartman, J., Holsinger, A., and Turner, M. (1999). The effects of aggressive policing of disorder on serious crime. Policing, 22, 171–190 Oliver, M. L., and Shapiro, T.M. (1995) Black Wealth/White Wealth: A New Perspective on Racial Inequality London: Routledge. Osborne, S. (2019). Water Supply and Sanitation Sector Cost Benefit Analysis Guidance. Millennium Challenge Corporation. https://www.mcc.gov/resources/doc/water-sector-cost-benefitguidance#4-estimation-of-mcc-wash-project-costs Ozar, D. (1985). Do corporations have moral rights? Journal of Business Ethics, 4(4), 277–281. Passavant, P. A. (2015). I can’t breathe: Heeding the call of justice. Law, Culture and the Humanities, 11(3), 330–339.

Pietsch, B. (2021) Horse owners can’t find ivermectin as Americans flock to unproven coronavirus cure. Washington Post. https://www.washingtonpost.com/health/2021/09/19/ivermectinhorse-dewormer-owners-covid/ Pinsker, J. (2015). Finland, Home of the $103,000 Speeding Ticket. The Atlantic. https://www.theatlantic.com/business/archive/2015/03/finland-homeof-the-103000-speeding-ticket/387484/ Polasky, S., and Segerson, K. (2009). Integrating ecology and economics in the study of ecosystem services: some lessons learned. Annu. Rev. Resour. Econ., 1(1), 409–434. Portney, P. R. (2008). Benefit-Cost Analysis. In The Concise Encyclopedia of Economics. Henderson D.R. (ed.), 38–40. Prenzler, T., Porter, L., and Alpert, G. P. (2013). Reducing police use of force: Case studies and prospects. Aggression and Violent Behavior, 18(2), 343–356. Proulx, G. and Crane, N.J. (2019). “To see things in an objective light”: the Dakota Access Pipeline and the ongoing construction of settler colonial landscapes, Journal of Cultural Geography, DOI: 10.1080/08873631.2019.1665856 Raitzer, D. A., Lavado, R. F., Rabajante, J., Javier, X., Garces, L., and Amoranto, G. (2020). Cost–Benefit Analysis of Face-to-Face Closure of Schools to Control COVID-19 in the Philippines. Asian Development Bank https://dx.doi.org/10.22617/BRF200405-2 Rao, V., and Woolcock, M. (2007) The Disciplinary Monopoly in Development Research at the World. Global Governance: A Review of Multilateralism and International Organizations, 13, 479–484. https://heinonline.org/HOL/LandingPage? handle=hein.journals/glogo13&div=43&id=&page= Rawls J. (1971). A Theory of Justice. The Belknap Press of Harvard University Press Cambridge Massachusetts.

Raz, J. (1986). The Morality of Freedom. Oxford, New York: Oxford University Press. Robinson, L. A., Hammitt, J. K., and O’Keefe, L. (2018) Valuing Mortality Risk Reductions in Global Benefit ‐ Cost Analysis. Guidelines for Benefit-Cost Analysis Project. Robinson, L.A., Sullivan, R. and Shogren, J.F. (2021), Do the Benefits of COVID-19 Policies Exceed the Costs? Exploring Uncertainties in the Age–VSL Relationship. Risk Analysis, 41: 761–770. https://doi.org/10.1111/risa.13561 Rodriguez, A. (2020) “Texas’s lieutenant governor suggests grandparents are willing to die for US economy.” USA Today. https://www.usatoday.com/story/news/nation/2020/03/24/covid-19texas-official-suggests-elderly-willing-die-economy/2905990001/ Rose, S. and Wiebe, F. (2015) An Overview of the Millennium Challenge Corporation. Center for Global Development. https://www.cgdev.org/sites/default/files/overview-mcc-brief.pdf Rosenfeld, R., Fornango, R., and Rengifo, A. F. (2007). The impact of order-maintenance policing on New York City homicide and robbery rates: 1988-2001. Criminology, 45, 355−384. Sagoff, M. (2004) Price, Principle, and the Environment. Cambridge University Press. Sahm, C. (2019) Direct Stimulus Payments to Individuals. Brookings Institute. https://www.brookings.edu/wpcontent/uploads/2019/05/ES_THP_Sahm_web_20190506.pdf Scanlon, T. M. (2000). What We Owe to Each Other. Belknap Press of Harvard University. Scioto Analysis (2020) Closing Schools for Covid-19. https://static1.squarespace.com/static/5bdb6f642714e55b84ebe507/t/5 ef0e19b03e56c3a4a905c46/1592844726518/schoolclsing.pdf Shapiro, T., Meschede, T., and Osoro, S. (2013). The roots of the widening racial wealth gap: Explaining the black-white economic divide. Institute on Assets and Social Policy.

Sharp, E. B., and Johnson, P. E. (2009). Accounting for variation in distrust of local police. Justice Quarterly, 26(1), 157–182. Sherman, L. W. (1990). Police crackdowns: Initial and residual deterrence. In M. Tonry & N. Morris (Eds.), Crime and justice: A review of research. 12 1−48. Chicago: University of Chicago Press. Silver, K. (2019). Can a corporation be worthy of moral consideration? Journal of Business Ethics, 159(1), 253–265. Singer, P. (1972). Famine, Affluence, and Morality. Philosophy and Public Affairs, 1(1):229–243. Sinnott-Armstrong, W. (2021). Consequentialism. In The Stanford Encyclopedia of Philosophy. Zalta E. N. (ed.). https://plato.stanford.edu/archives/fall2021/entries/consequentialism/ Slote, M. (2005) Utilitarianism. In The Oxford Companion to Philosophy New Edition. Honderich, T. (ed.). Smith, M. R., Rojek, J. J., Petrocelli, M., and Withrow, B. (2017). Measuring disparities in police activities: A state of the art review. Policing: An International Journal of Police Strategies & Management. Steinberger, M. (2022). Novak Djokovic, a Master on the Court, Keeps Making Errors Off It. New York Times. https://www.nytimes.com/2022/01/15/sports/tennis/novak-djokovicaustralian-open.html?smid=url-share Strike P.C., Steptoe A. (2005) Behavioral and emotional triggers of acute coronary syndromes: a systematic review and critique. Psychosom Med. (67), 179–186 Sunstein, C. (2018) The Cost Benefit Revolution. The MIT Press. Sunstein, C. (2004). Valuing Life: A Plea for Disaggregation. Duke Law Journal, 54(2):385–445. Tatem Jr, R. M. (2021). Social Work Policing: An Embedded Autonomous Model (Doctoral dissertation, Arizona State University). Thaler, R. and Rosen S. (1976). The Value of Saving a Life: Evidence from the Labor Market. National Bureau of Economic Research.

Thorbecke, (2016). Why a Previously Proposed Route for the Dakota Access Pipeline Was Rejected. ABC News. https://abcnews.go.com/US/previously-proposed-route-dakota-accesspipeline-rejected/story?id=43274356 Thunstrom, L., Newbold, S., Finnoff, D. and Ashworth, M. and Shogren, J. F. (2020). The Benefits and Costs of Using Social Distancing to Flatten the Curve for COVID-19. Forthcoming Journal of Benefit-Cost Analysis, Available at SSRN: https://ssrn.com/abstract=3561934 or http://dx.doi.org/10.2139/ssrn.3561934 Tilley, A. and Jenkins, E. (2020). Aid Transparency Index 2020. Publish What You Fund. https://www.publishwhatyoufund.org/theindex/2020/ Toossi, M. (2002). Consumer spending: an engine for US job growth. Monthly Lab. Rev., 125, 12. Trochmann, M. B., and Gover, A. (2016). Measuring the impact of police representativeness on communities. Policing: An International Journal of Police Strategies & Management. United Nations (2020). Contributions received for 2020 for the United Nations Regular Budget. United Nations. Retrieved 28 December 2020 from https://www.un.org/en/ga/contributions/honourroll.shtml United Nations (1945) The Charter of the United Nations. Retrieved 22 March 2019 from https://www.un.org/en/charter-unitednations/index.html Valette, J. (1999) Larry Summers’ War Against the Earth. CounterPunch Global Policy Forum. https://archive.globalpolicy.org/socecon/envronmt/summers.htm Viscusi, W. K., and Masterman, C. J. (2017). Income Elasticities and Global Values of a Statistical Life. Journal of Benefit Cost Analysis, 8(2):226–250. Viscusi, W. K. (2009). The devaluation of life. Regulation & Governance, 3:103­–126.

Wacquant, L. (2009) Punishing the Poor: The Neoliberal Government of Social Insecurity

Durham, NC: Duke University Press, 2009, 84–85. Wacquant, L. (2008) Urban Outcasts: A

Comparative Sociology of Advanced Marginality Malden, MA: Polity, 80–81. Wainer, A. and Zabel, J. (2020). Homeownership and wealth accumulation for low-income households. Journal of Housing Economics. 47. Wenz P.R. (2005). Environmental Ethics. In Encyclopedia of Philosophy Borchert D. (ed.) vol. 3 ed. 2, 258–261. Whirled Bank (2001). Lawrence Summers Memo. Retrieved 26 March from http://www.whirledbank.org/ourwords/summers.html World Bank Data (2019). Population, Total. Retrieved 22 March 2019 from https://data.worldbank.org/indicator/sp.pop.totl. World Economic Outlook (2020). General Government Total Expenditure. International Monetary Fund. Retrieved 28 December 2020 from https://www.imf.org/en/Publications/WEO/weodatabase/2020/October Worrall, J. L. (2002). Does “broken windows” law enforcement reduce serious crime? California Institute for County Government (CICG) Capital Office. https://www.ojp.gov/ncjrs/virtual-library/abstracts/does-brokenwindows-law-enforcement-reduce-serious-crime Xu, J., Murphy, S. L., Kochanek, K. D., Bastian, B, and Arias, E. (2018). Deaths: Final Data for 2016. National Vital Statistics Reports, 67(5). Zhao, H., Feng, Z., and Castillo-Chavez, C. (2014). The dynamics of poverty and crime. Journal of Shanghai Normal University (Natural Sciences· Mathematics), 43(5), 486–495.

   

 

               

DISCLAIMERS     The views expressed herein are those of the author and should not be construed as an express or implied endorsement of those views by the World Bank, the United Nations, the Gates Foundation, nor the U.S. government or any of its agencies. Unless explicitly stated, the thought experiments in this book are works of fiction for the purpose of elucidating philosophical consequences. Any similarity to actual persons, living or dead, organizations, corporations, sovereign entities, or actual events, is purely coincidental.  







ABOUT THE AUTHOR   Carneades is the creator of the educational philosophy YouTube channel Carneades.org, as well as an international development practitioner with years of experience bringing philosophical analysis to bear on the myriad problems in global development. The channel was founded with the goal of repaving the agora with the rubble of the ivory tower, making philosophy and philosophical debate more accessible for a broader audience, and championing philosophical skepticism. Within global development, Carneades has extensive experience working on World Bank projects, Gates Foundation research, and in both policy and project implementation for multiple U.S. government development agencies.

    [1]

Ackerman and Heinzerling, 2004 p. 150; Whilred Bank, 2001; New York Times, 1992; Valette

1999. [2]

Ackerman and Heinzerling, 2004; Mokhiber and Weissmann 1999.

[3]

Whilred Bank, 2001; Valette 1999; see Appendix A for the complete memo.

[4]

Ackerman and Heinzerling, 2004, pp. 73–74.

[5]

Robinson, L. A., Hammitt, J. K., and O’Keefe, L., 2018; Millennium Challenge Corporation,

2021; EPA, 2018. [6]

Viscusi and Masterman, 2017, p. 226.

[7]

Viscusi and Masterman, 2017, pp. 245–247

[8]

Bill and Melinda Gates Foundation, 2019.

[9]

Robinson, L. A., Hammitt, J. K., and O’Keefe, L., 2018.

[10]

Millennium Challenge Corporation, 2021; EPA, 2018.

[11]

For the purposes of this thought experiment, we use a simplified version that ignores facts like

the ability of disease to cross borders and the wide range of public and private institutions that are able to implement public health restrictions and research vaccines. Chapter 15 discusses a more realistic version, focusing on the implications of these theories and the solutions offered here for the Covid-19 pandemic. [12]

World Bank Data, 2019; Xu et al., 2018.

[13]

World Bank Data, 2019.

[14]

Estimates in line with Viscusi and Masterman, 2017.

[15]

EPA, 2018.

[16]

See Appendix C for a note on how this view meshes with the philosophical skepticism I present

on Carenades.org. [17]

Sinnott-Armstrong, 2019; Hooker, 2005.

[18]

Slote, 2005; Chappell and Crisp, 1998.

[19]

Boardman et al., 2014; Portney, 2008.

[20]

Sunstein, 2018, pp. 21–25.

[21]

Sinnott-Armstrong 2019; Hooker 2005; Chappell and Crisp, 1998.

[22]

Sagoff, 2004; Anderson, 1993.

[23]

Boardman et al., 2014; Portney 2008.

[24]

Boardman et al., 2014; Portney 2008.

[25]

Chavas, 2017; Polasky and Segerson, 2009; Boyce, 2001; Adams 1996.

[26]

Sunstein, 2018; Portney, 2008.

[27]

Boardman et al., 2014, pp. 1–2.

[28]

Sunstein, 2018, pp. 21–25.

[29]

Note that many authors claim that this is merely one tool among many to help inform policy

decisions, but still should have a bearing on the decisions to be made (e.g., Portnoy 2008 “While not intended to be the only basis for decision making, BCA can be a valuable aid to policymakers”). [30]

Boardman et al., 2014, pp. 1–2; Portnoy, 2008, p. 38.

[31]

Boardman et al., 2014, pp. 28–29.

[32]

Boardman et al., 2014, p. 33.

[33]

Millennium Challenge Corporation, 2020a.

[34]

Wenz, 2005; Ball, 1998.

[35]

Boardman et al., 2014, p. 13.

[36]

Wenz, 2005; Ball, 1998.

[37]

Sunstein, 2018, pp. 39–40; Boardman et al., 2014, p. 394. As Sunstein explains “We are not

speaking here of identifiable lives—of the lives of your son, your mother, your lover, or your best friend. …the government is assessing the value of reducing statistical risk of death.” [38]

To be clear, there are some philosophers that make such a mistake, and confuse average risk

reduction with assassination for hire (which is one reason that economists are so defensive of this concept). This error can be found in the works of Stephen John and Emma Curran (2021), or T. M. Scanlon (2000), who argue that cost-benefit analysis would justify killing or maiming a specific single person to keep the World Cup on television. These views and their failings are further addressed in Chapter 15. [39]

Sunstein, 2018 pp. 42–45; Boardman et al., 2014, pp. 394–397, 466–468; Thaler and Rosen,

1976. [40]

Consumer purchase and value of future earnings are also used to value statistical lives, but I do

not focus on them as they are not used by Viscusi and Masterman (2017) to arrive at their crosscountry estimates. [41]

See Boardman et al., 2014, pp. 394–397, 466–468 for a more complete explanation of these

techniques and how they are operationalized. [42]

Boardman et al., 2014, p. 395.

[43]

Robinson, Hammitt, and O’Keefe, 2018; Viscusi and Masterman, 2017.

[44]

Boardman et al., 2014, p. 422.

[45]

Chavas, 2017; Polasky and Segerson, 2009; Boyce, 2001; Adams 1996.

[46]

Boardman et al., 2014.

[47]

Boardman et al., 2014, p. 392.

[48]

Viscusi and Masterman, 2017, p. 227.

[49]

Bill and Melinda Gates Foundation, 2019.

[50]

Robinson, Hammitt, and O’Keefe, 2018.

[51]

Singer, 1972.

[52]

Beatty, et al., 2020; Millennium Challenge Corporation, 2021; Millennium Challenge

Corporation, 2020b; Osborne, S., 2019; Millennium Challenge Corporation, 2007. [53]

Cameron, 2010, p. 165.

[54]

Cameron, 2010, p. 168.

[55]

Robinson, Hammitt, and O’Keefe, 2018, p. 3.

[56]

Boardman et al., 2014, p. 396; Ackerman and Heinzerling, 2004, p. 77; Anderson, 1993, p. 197.

[57]

Boardman, 2014, p. 396.

[58]

Ackerman and Heinzerling, 2004, p. 79.

[59]

Ackerman and Heinzerling, 2004 p. 82; Anderson, 1993, p. 196.

[60]

Sagoff, 2004, pp. 180–181.

[61]

Sagoff, 2004, p. 8.

[62]

Sagoff, 2004, p. 181.

[63]

If you think that we should never approve such a policy, then you surely support lowering all

speed limits to five mph, and mandating complete lockdowns during flu season. [64]

Ackerman and Heinzerling, 2010, pp. 34–35.

[65]

An economist might object to this characterization by claiming that the lives lost in the other

country are an opportunity cost. However, under such a framework it seems that no policy would satisfy the Pareto Efficiency criterion, as there will always be an opportunity cost to spending money, even if that is only the small amount of interest it would collect sitting in a bank account. [66]

Anderson, 1993.

[67]

Anderson, 1993, p. 213.

[68]

Sagoff, 2004, p. 181.

[69]

Raz, 1986, p. 322.

[70]

Raz, 1986, p. 325.

[71]

Boardman et al., 2014, pp. 34–36.

[72]

Raz, 1986, p. 327.

[73]

Anderson, 1993, p. 211.

[74]

Sagoff, 2004.

[75]

EPA, 2018.

[76]

Viscusi, 2009, p. 110.

[77]

Viscusi, 2009.

[78]

Ackerman and Heinzerling, 2004, p. 73.

[79]

Ackerman and Heinzerling, 2004, p. 74.

[80]

EPA, 2018.

[81]

Note that economists do have a concept that is used in these situations: Disability Adjusted Life

Years (DALYs). We will not delve into the specifics of this here, but it will have many of the same issues and solutions as the value of a statistical life.

[82]

Sunstein, 2004.

[83]

Sunstein, 2004, p. 417.

[84]

Sunstein, 2004, p. 418.

[85]

Sunstein, 2004, p. 395.

[86]

It is important to note that, as we show in Chapter 16 (response 1.1), despite Sunstein’s

protestations, some donor agencies do use statistical lives this way. [87]

United Nations, 1945.

[88]

Hernandez, 2017; Moyo, 2009; Dreher, 2002.

[89]

E.g., Hartman, 2017.

[90]

United Nations, 2020; World Economic Forum, 2020; author’s calculations.

[91]

IRS, 2017, p. 11.

[92]

Moyo, 2009.

[93]

Portney, 2008, p. 39.

[94]

Portney, 2008, p. 39.

[95]

E.g., Portney, 2008, p. 39.

[96]

Daley, 2015.

[97]

Pinsker, 2015.

[98]

Portney, 2008, p. 39.

[99]

Portney, 2008, p. 39.

[100]

Boardman et al., 2014, p. 50. In fact, as we will see in Chapter 16, Boardman et al. admits that

willingness to pay can overstate utility for wealthy and understate it for the poor (Chapter 16 also shows why Boardman et al. go wrong in dismissing this disparity). [101]

Pinsker, 2015.

[102]

Sunstein, 2018, p. 41.

[103]

Thorbecke, 2016.

[104]

da Cunha, 2016.

[105]

Proulx and Crane, 2019; Thorbecke, 2016.

[106]

Thorbecke, 2016.

[107]

Viscusi and Masterman 2017.

[108]

Viscusi and Masterman, 2017; author’s calculations.

[109]

Sunstein, 2004.

[110]

Aristotle, 1989, Nicomachean Ethics, 1106b.

[111]

Rawls, 1971.

[112]

Wenz, 2005; Ball, 1998.

[113]

Working class’s benefits 900 x $400 = $360,000 minus millionaires’ costs 100 x $9,500 =

$950,000. [114]

Working class’s benefits 900 x 4% = 3600%, minus millionaires’ costs 100 x 0.95% = 95%.

[115]

For simplicity of this example, imagine that the costs of moving and rebuilding are identical.

The next example looks at how to value government costs using percentage willingness to pay. [116]

Benefit of $17,000 per millionaire for 100 millionaires, the risk they were feeling before would

be reduced. [117]

Benefit of $1,700 per working class person, for 900 people, they did not have the risk before, but

now would. [118]

1.7% benefit for 100 millionaires.

[119]

17% benefit for 900 working class people.

[120]

Jaworska and Tannenbaum, 2021; Kuflik, 1998.

[121]

Quite a few philosophers doubt that corporations have any moral status, or moral

consideration/rights: Cieply, 2017; List and Pettit, 2011; Ozar 1985; Manning, 1984; Hessen, 1978; Jensen and Meckling, 1976, etc. Silver, 2019, is the only, or one of the few, that has attempted to defend the claim that corporations have rights. I do not address Silver in this work, as his claims are largely tangential to the present case as noted above and he contends that corporations are conscious beings that experience pain and pleasure, are able to care for others, are capable of making decisions independent from their managers, and have goals independent of the people running them: a view sufficiently ludicrous such that merely stating it should suffice as a disproof of it, nor do I suspect that any economist defending absolute willingness to pay holds it. [122]

Beck, 2020; Bratton and Kelling, 2015; Harcourt, 2001.

[123]

Johnson, 2019; Morrow and Shjarback, 2019; Smith et al., 2017.

[124]

Imran, Hosen, and Chowdhury, 2018; Zhao, Feng, and Castillo-Chavez, 2014; Heller, Jacob, and

Ludwig, 2011. [125]

Trochmann and Gover, 2016; Sharpe and Johnson, 2009. Alpert, MacDonald, and Dunham,

2005. [126]

Masera, 2021; Delehanty, Mewhirter, and Welch, 2017; Edwards and Harris, 2015.

[127]

Tatem, 2021.

[128]

Headley, 2021; Prenzler, Porter, and Alpert, 2013.

[129]

Beck, 2020; Bratton and Kelling, 2015; Hinkle and Weisburd, 2008; Worrall, 2002; Katz, Webb,

and Schafer, 2001. [130]

Passavant, 2015; Wacquant, 2009; Wacquant, 2008.

[131]

Legewie and Fagan, 2019.

[132]

Hinkle and Wiesbund, 2008; Rosenfeld, Fornango, and Rengifo, 2007; Messner et al., 2007;

Katz, Webb, and Schaefer, 2001; Novak et al., 1999; Sherman, 1990. [133]

Harcourt and Ludwig, 2006; Eck and Maguire, 2000.

[134]

Passavant, 2015; Wacquant, 2009; Wacquant, 2008.

[135]

McIntosh et al., 2020; Darity et. al., 2018; Charles and Hurst, 2002; Barksy et al., 2002; Oliver

and Shapiro, 1995; Blau and Graham, 1990.

[136]

McIntosh et al., 2020; Darity et. al., 2018.

[137]

Darity et al., 2018.

[138]

Choi et al., 2019; Shapiro, Meschede, and Osoro, 2013; Charles and Hurst, 2002.

[139]

Wainer and Zabel 2020; Di, Belsky, and Liu 2007.

[140]

Choi, 2020; Choi et al., 2019.

[141]

Rodriguez, 2020.

[142]

Cole, 2020.

[143]

Acemoglu et al., 2020; Robinson, Sullivan, and Shogren, 2021; Raitzer et al., 2020; Scioto

Analysis, 2020; Thunstrom et al., 2020. [144]

John and Curran, 2021, pp. 2–3.

[145]

BBC, 2022; Maturana et al., 2021; Leeka, Schwartz, and Kloner, 2010; Baumhäkel et al., 2007;

Strike and Steptoe, 2005. [146]

This will focus only on the individual benefits of vaccines to show that even ignoring

externalities, and questions of public goods, we can make the case that absolute willingness to pay fails to maximize real consumer benefit. [147]

Guardian, 2022.

[148]

Al Jazeera, 2021; Steinberger, 2022.

[149]

Loofbourow, 2021.

[150]

Carmichael and Goodman, 2020.

[151]

Carmichael and Goodman, 2020.

[152]

Pietsch, 2021.

[153]

Millennium Challenge Corporation, 2022; Rose and Wiebe, 2015.

[154]

Fletcher and Hayes-Birchler, 2020; Tilley and Jenkins, 2020.

[155]

Millennium Challenge Corporation, 2020a.

[156]

Millennium Challenge Corporation, 2020a.

[157]

Millennium Challenge Corporation, 2021.

[158]

Sunstein, 2018, p. 41.

[159]

Boardman et al., 2014, p. 38.

[160]

Boardman et al., 2014, p. 38.

[161]

A clearly racist method to be sure, but given that some like John and Curran (2021) have

advocated for politicians to be able to ignore the costs and benefits that they don’t find valid, and there are still racists politicians in the world, it is surely a possible method. [162]

Mitnik, Bryant, and Grusky, 2018.

[163]

Narayan et al., 2018.

[164]

Davidai and Gilovich, 2018; Davidai, 2018; Kraus and Tan, 2015.

[165]

Sahm, 2019; Toossi, 2002.

[166]

John and Curran, 2021; Sagoff, 2004; Anderson, 1993; etc.

[167]

In 2007, the World Bank had 83 research staff, 80 of which had PhDs in economics. The other

three had degrees in government and public policy, anthropology and development, and sociology. Rao and Woolcock, 2007. [168]

Whirled Bank, 2001.

[169]

Empiricus, 1990, p. 87.