The Costly Betrayal of Cost-Benefit’s Promise

The Costly Betrayal of Cost-Benefit’s Promise

September 26, 2023

Jerry Cayford

Originally published here

A conversation, eighty-some years ago, shaped our world today. The conversation was among three papers in economics. In 1939, Nicholas Kaldor and John Hicks published papers that steered economics through a major crisis, one that threatened its legitimacy as an advisor on public policy. With the Great Depression still raging and capitalism in tatters, they engineered an ingenious compromise that not only salvaged the scientific standing of their discipline, but also guided it toward a vision of a new, fairer world. We will come later to the third paper, which betrayed that vision and opened up fault lines now fracturing our culture.

Few outside the economics profession have heard of the “Kaldor-Hicks potential compensation criterion,” but it forms the heart of modern cost-benefit analysis, which is used whenever politics appeals to economics to judge if an action will benefit society. Many hundreds of billions of dollars have been allocated based on cost-benefit analysis. In the United States, Executive orders from Reagan to Biden mandate that policies with impacts of over $100 million justify themselves by cost-benefit analyses. At that scale, interpretations of the criterion have vast consequences for the distribution of wealth in our society.

The Kaldor-Hicks criterion says, roughly, if the amount of money generated by an action is sufficient to fully compensate those who are harmed by the action, plus some profit, then do it. This is a way of describing what it means for benefits to outweigh costs. There is a catch, though, a compromise.

The story of this compromise, what it is, how it came about, and what has happened since, will have a familiar structure. Legends and morality tales often describe a bargain whereby the protagonist gets a position of privilege but with a string attached, some prohibition: stay out of that locked room; don’t eat from that forbidden tree; be back before dark. Violating the prohibition sends the protagonist on a journey. The odyssey of Kaldor-Hicks from its origins to today has something to teach our own troubled times.

Value-free judgment

Occupy Wall Street challenged the legitimacy of how wealth is distributed in our society. It claimed that “99 percent” of us are short-changed by our major institutions, which are serving the privileged one percent. On the other side of the debate, a more hard-nosed, laissez-faire approach argues that people acting freely within our society’s rules sometimes win and sometimes lose; we complain about it at the peril of processes that have served us well.

A dispute about who “built it” exemplifies this disagreement. From the 2012 campaign of businessman Mitt Romney to the administration of businessman Donald Trump, one side in this dispute claims business entrepreneurs build their own wealth and are entitled to enjoy it free from redistribution of that wealth by the government. The opposing idea is captured by the Depression-era laborer’s lament:

Once I built a railroad, made it run,

Made it race against time.

Once I built a railroad, now it’s done,

Brother, can you spare a dime?1

These and other disputes polarizing our country can seem just different opinions or ideologies or values talking past each other. They can seem to turn on vague and subjective ideas like “fairness,” ideas that provide no common ground on which to hash out differences. But they don’t have to. Kaldor-Hicks is the story of figuring out how to assess different paths forward, without knowing the ground. That search for objectivity—during economists’ darkest time, one of the world’s darkest times—can help us find solid footing through the ideological swamps today.

A tension deep within the history of economics informs this story, a tension between philosophy and politics. Like many disciplines, economics grew out of philosophy: the philosopher Thomas Hobbes created the framework for modern social science; the philosopher Adam Smith started thinking systematically about macroeconomics; Jeremy Bentham and John Stuart Mill, philosophers of utilitarianism, created the ethical theory that remains fundamental to much economic analysis. Politics, however, needs answers to economic questions, and it needs them now. The tension in economics is between its inheritance of philosophy’s austere commitment to reason, and its concrete involvement with politics’ more free-wheeling demand for sensible value judgments.

To provide politics with value judgements while staying within philosophy’s constraints of reason, economics has always employed utilitarianism, the ethical theory that the greatest happiness for the greatest number should be our goal. Utilitarianism offers a method to calculate scientifically the best course of action: add up the impacts, positive or negative, on the happiness (or pleasure or utility) of all the people affected by each option, and select the option that produces the greatest total. This method avoids saying some people’s values matter more than other people’s values. Whatever makes anyone happy goes into the calculation. No judgment. Economics would be the morally neutral science of calculating the effects of policies on overall social welfare.

There is a stumbling-block, and it’s a beauty: there is no value-neutral way to compare one person’s happiness with another’s. This difficulty is completely intuitive. Anyone who has ever wasted time wondering who enjoys sex more, men or women, has encountered it. Each person’s pleasure is entirely subjective, and no metric of subjective experience can be carried from one person to the next to compare them.

If we cannot even measure, say, whether Albert enjoys his cup of coffee as much as Janet enjoys her piece of pie, the whole project of aggregating policy impacts on individual well-being is a non-starter. This difficulty is known as “the problem of interpersonal utility comparisons.”

Utilitarians recognized the problem from the beginning, but largely left it unaddressed. When the Great Depression forced a reexamination of economics, the nagging theoretical problem of interpersonal utility comparisons became an acute practical challenge to the young discipline’s policy prescriptions. Failing to meet the challenge threatened to marginalize economics entirely. The Kaldor-Hicks criterion was and remains economists’ answer to that challenge.

Damage control

Nicholas Kaldor arrived in London by an immigrant’s path. Son of a prosperous, middle-class, Jewish lawyer, Kaldor left post-WWI Hungary, where rising authoritarianism and antisemitism were the defeated country’s response to shrinking territory and economic collapse. Hungary passed the first of many European “Jewish laws” in 1920, limiting Jewish enrollment in universities. The teenaged Nicholas Kaldor did get one of those limited spots at the University of Budapest, but soon left the country to study in Berlin, and then England.

Less interested in economics than journalism, and slightly averse to mathematics, Kaldor registered at the London School of Economics as a non-degree student. LSE was a dynamic, young university, barely three decades old, and avowedly welcoming to “students from overseas.”2 There, Kaldor found his calling, taking his degree in 1930 and becoming Lecturer, then Reader in economics. He would go on to become one of the most celebrated economists of the 20th century, a Director at the U.N. Economic Commission for Europe, and a Professor at Cambridge University.3

Kaldor confronted the comparison problem in “Welfare Propositions of Economics and Interpersonal Comparisons of Utility.”4 He opens his paper by acknowledging that the problem is insoluble, quoting other economists’ expressions of alarm at this fact: “If the incomparability of utility to different individuals is strictly pressed, not only are the prescriptions of the welfare school ruled out, but all prescriptions whatever. The economist as an adviser is completely stultified.”5 Kaldor notes that it is “assumed, on both sides, that the scientific justification of such comparisons determines whether ‘economics as a science can say anything by way of prescription’.”6 This is the Slough of Despond out of which he aims to pull his colleagues.

Kaldor did not aspire to solve the insoluble, however, but to avoid it. His short, elegant paper argues that a great deal can be done without facing the comparison problem at all. He builds on an earlier argument by Vilfredo Pareto that people’s rankings of their own subjective preferences can sometimes be sufficient to draw conclusions about social welfare.

Consider a simple example. If Janet would rather have a dollar than her piece of pie, and Albert would rather have the pie than a dollar, both of them will be better off if Janet sells her pie to Albert for a dollar. We can infer that the pie sale raises the utility of the whole system because both consider themselves better off. We still can’t say that Albert gets more utility from the pie than Janet does—we don’t know that—only that the deal, as a whole, benefits them both.

A change that makes some people better off without making anyone worse off (a “Pareto improvement”) raises overall welfare, no interpersonal utility comparisons necessary. Left to chance, though, it would seem impossible that a change affecting many people would hurt no one. Kaldor’s striking insight was that compensating those hurt by a change could turn a wide range of changes into such Pareto improvements.

Through compensation, many actions can be shown to raise welfare. If all the people who gain from, say, building a freeway benefit so much that they could pay off all the people hurt by that action and still benefit, and all the people hurt would willingly accept the payment in exchange for allowing the action, then we can conclude that everyone is better off when we build the freeway and transfer compensation from its beneficiaries to those whose interests have been harmed.

It’s a clever and well-crafted argument. Without making any interpersonal comparisons of utility or privileging anyone’s values, we now have a procedure that can infer from individual preferences alone whether an action will increase the collective social welfare of the group. Problem solved, right?

The catch

Kaldor rescued economics from policy irrelevance by showing how economists and politicians can collaborate to make the world better: “This argument lends justification to the procedure, adopted by Professor Pigou in The Economics of Welfare, of dividing ‘welfare economics’ into two parts: the first relating to production, and the second to distribution” (551). Economists can tell politicians which policies create enough surplus product to more than offset their costs, and politicians can secure a real improvement by distributing that surplus so that everyone benefits. The two parts work together to raise welfare.

Easier said than done. The required partnership with politics has proved a challenge for economics. John Hicks was well aware of this challenge when he took up and fleshed out Kaldor’s core idea in “The Foundations of Welfare Economics.”7 Hicks had taught at the London School of Economics during Kaldor’s early years there. In 1939, he was at the University of Manchester. Later, he became a professor at Oxford University and a winner of the Nobel Memorial Prize in Economic Sciences.

Hicks foreshadows the troubles that will be encountered by the argument he and Kaldor are developing:

The main practical advantage of our line of approach is that it fixes attention upon the question of compensation. Every simple economic reform inflicts a loss upon some people; the reforms we have studied are marked out by the characteristic that they will allow of compensation to balance that loss, and they will still show a net advantage. Yet when such reforms have been carried through in historical fact, the advance has usually been made amid the clash of opposing interests, so that compensation has not been given, and economic progress has accumulated a roll of victims, sufficient to give all sound policy a bad name. (711)

Compensating those hurt by public policies is both practically and politically difficult. Though key to Kaldor’s whole argument, the requirement has encountered great resistance. This is where reason meets pushback from politics.

On this crucial issue of compensation, the later development of cost-benefit analysis abandons Kaldor and Hicks. Economists start claiming that the mere potential to compensate is sufficient to infer increased welfare. That is, if the benefits are enough so that we could fully compensate the victims (even if we do not actually compensate them), cost-benefit analysis concludes that the benefits are greater than the costs. Bluntly, more total wealth is better, even if creating it “accumulates a roll of victims.” This faulty reasoning is a wrong turn of enormous consequence, as we will investigate below. Before we do, though, let’s be clear about Kaldor’s and Hick’s argument on this point: do we actually need to compensate losses? If we know the freeway creates more than enough new wealth to pay off those it hurts, why can’t we conclude that on balance it does more good than harm?

Our simple example shows why Kaldor and Hicks are right that actual compensation is essential. Contrast the pie sale (actual compensation) with a potential compensation scenario: what if we just give Janet’s pie to Albert without payment? The benefit to Albert is more than a dollar (he prefers the pie to a dollar), and the cost to Janet is less than a dollar (she prefers a dollar to the pie). Doesn’t it follow—as cost-benefit analysis would conclude—that the benefit to Albert is more than the cost to Janet? No; there is a logical error here. The fallacy in this reasoning is that it assumes the subjective value of a dollar is the same to Janet and Albert. And that is exactly what cannot be known, not without making an interpersonal utility comparison. Unless Janet gets her dollar, we have no idea if transferring the pie makes things better or worse.

Is it really plausible, though, for money and utility to diverge? Certainly. Imagine the pie is all Janet has eaten today, but a dollar would give her bus fare she badly needs. Albert is well fed and the pie means little, but a dollar means even less. In this situation, sale of the pie for a dollar does make them both better off, but uncompensated transfer of the pie would make Janet miserable for piddling benefit to Albert. The fact that the transfer would generate enough value for Albert so he could potentially compensate Janet and still benefit implies absolutely nothing about their collective welfare.

Potential compensation, then, is an unsound criterion of welfare. Multiply this simple example to the scale of the national economy, and we can see how economic policies guided by that criterion can damage the interests of millions of people for piddling benefit to others. They can create huge net declines in social welfare.

Potential compensation is properly a criterion not of welfare but of what Kaldor calls “the economist’s case for the policy” (550): it shows that “[t]he first, and far the more important part,…which relate to the increase in aggregate production” (551) is adequate to cover costs. In the larger argument of Kaldor and Hicks, the politician’s part is distributing that increased production to compensate for the harm it has done (i.e. paying those costs). The compromise intrinsic to this argument is that only both parts together can be evaluated. Production and distribution cannot be separated without forfeiting the goal: an objective verdict on welfare. Unless the fruits of a good policy are used to make whole those it has harmed, economists have no argument that the policy is in fact good.

Since compensation to victims is essential for a sound comparison of costs to benefits, it must be integral to economists’ policy analyses and recommendations, and Kaldor and Hicks accordingly integrate it:

Since almost every conceivable kind of compensation (re-arrangement of taxation, for example) must itself be expected to have some influence on production, the task of the welfare economist is not completed until he has envisaged the total effects of both sides….If, as will often happen, the best methods of compensation feasible involve some loss in productive efficiency, this loss will have to be taken into account. (Hicks 712)

Actual compensation, and all its attendant costs, is a necessary part of a scientific analysis of any policy’s costs and benefits.

The objectivity economics seeks carries a powerful status: everyone is supposed to agree to it. To earn that status and to justify its advice to politics, economics must maintain its purity, its scientific detachment, following where reason leads. The crisis that Kaldor and Hicks addressed was that economic advice on social welfare seemed to require interpersonal utility comparisons, value judgments that reduce economics from science to political advocacy. Kaldor and Hicks devised a sound procedure to avoid such comparisons and give objective advice to policymakers.

Their achievement came at a cost, though, one that tests the commitment of economists to reason. It includes an unpalatable prohibition: do not try to evaluate policies isolated from compensation; do not separate production from distribution; “accustom ourselves, whenever we can, to thinking of every economic reform in close conjunction with some measure of compensation, designed to render it approximately innocuous from the distributive point of view” (Hicks 712). Economists immediately began to chafe under this constraint. If “the economist’s case” is “far the more important part,” why should it be held hostage to the awkward, difficult (and, presumably, far less important)8 political task of redistribution? Like the character of legend prohibited from the locked room or the forbidden tree, these apostles of reason found the constraint unreasonable.

Going rogue

The idea initiated by Kaldor and elaborated by Hicks did not assume its final form until further refined two years later by Tibor De Scitovszky in “A Note on Welfare Propositions in Economics,” the third founding paper of cost-benefit analysis.9 Like Kaldor, Scitovszky was born and educated in Hungary, then went to England between the wars and studied at the London School of Economics. Scitovszky, though, was from the Hungarian nobility; his father was Hungary’s Minister of Foreign Affairs in the mid-1920s. Scitovsky—he dropped the aristocratic “De” and simplified his name’s spelling—continued his emigration from England to the United States and spent most of his career teaching at Stanford and U. C. Berkeley.

The standard summary of Scitovsky’s role in this history focuses on a paradox he points out in the argument of Kaldor and Hicks and resolves in his 1941 paper, a technical point on which he is correct. It is what Scitovsky gets wrong, however, that sets cost-benefit analysis wandering.

Scitovsky’s goal was less to refine Kaldor and Hicks than to refute them: “The present note is a criticism of the principle enunciated in Mr. Kaldor’s first-quoted article and underlying the argument of the others” (77n1). What Scitovsky disputes, explicitly, is that economics either can or should be “scientific” and value neutral. The corollary that implicitly accompanies this stance—and over time comes to obscure and then damage the community’s interpretation of the whole argument—is that the attempt at rigor here is misplaced, that it amounts to a bogus “claim to purity.” According to Scitovsky, confining the economist’s judgments to the modest limits Kaldor observes “would hardly be satisfactory” (79). He adopts an acerbic tone toward efforts to avoid value judgments, suggesting the economist “may also renounce his claim to purity and base his own recommendations on both criteria [economic efficiency and social justice]” (80n1). He routinely refers to those recommendations as “value judgments.”

Scitovsky makes only one actual argument, though, against the scientific character of Kaldor’s and Hicks’s procedure, and that argument is unsound. He claims that making whole those hurt by a policy is as much a value judgment as is letting them take their lumps:

Favouring an improvement in the organisation of production and exchange only when it is accompanied by a corrective redistribution of income…such propositions are not independent of value judgments between alternative income distributions either. For, going out of their way to preserve the existing distribution of income, they imply a preference for the status quo. (79; emphasis in original)

This argument does not work; it is like saying we only think sea level is rising if we have an arbitrary attachment to the present level. Kaldor and Hicks require no such value judgment; their use of the status quo is not moral—not a preference for it—but rather epistemological: the status quo—whether of sea level or income distribution—is the benchmark that defines “change,” and identifying change is essential.

Kaldor and Hicks do not argue for preserving the status quo, because that is not the economist’s business: “Whether [those who are harmed]…should in fact be given compensation or not, is a political question on which the economist, qua economist, could hardly pronounce an opinion” (Kaldor 550). Their point is that without compensation it is conceptually impossible to know whether the change in welfare is positive or negative, whether the policy makes the world a better or worse place. This fundamental point disappears from Scitovsky’s reinterpretation, and with it the whole ambition to “scientific justification” of economic prescriptions.

Notice that Scitovsky is well aware that Kaldor and Hicks consider corrective redistribution necessary, though he disagrees with them. The misinterpretation of their argument as calling for merely potential compensation comes from others later, not from Scitovsky.

Scitovsky defies the prohibition that comes with Kaldor’s and Hicks’s argument, and decisively separates production from distribution: “the economist may put forward his welfare propositions with due emphasis on their limitations, as being based on the sole criterion of efficiency” (79-80). Distribution is an afterthought: “the economist…may then point out the nature of eventual redistributions of income likely to accompany a given change, and stress the necessity of basing economic policy on considerations both of economic efficiency and of social justice” (80). This advice sounds like exactly what it is: a subjective moral opinion that social justice is a worthy policy goal. But the key fact here is that the same is equally true of productive efficiency: once divorced from distribution, prescriptions “based on the sole criterion of efficiency” are just as much subjective moral opinions as are pleas for social justice.

The limitation of efficiency-only recommendations is that we know nothing at all about their welfare effects. This fact is crucial. Scitovsky, though, hides this crucial fact. He performs an intellectual sleight of hand by redefining terms. Consider these two sentences:

In welfare problems, of course, we can aim neither at a ‘true’ answer nor at its quantitative expression without measuring satisfaction and comparing different people’s. (87) (which he grants we cannot do)

[T]he general welfare can be conceived of as average welfare (87n1).

On their face, these are contradictory: if welfare cannot be expressed quantitatively, then plainly it cannot be averaged or totaled into “general welfare.”

The only way the two sentences could not be contradictory is if the meaning of “welfare” changes between them. That’s exactly what Scitovsky does. He changes the subject by redefining the key concept. He is no longer talking about welfare at all but about wealth. Welfare is now dismissed as ‘true’ welfare—in scare-quotes—and replaced by a “general welfare” that is simply a euphemism for productive efficiency as measured in money. We are now to maximize average wealth, regardless of its effects on true welfare, and label the result “general welfare.”

The replacement of social welfare by per capita wealth changes everything. Where Kaldor and Hicks “fix attention upon the question of compensation” (their approach’s “main practical advantage”) (Hicks 711), Scitovsky eliminates the question of compensation entirely by substituting an average measure, which is indifferent to who has the money. By design or by accident, the interests of Scitovsky’s class have pushed back against the interests of Kaldor’s. The one percent, whether Old World nobility then or Wall Street barons now, are in Scitovsky’s debt.

So, out with distribution; production is now the whole game. In this way, Scitovsky unburdens economists of the compromise Kaldor and Hicks devised. Politicians, too, are freed from the nasty duty of redistributing wealth. The only losers are the citizens, because the goal of pursuing policies that truly enhance welfare has been lost in the shuffle. Also lost is the objectivity that follows where reason leads.

The devilish details

Scitovsky’s essay—with all its faults—set welfare economics on its subsequent journey. Enchanted with the power of simple calculation, economists forget the warning given them. They adopt Scitovsky’s convenient reliance on “the sole criterion of efficiency.” Generally more loath than he was to embrace “value judgments,” though, they also want the rigor and objectivity of Kaldor and Hicks, and claim they have it. Over time, economists forget that the argument of Kaldor and Hicks requires “corrective redistribution of income,” and even attribute to them Scitovsky’s rejection of such compensation. Here is a typical modern reading:

The Kaldor-Hicks criterion argued that a public policy was justified if it produced gains in excess of losses so that it was possible for winners from that policy to compensate losers—even if such compensation did not actually occur. This approach soon became the foundation of a new applied welfare economics known as cost-benefit analysis.10

Kaldor and Hicks argued no such thing—quite the opposite—but today their argument is widely misinterpreted to justify policies as welfare-enhancing without compensating people harmed by those policies.

Still, logic is not so malleable, and economists know there is a problem. Naming the potential compensation criterion after Kaldor and Hicks cannot change the fact that the potential to compensate does not imply improved welfare. The literature is full of articles that recognize something is wrong with the criterion.11 E. J. Mishan, the fourth player in this tale, embodies the tension beautifully.

Mishan was, like Kaldor, a Jewish emigrant from a WWI-torn country. Or rather, his parents were, and Mishan was born shortly after they left Egypt during the war and came to England. Another long-time lecturer at the London School of Economics, he became an early expert in the new field of cost-benefit analysis, and the author of the widely-used textbook Cost-Benefit Analysis (now in its sixth edition (first posthumous) and cited repeatedly by the U. S. Office of Management and Budget (see next section)). In our tale, Mishan plays the Greek chorus, observing the drama, articulating the danger. Late in his life, seventy years after our story began, Mishan reflected pessimistically on his chosen field in “Reflections on the Foundations of Economic Valuation”: “for the economist interested in Cost-Benefit Analysis, a realization of the weakness of the foundations of economic valuation—a weakness, that is, of its moral appeal—cannot but act to dampen his spirits.”12

Nevertheless, widespread awareness that the potential compensation criterion is unsound does not fully dampen anyone’s spirits, even Mishan’s: “Yet no matter how disenchanted [the economist] may become, he should resist any temptation to abandon the techniques of Cost-Benefit Analysis in the evaluation of public projects” (541). With Mishan and others, awareness of the problem seems to coexist with misunderstandings that mask the real gravity of the problem. Let’s consider some of those masks.

Perhaps the most important misunderstanding that pervades welfare economics is that cost-benefit analysis, despite its unsoundness, still somehow approximates welfare, as if the unsoundness threatened only peripheral issues, such as distributive justice, but left the core task of assessing social welfare still roughly on track. This is incorrect. Cost-benefit analysis does not approximate the measuring of welfare any more than casting chicken bones approximates predicting the future.

The mistake lies in assuming (or stipulating) that what we can measure in some convenient way—some way that requires neither interpersonal utility comparisons nor compensation through redistribution—is, in fact, welfare. Scitovsky’s stipulation that productive efficiency be defined as “general welfare” is an example of such an assumption. Economists’ old friends the philosophers have named the logical fallacy he commits:

Almost every book on a technical subject employs stipulative definitions….One way of misemploying stipulative definitions in an argument is so common and fallacious, however, that it deserves special consideration. The technique consists in making some controversial statement true, indeed, analytic, by stipulating a definition for some key term and then claiming to have shown the original statement to be true. When this happens, a stipulative definition is masquerading as a reportive one. We shall refer to this dubious procedure as the redefinist fallacy.13

The redefinist fallacy occurs when, instead of working through a question to find a real answer, we simply define one possibility to be the answer. Defining social welfare to be enhanced by productive efficiency (i.e. by the potential to compensate)—as Scitovsky does—is no more valid than, say, stipulating that the country’s welfare be measured in territory, so that wars of conquest are by definition welfare-enhancing.

The redefinist fallacy is sort of a formal way to explain what the drunk has done wrong in searching for his lost keys here under the streetlamp where the light is better, even though he lost the keys over there. The light may be better under productive efficiency, but that does not mean it illuminates welfare. Almost every use in the welfare economics literature of phrases like “general interest,” “overall welfare of the community,” or “benefit for the country as a whole,” relies on the redefinist fallacy. These phrases are meant to describe social welfare, the phenomenon we are trying to understand. Kaldor and Hicks showed a method to discover it in limited circumstances (that include actual compensation), but otherwise it remains elusive, and stipulations don’t help.

Another group of misunderstandings defends potential compensation as implying improved welfare under certain conditions. All these defenses fail because the required conditions are entirely implausible. The simplest is the blunt assumption that distribution does not matter, the value (utility) of the next widget is the same no matter where it goes. This assumption violates our most basic intuition about utility: when you need something, it is valuable; when you have enough, another one is not valuable. Economists call this fundamental idea “diminishing marginal utility.”

Santa’s elves spend the whole year making toys, and on Christmas Eve Santa spreads joy to children around the world. If marginal utility were constant, not diminishing, he could just give all the toys to one favorite child and the amount of joy would be the same. A child can tell you this is nonsense (maybe not that favored child).

The assumption of constant marginal utility is worse than just false, though; it practically annihilates the very concept of value. Distributing stuff to where it will do good is how value is created. Constant marginal utility means the truckload of fabric has the same value delivered to the furniture upholsterer’s or the skating rink. Thread has the same value whether woven into fabric or left on the spinning floor. Following this logic back to the cotton seed in the ground and even further, we see the whole chain of value neutralized by the assumption that marginal utility is constant. This nihilistically strong assumption almost accomplishes the work of entropy itself: we can glimpse a rough proof that constant marginal utility implies a thin, gray soup of evenly distributed atoms has the same value as our own vibrant universe. But only philosophers ponder such silly extremes. Let’s leave it at this: the assumption is false, and everyone knows it is.

Another line of defense claims that the logical inadequacy of potential compensation is not really a practical problem, because things work out over time. One such argument is that the benefits of public policy fall randomly, and therefore cancel out, without the need for cumbersome, costly, case-by-case compensation. But do the effects of policy fall randomly?

Not according to Who Stole the American Dream? by Hedrick Smith.14 Smith examines the very unevenly distributed effects of American public policy over the last four decades. He shows that the benefits fall mainly to the rich, the costs to the middle class. Lobbying illustrates this non-randomness. Spending on lobbying averages around three and a half billion dollars per year, almost all of it by corporations. The return on investment in lobbying, though complex, is universally agreed to be extremely high; one academic study concludes that “results indicate the market value contribution of an additional dollar of lobbying is roughly $200.”15 By this analysis, over half a trillion dollars per year of benefits from public policy skew to corporations.

The real distortion is even worse because the baseline is already biased toward the rich. To see the bias, think of wealth accumulation as an industry with economies of scale and barriers to entry (meaning you have to have money to make money). In that case, policy guided by cost-benefit analysis that maximizes wealth will favor those who already have it, widening the disparities between rich and poor even before lobbying skews the results further. The effects of public policy do not fall randomly.

A final defense of potential compensation is the one that, according to Mishan, economists actually believe:

The widespread acceptance of the KaldorHicks criterion, or the potential Pareto improvement criterion,…derives from a belief that its continued application (rather than that of any other criterion) would tend to produce an actual Pareto improvementespecially within an economy having a progressive system of taxation. (530; see also 541)

Notice several things in this revealing passage. First, Mishan implicitly acknowledges that only an actual Pareto improvement, not a potential one, increases welfare; this point already repudiates the central claims of cost-benefit analysis. Second, this defense of potential compensation is similar to Kaldor’s and Hicks’s original argument: potential improvement can produce actual improvement through appropriate taxation: “by compensating the [losers] for any loss of income and by providing the funds for such compensation by an extra tax on those whose incomes have been augmented” (Kaldor 550), we can render “every economic reform…approximately innocuous from the distributive point of view” (Hicks 712). This is a sound argument, as discussed above.

Where this defense differs from Kaldor and Hicks, though, is that their argument requires actual compensation, whereas the belief that Mishan says enjoys “widespread acceptance” is that “continued application” of merely potential compensation “would tend to produce” the same result. In other words, continually maximizing wealth while ignoring its distribution will tend to lift all boats, which is the essence of “trickle-down economics.” (Perhaps economists’ widespread faith in “continued application” accounts, in part, for the trickle-down theory’s apparent survival, despite repeated refutation.)

The idea that wealth maximization alone promotes welfare is not backed by either argument or evidence, and is equivalent to the widely rejected trickle-down theory. Why, then, is it believed? Because tacitly it presumes that actual compensation will occur. What really “tends to produce an actual Pareto improvement” in Mishan’s summary is the “progressive system of taxation,” not the potential compensation criterion. The progressive taxation that Mishan presents as merely added insurance implicitly carries the weight of the argument by delivering the actual compensation the argument requires. But that crutch itself is threatened by the faulty criterion. Let’s look at how.

In morality tales of broken faith, it is a magical or higher being who imposes the prohibition, a being who represents something vital to human life. In breaking faith with this vital power, the protagonist is driven into the wilderness, dependent solely on his or her own limited resources. And so it is in this case.

Remember the key prohibition: do not separate production from distribution. It is the rule that cannot be broken if cost-benefit analysis is to show us the way to increased social welfare. Economists try to finesse, ignore, or define away this prohibition, and they simply violate it. In endorsing the unsound argument that potential to compensate is a criterion of increased welfare—meaning the distribution of policy effects does not matter—economists break faith with reason; their welfare judgments then depend on their own limited political intuitions instead of on reason. Endorsing a bad argument causes a cascade of consequences that haunt the history of welfare economics.

Once we buy the fallacy that productive efficiency alone is welfare-enhancing, distributive measures become superfluous. But that’s just the beginning. Once we enshrine the potential to compensate as our test of welfare, distributive measures actually appear welfare-diminishing. Consider a simple transfer of money. It does not increase wealth; rather, due to transaction costs, it slightly decreases total wealth. The provider gives up X dollars, but the recipient gains only X-minus-transaction-costs dollars. So, the recipient could not fully compensate the provider and still show a profit. The transfer fails our test, as every simple transfer must.16 In this way, the potential compensation criterion erodes the warrant for redistribution of any kind, including a system of progressive taxation. And this erosion of respect for redistribution and taxation is exactly what has happened. Far from tending to actual improvement, then, “continued application” of this faulty criterion tends to undermine improvement by undermining the corrective redistribution that Mishan implicitly assumes and that Kaldor and Hicks so compellingly argued was necessary.

The consequences cascade on. Since compensation is essential to creating objectively justified policies, redistribution will not go quietly. To explain its mysteriously stubborn appeal, then, we imagine that the need for redistribution isn’t an objective, logical requirement at all; instead, it must be moral and subjective. In this way, the false appearance of a great clash of values over wealth distribution is artificially created.

In truth, there is nothing ideological going on here, just facts about the logic of arguments. Distributive measures are required to assure social welfare improvements; productive efficiency is not enough. These are just facts. No opinion about distributive justice, fairness, the wealth gap, or any other controversial subject is required.

The cascade continues. Unleashing a cost-benefit criterion that flunks all compensation measures means the original goal of producing policies known to be welfare-enhancing can now only end in failure. Economists know this. Awareness of the logic is palpable in the literature. If there is only a hint of defensiveness in Scitovsky’s “the economist may put forward his welfare propositions with due emphasis on their limitations” (7980), the same concern has grown to a plaintive plea for leniency in Mishan’s reflections seven decades later:

For no intellectually honest economist would knowingly mislead the public, or the political decision-makers, about what a properly conducted economic assessment can deliver. Provided the economist…makes it abundantly clear to the political decision-maker…that the economic criterion…ensure no more than a potential potential Pareto improvement within the community. Moreover, provided that there still remains a penumbra of uncertainty…then the economist may indeed proceed to use the techniques of economic assessment with a clear conscience. For after clearly explaining to the political decision-makers the nature of an economic assessment, the economist cannot be held responsible for a political decision to adopt or reject the proposed project. (5412; emphasis and double-potential remove from real improvement are in the original)

This sorry scramble to avoid blame is what remains of the scientific procedure to make a better world that Kaldor and Hicks created. Our protagonist is deep in the wilderness, producing economic assessments that cannot claim to assess welfare.

The argument that the potential compensation criterion implies real welfare is wrong. Examine the details, and it only looks worse. Cost-benefit analysis built on that criterion is intrinsically unable to render any valid judgment on net social welfare. This fact is widely known yet even more widely ignored, which is puzzling in a discipline deeply committed to following reason. However it came about, though, this bad argument is now entrenched. What consequences, then, for the rest of us?

A reasoned determination

If economists have been “clearly explaining” to politicians “the nature of economic assessments” (Mishan) with “due emphasis on their limitations” (Scitovsky), the message has not gotten through. The experts may be ducking for cover, but the hapless client does not even know the project is going bust. This ignorance among politicians is demonstrated by the institutionalization of the flawed Kaldor-Hicks potential compensation criterion into federal regulatory practices (which is only part of the impact of cost-benefit analysis).

We can trace the modern trend from Jimmy Carter’s Executive Order 12044 (1978), which mandated that “regulatory analyses are performed for all regulations which will result in…an annual effect on the economy of $100 million or more.”17 Ronald Reagan’s Executive Order 12291 (1981) explicitly required comparing costs and benefits: “Regulatory action shall not be undertaken unless the potential benefits to society from the regulation outweigh the potential costs to society.”

The primary document governing regulation, though, is Bill Clinton’s Executive Order 12866 (1993), which replaced Reagan’s order and is still in effect, with amendments by George W. Bush and Barack Obama. On Joe Biden’s first day as president, he issued an order for a new revision of regulatory review, but his order explicitly “reaffirms the basic principles set forth in that order [12866].”18 12866 discusses costs and benefits in much more detail than earlier orders, setting out in its “Regulatory Philosophy” that “agencies should select those approaches that maximize net benefits” and in its “Principles of Regulation” that “Each agency shall assess both the costs and the benefits of the intended regulation and…propose or adopt a regulation only upon a reasoned determination that the benefits of the intended regulation justify its costs.”

To accomplish this “reasoned determination,” the Office of Management and Budget (OMB) published—and is now updating—Circular A-4 to provide “guidance to Federal agencies on the development of regulatory analysis.”19 This is where cost-benefit (or benefit-cost) analysis formally enters:

Benefit-cost analysis [BCA] is a primary tool used for regulatory analysis. Where all benefits and costs can be quantified and expressed in monetary units, benefit-cost analysis provides decision makers with a clear indication of the most efficient alternative, that is, the alternative that generates the largest net benefits to society (ignoring distributional effects). (p. 2; parenthetical in original)

Executive branch agencies write their own manuals to guide staff in producing Regulatory Impact Analyses that conform to OMB standards. At this final level, it becomes fully explicit that the Kaldor-Hicks potential compensation criterion is the measure used to decide the social welfare effects of regulation. Here, as an example, is the Environmental Protection Agency’s Guidelines for Preparing Economic Analyses (2010): “The Kaldor-Hicks criterion is widely applied in welfare economics and managerial economics. It forms an underlying rationale for BCA.”20

To summarize, then, at the heart of the procedures that all federal agencies are required to follow in order to justify government action is this claim: the Kaldor-Hicks potential compensation criterion “provides decision makers with a clear indication” of which course of action “generates the largest net benefits to society.” And this claim is just not true.

Violating the prohibition on “ignoring distributional effects” that is built into the logic of Kaldor’s and Hicks’s argument does not result, as Circular A-4 implies, in a slightly-imperfect approximation of net benefits, but in analysis that is fully blind as to net effects. We just don’t know what transferring Janet’s pie to Albert means; the welfare effects of uncompensated transfers are indeterminate. Consequently, a “reasoned determination” is exactly what a cost-benefit analysis based on merely potential compensation cannot produce.

We are starting to get a complete picture. In the Great Depression, welfare economists confronted a deep problem: the impossibility of interpersonal utility comparisons meant that the social welfare effects of public policies could not be aggregated without arbitrary value judgments; the reasoned determination they sought for their policy prescriptions foundered on that impossibility. Nicholas Kaldor and John Hicks argued that redistributing the gains due to wealth-increasing policies from the winners to the losers—so that the policies make everyone better off—created a value-neutral path to an objectively better world. But this path required close collaboration between economists and politicians to ensure that wealth enhancement and wealth distribution happened together.

Starting with Tibor Scitovsky and continuing through the development of cost-benefit analysis, many economists argued—incorrectly—that wealth enhancement alone was sufficient, that the potential to compensate losers, without actually doing so, was all the argument of Kaldor and Hicks required to infer welfare gains. In this way, cost-benefit analysis claimed the objectivity that Kaldor and Hicks had earned, while replacing their objectively justified enhance-and-redistribute procedure with a morally contentious wealth maximization. This unsound distortion of their argument is now institutionalized as the standard and even mandated method of evaluating economic actions.

Based on faulty cost-benefit analysis, thousands of analysts are trained by hundreds of agency documents to do Regulatory Impact Analyses that guide billions of dollars of projects and justify countless legal contracts, and on and on. It can seem overwhelming to address a problem so large and so entrenched. Only platitudes come to mind on where to start. The first step, though, always, is to say what is true, and say what is wrong. Get more minds to focus on the problem.

And stop to look at the big picture, or a bigger one, for there is always a still bigger picture.

Occupy economics

But we know that people’s frustrations run deeper than these most recent political battles. Their frustration is rooted in…the nagging sense that no matter how hard they work, the deck is stacked against them. (Obama, “Remarks by the President on Economic Mobility”21)

A stacked deck is a suggestive metaphor. Stacking the deck violates the rules. It relieves those against whom the deck is stacked of the obligation to accept the game’s outcome. Conversely, the arguments that do justify binding obligations tell us what really are society’s rules and not just the day’s intellectual fashion. Consider an example.

Imagine we are all engaged in a joint project: running a baseball team. Various jobs need to be done: playing the game, coaching, maintaining the field, cleaning the stadium, etc. Since our money comes in at the ticket booths, everybody takes a turn selling tickets to get a share of our earnings. Now, an economic study shows customer willingness-to-pay goes up if players don’t leave the game to man the ticket booths. A new policy of dedicated cashiers selling all tickets raises productive efficiency and increases our total wealth. Like all economic policies, it also changes the distribution of wealth, in this case leaving all the money in the cashiers’ hands. Having read Kaldor and Hicks, we sensibly allow this new policy, and redistribute our increased total wealth by taxing the cashiers to pay everyone else.

Would we find this wealth-enhancing policy equally sensible if the cashiers’ newfound wealth was not used to compensate us for the loss of our share in the joint project? Would we be convinced by the argument that since this could hypothetically be done our overall welfare must therefore be higher? What if we distributed the wealth for forty years, and only then did the cashiers start hiring lobbyists to convince our representatives that redistribution is market-distorting theft that violates the natural order?

When Kaldor and Hicks wrote, the top income bracket’s tax rate in America was 79 percent. Shortly after they wrote, the top rate went up to over 90 percent, where it stayed until a mid-1960s drop back to 70 percent (at the same time as a Great Society expansion of social programs). The 1980s tax cuts brought it to about 39 percent, around today’s rate. From the Depression to the 1980s, then, the tax system was much more progressive than it is now.

One way to look at this history is that for forty years or so, politicians understood the Depression-era compromise with economists—articulated by Kaldor and Hicks—that we would pursue growth (wealth) and compensate the victims of our policies using extensive redistribution through progressive taxation and social programs. Over time, though, politicians and economists forgot that redistribution is compensation for real losses suffered for the common good, not a handout. Part of that forgetting was a carelessness about the arguments that justify cost-benefit analysis, a carelessness that over the last forty years has become embedded in the accepted rules by which we debate and judge our public policies.

In the years since Occupy Wall Street focused public attention on the unequal distribution of wealth, our awareness of the scope and consequences of the problem has only deepened.22 Occupy challenged the idea that the accepted rules are working for our collective betterment. It was widely criticized as somewhat inarticulate, but the pronouncement “I smell a rat” can be a very good start to a more systematic critique.

The investigation of cost-benefit analysis above is one part of such a systematic critique. It shows that we have institutionalized a corrupted version of the analysis. It is a step toward showing that Occupy Wall Street’s complaints—and the even more widespread nagging sense that the deck is stacked—are based not on resentment or ideology or a squishy, self-serving gut feeling about fairness, but on an accurate assessment of society’s mechanisms. Among those mechanisms, the cost-benefit analysis we use to evaluate public policies is not only inherently unable to draw valid conclusions about social welfare, but is actively hostile to exactly those redistributive measures that a sound argument for social welfare requires. Instead of social welfare, cost-benefit analysis promotes wealth.

Occupy’s challenge is rooted in ground that is shared by philosophy and economics. Social contract theory, created by Thomas Hobbes, explains the authority of society’s rules as based on rational self-interest (though the label is often misapplied to any currently prevailing opinion on what those rules are). The social contract is the rules it would be rational for individuals to accept, and which therefore we are justified in presuming each other bound by, even though no one has explicitly accepted them. Economists and politicians widely employ this logic. It is the logic that underlies the claims of economic policy to deference by citizens.

Institutionalizing faulty cost-benefit analysis is, arguably, a fundamental breach of the social contract. Uncompensated victims have no rational reason to accept an outcome “justified” by the potential compensation criterion. That a fundamental breach is occurring is suggested by the evidence in Who Stole the American Dream? and the back-of-the-envelope calculation above about lobbying skewing policy benefits to the wealthy. Why, then, should the 99 percent honor the contract? What keeps them from rejecting society’s norms? What holds us together? Fear? Habit?

Social contract theory—like economics, like philosophy—places a lot of weight on reason. It is a theory that takes seriously both the necessity and the possibility of justifying by reason the demands we make on each other. It takes reason to be a tangible and practical force in the world (a force we might sensibly personify in a morality tale of faith betrayed). When society starts to unravel, when people stop feeling bound by social norms, social contract theory looks not to bad parenting or poor nutrition or television or lead in the soil but to a loss of integrity in the arguments supporting those norms. It predicts that when enough people have the nagging feeling the deck is stacked, it is, in fact, stacked.

Conversely, to restore society’s rules to respect, we need to bring those rules back in line with reason. The danger of a stacked deck is that people will stop accepting the outcome. Thirty or forty years of policies skewed to the rich have led to the pointed question Occupy Wall Street asked: in the terms of our imaginary ball team, that question is, who really owns the money put in the cashiers’ hands by policies meant to help us all? Put another way, if policies are justified by the argument that they raise overall welfare, and this argument only holds water if compensation is actually paid, then is compensation still owed the 99 percent for the costs they bore during four decades of improvements in productive efficiency whose benefits went to the one percent?

The ambition of early welfare economists to address the failure that was the Great Depression by cleaning up the foundations of their work, finding a scientific basis in reason for their policy prescriptions, was the right ambition. An ingenious argument by Nicholas Kaldor and John Hicks eighty years ago accomplished that goal, if only for a while. Their accomplishment was built on a compromise: don’t separate production from distribution; redistribute the fruits of policy to make whole the people it hurts.

Beginning with a specious rebuttal to Kaldor’s and Hicks’s argument by Tibor Scitovsky, a corrupted version of their work came to be institutionalized as the Kaldor-Hicks potential compensation criterion of cost-benefit analysis. This corrupted version grows in strength with each new Executive order, contributes to the wealth inequality that is so acute today, and has brought on a legitimacy crisis very like the one Kaldor and Hicks faced in 1939. It is time to reform the cost-benefit analysis so dubiously created in their names.

I will let the argument of Kaldor and Hicks speak to such a reform: “The main practical advantage of our line of approach is that it fixes attention upon the question of compensation.” We face today the same political habits they faced then: “compensation has not been given, and economic progress has accumulated a roll of victims, sufficient to give all sound policy a bad name” (Hicks 711). The first step, then, would be to disabuse policymakers of the idea that merely potential compensation implies anything whatsoever about net social welfare. It does not.

Once the need for actual compensation is recognized as a requirement of a sound analysis, the second step would be to produce a specific accounting of who gains and who loses from any policy change; this is the minimum necessary to know what compensation could “render it approximately innocuous from the distributive point of view” (Hicks 712).

To redeem sound policy from its bad name, though, requires more than the minimum. It requires us, as Kaldor puts it, “to secure unanimous consent” (551n1) “or at any rate to make some people better off without making anybody worse off” (550), as he more practically puts it. Third, then, we must incorporate factors beyond money, “for individuals might, as a result of a certain political action, sustain losses of a non-pecuniary kind” (551n1). These, too, must be compensated to secure a net benefit.

Lastly, since “the best methods of compensation feasible involve some loss in productive efficiency, this loss will have to be taken into account” (Hicks 712). We have to pay the transaction costs of actual compensation. All four of these steps comprise what Kaldor and Hicks called the “economist’s part” of a sound analysis of the costs and benefits of any policy change, for “the task of the welfare economist is not completed until he has envisaged the total effects” (Hicks 712).

Inescapably, after the economist’s part is completed, the decision still rests with the politician whether to implement a given policy, and whether to compensate those it harms. But if cost-benefit analysis deserves a bad name for its part over the decades in creating bad policies, Nicholas Kaldor and John Hicks do not deserve the blame. The faulty potential compensation criterion that today bears their names is thoroughly contrary to the logic and the spirit of the ingenious program they created in 1939. The path they showed us remains the path to reforming cost-benefit analysis.

Notes

Jerry Cayford is a philosopher and policy analyst in the Washington, DC area. He is the co‑author with Thomas C. Beierle of Democracy in Practice: Public Participation in Environmental Decisions and with Michael R. Taylor of “American Patent Policy, Biotechnology, and African Agriculture: The Case for Policy Change.” Dr. Cayford’s solo works include “Breeding Sanity into the GM Food Debate.” His Ph.D. in philosophy is from Northwestern University, and bachelor’s in economics from U. C. Berkeley.