The Axioms Explained

KH: An Adjusted Kaldor-Hicks Measure

By the KHZ tests, I mean a set of tests in which a project is evaluated by willingness to pay (WTP) measures for gains and by willingness to accept (WTA) measures for losses. I will show that these are not equal to the sum of either the compensating CVs or the compensating EVs but rather to some combination of them. Unlike the CV or EV measures alone, I will show further that these measures satisfy a Kaldor compensation test.

By building on the KH measures we help to ensure that our new criteria will meet the proposed tests of acceptability and usefulness. The KH measures have been the practical standard for project analysis since they were introduced in the 1930s. They have persisted as the standards of practical art despite voluminous technical economic criticism, trenchant ethical criticisms, and overly extravagant claims by some of their supporters.5 Thus they represent a good starting point for a measure attempting to be acceptable and useful.

The KH measures, however, have rested on using either a CV measure or an EV measure. Yet, as Chapter 1 noted, some projects which have a positive sum of CVs may not meet the Kaldor potential compensation test.

The KHZ measure requires no potential compensation test rationale. The revolutionary aspect of the KHZ approach to benefit-cost analysis is that it abandons the traditional potential compensation test, yet it retains the essential features of the Kaldor-Hicks criteria. To abandon the potential compensation test is to jettison the baggage of the efficiency-equity dualism that, as we have seen, has been a part of normative economic analysis since the 1930s. Yet the proposed test for economic efficiency remains true to the essence of the KH criteria in its reliance on the sum of compensating and equivalent variations.

Many criticisms of benefit-cost analysis are result driven: for example, someone complains that the use of a discount rate is immoral in its treatment of future generations. Regardless of the merits of these criticisms (though the criticisms often have little merit), by assuming that the purpose of benefit-cost analysis is to provide information to decision makers, I can get benefit-cost analysis 'off of the hook' whenever its use leads to a result that is unappealing for ethical reasons. Moreover, my statement of the purpose of benefit-cost analysis accords with reality. Benefit-cost analysis itself is not capable of implementing a decision: only individuals can make decisions. At most, benefit-cost analysis can help an individual make an informed decision. There is probably no evidence that the public would want decision makers to be wholly constrained by the results of a benefit-cost analysis. In part, this may reflect an understanding that not all considerations are likely to be captured by a benefit-cost analysis. No benefit-cost analysis will be perfect.

Because benefit-cost analysis performed according to the KHZ criteria rests on existing social sentiments, its ethical value depends on the ethical value of those sentiments. If one argues that the 'right thing to do' should be determined without considering the current sentiments of a society, benefit-cost analysis does not provide the relevant framework to respond to that argument. For example, a study might conclude that it would be KHZ efficient to legalize narcotics, if there was evidence that the sentiments against the use of narcotics had declined or had disappeared. However, a person who felt that the use of narcotics was immoral - regardless of whether or not society was opposed to narcotics - would not find the study persuasive.

Psychological Nature of Benefits and Costs

There is a distinct tendency in economics and decision theory to view material assets as the main carriers of utility (Kahneman and Varey 1991, p. 130). As important as material assets may be, however, choices are not well explained without considering the psychological aspects of choice and of well-being (Kahneman and Tversky 1979). Economic analyses that incorporate regret (Loomes and Sugden 1983), that recognize loss aversion (Kahneman and Tversky 1979), and that recognize framing and positional considerations (ibid.) have improved predictability of decisions (Thaler 1991).

We seek a measure of value that tries to determine whether or not people will be better off as a result of a new rule or project and which determines which people are likely to be better off. Such a measure must then incorporate information about psychological states and subjective value. We will say then that the correct economic measure of value is one that best measures well-being and that this measure is a psychological measure (Knetsch 1997, p. 509). The psychological reference point is the correct one because benefit-cost analysis rests - and has always been thought to rest - on the preferences of individuals (Kahneman and Varey 1991).6 This reference point represents what I will call 'psychological ownership.' Both losses and gains are subjective. Both are felt psychological states (ibid.). Moreover, the measure of a loss is different from that of a gain because it is frequently felt differently psychologically (Kahneman, Knetsch, and Thaler 1991). The economic theory, then, is clear: one should use willingness to pay (WTP) for gains and willingness to accept (WTA) for losses. A loss is determined by what one has to lose. A psychological reference point determines whether a positive change is a gain or a loss restored, or whether a negative change is a loss or a gain forestalled.

Information Available information

The KH measures rest on the compensating or equivalent variations, which are measurable, or on consumer and producer surpluses, which are also measurable estimates of the CV or the EV. The fact that the proposed criteria rest on measurable data and on information that can be gathered (even if imperfectly) is, of course, a major reason for their usefulness and success. The KHZ measure suggests expanding the range of goods or values that must to be represented and, to this extent, imposes greater information requirements. These requirements do not, however, go beyond the possible and are justified by the gain in usefulness of the criteria.

Better information

This aspect of the axiom has implications for situations in which different preferences arise from different information. The choice one makes quickly may, for example, differ from the one that would have been made after greater reflection. This assumption allows us to include the more reflective decision in the benefit-cost analysis, where this can be determined. I do not say, however, that more information is always worth the costs.

No utility

In economics the clearest and most trouble-free use of the term 'utility,' is that which represents 'preferences' and not 'good' (Broome 1991, p. 10).7 Utility in economics is a concept that is convenient for modeling choice decisions. It need be nothing more. It need not be a measure of hedonistic pleasure, or a measure of satisfaction, or of happiness, nor should it. Even when using utility as an argument in modeling the economics of normative decisions, economists should not define utility as a hedonistic measure. Why should the profession have a measure of something that cannot be measured? To claim such a measure is both unscientific and unnecessary, as well as false.

There is something unscientific about using, as a measure of welfare, a term such as 'utility,' which cannot be measured. Economists must feel ambiguity about retaining the notion of utility at least in their more secret moments. Like the old 'ether' of physics, the concept of utility does little work for us. The convenience in its use is now outweighed by the confusion it causes. Most approaches to normative economics rest on utility-based formulations (Hammond 1985). I drop that approach here. The KHZ criteria approach to determining preferences is to determine the purchase or sale price of anything that people care about.

Costs of Change Transactions costs

A fundamental insight of Coase was that in determining inefficiency economists compare a world without transactions costs with the real world. This leads to absurd results but more important it also taints the usefulness of KH efficiency. When transactions costs are ignored every transaction is inefficient as compared to a transaction in a hypothetical world in which transactions are costless. When transactions costs are ignored, externalities are ubiquitous as Zerbe and McCurdy (1999) have shown.

There is now a large, not to say enormous, literature on this issue that suggests that it is appropriate to include transactions costs in the determination of what is efficient (e.g. Allen 1991; Coase 1960, 1964, 1974, 1988; Eggertsson 1990; Barzel 1985; Baumol 1979; Randall 1983; Nelson 1987; Medema and Zerbe 1999a, 1999b; Zerbe and McCurdy 1999). The seminal source is Ronald Coase (1960). As Coase has pointed out, little can be learned from the study of theoretical optimal systems (Coase 1964, p. 195).8 Analysts who become enamored of 'blackboard economics,' in which equations are substituted for underpinnings, produce concepts that bear little correspondence to the actual social system. The world portrayed is one that exists only on the blackboard: 'the analysis is carried out with great ingenuity, but it floats in the air' (Coase 1988, p. 10).

This assumption addresses the criticisms quite properly raised by Coase that a welfare analysis is improper that ignores transactions costs. Coase (1960) has shown that in practical matters the exclusion of transactions costs in one's analysis can lead to absurd results. The absence of transactions costs is the source of the 'failure of market failure' (Zerbe and McCurdy 1999) and the alleged 'inefficiency of Christmas giving.' Economists or lawyers may speak of a particular change in the state of the world as being efficient when what they mean is that the state of the world would be efficient if transactions costs were zero.9 This practice, which is endemic throughout the legal and the economics literature, has led to considerable misunderstanding (Coase 1960; Zerbe and McCurdy 1999; Medema and Zerbe 1999a, 1999b) and to the development of concepts of market failure that are at worst incoherent and at best poorly thought out (Zerbe and McCurdy 1999).

A welfare measure will be more accurate, and therefore better, if it includes the property transactions costs associated with the change contemplated, in accord with Coasean welfare economics.10 The KH tests were devised to be practical measures that would be applied to changes in the real world, not to an ideal world in which transactions costs are zero. Not to include the costs of change is to ignore a component of transactions costs. Ignoring the costs of change has, in fact, opened the door for confusing comparisons that ignore various other components of transactions costs (Zerbe and McCurdy 1999).11 When two states of the world are compared and the analysis of the changed world does not include the costs of making the change, one cannot reasonably say that a change is desirable. Thus, anyone who ignores the property transactions costs of a proposed change itself cannot say that a change is 'efficient' if they mean to imply that a change is desirable. The failure to use the term 'efficiency' more precisely creates ambiguity about whether or not one is speaking of efficiency in a zero transactions cost world or not.

I wish to distinguish between the transactions of economic exchange and the transactions costs of making rule changes. The first, what I here call property costs, are the costs of making exchanges and protecting property given the legal regime (Allen 1991). These I define broadly to include search and information costs as well as the costs of registering ownership, negotiation over price and quality, and, in general, the whole panoply of property right maintenance costs and exchange costs. The second sort of transactions costs, that I call rule change costs, are the costs of changing the legal regime. The proposal here is that property transactions costs should be included in evaluating economic inefficiency.

Rule changes

At the same time I do not purpose to include rule change costs in the KHZ measure of efficiency. To do so would make the world tautologically efficient. For if the all costs associated with a rule change, including the costs of enacting the rule, are to be included, then in some sense this rule should already be in effect. That it is not in effect must be due to some (unspecified) costs. Thus when an economist or lawyer says that some new rule is efficient, or that some existing rule is inefficient, he or she will be taken to mean that the rule would be efficient or inefficient aside from the costs of enacting or eliminating a rule itself.

Compensation test costs

The hypothetical cost of administering a system of compensation should not be considered, unless the decision maker actually intends to compensate those harmed by a change. The purpose of KHZ is to furnish information about gains and losses from an actual change. To include costs that are not incurred would reduce the informational value (usefulness) of the KHZ measure.


Definition of a good

Axiom 6 (values) appears straightforward and perhaps innocuous. It is neither. Its formal adoption has substantial, perhaps profound, implications for normative economic analysis.

The axiom is, however, within the spirit of KH. Even for the KH criteria, no well-founded reason has been suggested for limiting the goods to be covered by the criteria. Consider the following description of the KH process suggested by Talbot Page (1992, p. 102):

You are asked to compare two worlds. The first is the status quo: the world the way it is now. The second is identical with the status quo except for a change brought about by the project. In the comparison, you take into account the ramifications of the project, differences in income to you and others, differences in habitat, and so on; but except for the changes brought on by the project, the two worlds are the same.

Suppose that you value the first world more highly than the second. Then you are asked what is the minimum you need to be compensated so that you would value the change (with the compensation) just as much as the status quo. If you value the world with the project more than the status quo, then you are asked how big a payment you could make in the changed world (with the project) so that you would just value equally the status quo ... The economic criterion says that if the sum of all the compensations (to those who would lose by the project) is less than the sum of the equilibrating payments (from the gains from the project), then the change from the status quo is worth making.

Somewhat more technically, in the opening remarks of their well-known book on welfare economics, Boadway and Bruce (1984, p. 1) note:

A social ordering permits one to compare all states of the world and rank each one as 'better than,' 'worse than,' or 'equally good as' every other. Ideally we would like the (social) ordering to be complete (so that all states could be ranked or ordered) and transitive . The term 'state of the world' can be interpreted as a complete description of a possible state of an economy including economic characteristics, political conditions such as freedom of speech and non-discrimination, physical characteristics such as the weather, and so on.

What is common to the statements of Page (1992) and Boadway and Bruce (1984) is that a 'good' is implicitly defined by what is of value.12 A 'good' in economic analysis is - or at least should be - defined by what people care about. This goes well beyond just physical attributes.13 People care about the fairness and efficiency of rules,14 for example, so that we can also call the fairness of a rule or a project a 'good.' Because people care about fairness, they will often not just care about the cost of a project but about who pays for it, about whether or not the project is an efficient one, and about whether or not the beneficiaries are deserving. They care about who pays for a good even when they do not expect to pay nor to be paid. Similarly, they may feel more strongly about being uncompensated for a loss when they consider compensation fair than when they feel otherwise. And, aside from specific compensation for harm, they often care about the equity implications of income distribution itself, so that the income distribution is itself a good.

Missing values

The economic literature, including very recent literature, has many examples in which goods that meet the test proposed here are, apparently arbitrarily, omitted in attempting to determine economic efficiency. The goods which are ignored include, for example, the ethical value attached to actions, the value associated with income distribution itself, sentimental value, and demoralization costs. These sorts of omissions reduce the acceptability of the KH criteria and should be avoided when using the KHZ criteria.

Income distribution Most of the criticisms of KH rest on its omission of income-distribution considerations in determining efficiency. Income distribution did not traditionally qualify under KH. However, income distribution does qualify as a good under KHZ, because people other than those directly affected nevertheless care about income distribution.15 In Chapter 1, I explained the historical decoupling of distributional equity and efficiency. This decoupling did not, however, achieve the value-free criteria that were desired. Other more recent reasons given for the decoupling are (1) that there is less agreement about income distribution questions than about traditional efficiency (Posner 1985a, p. 104) and (2) that as a practical matter it is better to separate the equity and the KH efficiency issues (Posner 1985a, pp. 104-105; Polinsky 1983, pp. 105-113).16 As a matter of fact, it is unclear whether there is less agreement about distributional matters than about traditional efficiency questions. But unanimous agreement is not required for KHZ analysis. Under KHZ analysis the purpose is to furnish useful information to the decision maker. Information about equity can be useful to the decision maker, as I will show. Whether it is practical to ignore equity and focus on traditional efficiency is a question of usefulness - a test for the acceptance of KHZ criteria. In Chapter 5 I will show when such separation is desirable and will show that in these cases such separation is consistent with KHZ.

I suggest adopting a criterion proposed by Harberger (1978) for valuing changes in income distribution. This criterion would allow a choice between different distributional states. Mishan (1981, pp. 365-367) has shown that where a choice between first-best states cannot be made on KH or hypothetical compensation grounds, the only remaining considerations in making a choice are distributional. By including a method of evaluating distributional choices, I increase the ability to make a choice between first-best states.

Compensation Because people who are not directly affected by a project will care about the fact of compensation or the fact that there is no compensation, the presence or absence of compensation is a relevant good. The issue of compensation will receive particular treatment here. The presence or absence of compensation is part of the regard for others.

The regard for others The omission of important goods from most normative economic analyses arises from their failure to consider the regard for others.17 Economic efficiency should be defined with respect to all states of the world that people care about. According to Ernst Mayr (1997),18 people care about others, including the welfare of people to whom they are not related. As Sen (1995, p. 15) notes, Adam Smith himself hardly saw an individual as the narrowly focused 'economic man.' Smith examined the role of 'sympathy,' 'generosity,' 'and 'public spirit'. I claim that people will care about a social change even when they themselves are not otherwise directly affected by it. They will care about fairness, about waste, and about income distribution. I will speak of 'the regard for others' and about the 'kindness of strangers.' By these terms, I mean the concern of some for what they regard as fair outcomes for others, whether or not the regarding parties are themselves directly affected. So, one expression of the regard for others is when we are concerned about the appropriate application of principles of justice in situations in which we ourselves are not directly affected. The failure of traditional KH analysis to consider the regard for others is particularly ironic, since under the logic of KH there is no reason to ignore it.19

One reason why people care about rules which are applied to others is that they believe those rules may also be applied to them. A second reason is that some people will be truly 'other-regarding' or altruistic; they care about the fairness of justice of rules and outcomes applied to others even when they do not expect to be subject to the rules or outcomes. I call this 'ethical' or 'altruistic regard for others.'20 Mayr maintains that genuine altruism (as opposed to reciprocal or kinship altruism) and ethics have evolved in humans because they have survival value for the relevant genes through group selection. He notes that 'as history has repeatedly illustrated, those behaviors will be preserved and those behavioral norms will have the longest survival that contribute the most to the well-being of the cultural group as a whole' (1997, p. 254). In other words, ethical behavior for humans is adaptive.21 Both the regard for others that arises from self-regard and the regard for others which is truly altruistic are part of our felt experience.22


Critics of benefit-cost analysis sometimes assert that it will inevitably ignore legal and moral rights (Lothrop 1986).23 This is profoundly wrong (Sen 1995; Zerbe 1998b). Admittedly, as Sen noted, 'the violation or fulfillment of basic rights and liberties tends to be ignored in utilitarian welfare economics ... particularly because of its "welfarism" whereby states of affairs are judged exclusively by the utilities generated.' To ignore rights is, however, incorrect even by 'welfarism's' standards, since the psychological consideration of value mentioned previously shows that rights affect one's valuation of goods. More to the point, my approach is not one of 'welfarism.'

The fact that a person has a recognized legal right to a good is some evidence that it is efficient to allow that person to possess the good. Similarly, the fact that a person does not have a legal right to a good is some evidence that it would be inefficient to allow that person to take the good from its lawful owner without permission. Although all values that may be measured by the WTP and the WTA should be considered, some goods and some individuals have been excluded from the judge's implicit benefit-cost analysis because those individuals have no legal right to those goods. The decision to deny a legal right to possess a good is itself frequently the product of a prior benefit-cost test. The interrelation between rights and benefit-cost analysis is a major theme of this book, which I explore in depth in the following chapters. The recognition of property rights, for example, may be held to rest also on a previous KHZ test that such recognition is desirable. No rule for normative decision making can be widely acceptable if it does not recognize as a starting point the existing pattern of rights. This is as it should be. How could it be otherwise? Why would one want it to be different? Judges recognize rights in making decisions. Should they not? As Heyne (1988, p. 56) notes, 'economic theory takes for granted, far more extensively than economists seem generally to recognize, the normative force of established rights and obligations.' This is even true of attempts to predict. Heyne goes on to say, 'economists who want to predict the effects of agricultural price supports must assume, inter alia, that the existing property rights of farmers will be respected, that public servants will carry out the provisions of the law, and that taxpayers will provide the amounts which they are assessed to subsidize the program' (1988, p. 56).

0 0

Post a comment