Pareto improvements and Kaldor-Hicks efficiency criterion
This page outlines the role and the limitations of the concepts of Pareto improvement, Kaldor-Hicks efficiency and cost-benefit analysis in the development and implementation of Government and regulatory policies, programmes and projects.
Nature and limitations of the Pareto and Kaldor-Hicks criteria
In a trading system with a fixed set of participants, a change is a Pareto improvement if it means that at least one participant would favour the effects of change (be "better off") and no participant would oppose the effects (be "worse off").
Within the same context, a Kaldor-Hicks improvement is defined as a change that is either a Pareto improvement or such that:
- the "winners" from the change would be able to compensate the "losers" and still be better off (Kaldor criterion); and
- the "losers" could not afford to bribe the "winners" to prevent the change (Hicks criterion).
Crucially, the compensation or bribe elements of the test for a Kaldor-Hicks improvement is a hypothetical one: the change is considered an improvement if the assessed winners' gain is greater than the assessed losers' loss, regardless of whether the change when implemented would actually involve the payment of any compensation.
A state of affairs can be said to be Kaldor-Hicks efficient if there exists no Kaldor-Hicks improvement away from that state.
There are a number of ways in which these concepts can be customised or refined. For example, the efficiency test can be applied at the level of the individual decision or action (e.g. whether to operate a polluting factory) or at the rule-making level (e.g. whether to establish a system of pollution permits).
An incomplete and imperfect way of ordering policy options
None of these tests can claim to provide a comprehensive guide to policy development in the sense of a complete ordering relationship by which different policy options can be ranked.
In many areas of policy, particularly where it affects a large and diverse population in complicated ways, no change is likely to be a Pareto improvement one because there is always likely to be at least one person who would be adversely affected by the change: in other words the ordering relationship provided by the Pareto criterion is such a partial one that "too many" states are Pareto efficient.
The Kaldor and Hicks criteria, and the Kaldor-Hicks combination, were developed to address this limitation. However, they suffer from other, similar, limitations: in particular Kaldor, Hicks or Kaldor-Hicks efficiency are only well-defined in terms of a "local optimum", which is to say if the only changes under consideration are (infinitesimally) small marginal ones. When comparing larger (finite) changes, the ordering relation implied by either of the Kaldor or Hicks compensation criteria may not be anti-symmetric (the Scitovsky paradox) or transitive.
An analysis limited by its assumptions about the affected population
The concepts of a Pareto improvement or the Kaldor-Hicks efficiency criterion (or other variants) relate to the preferences that each participant in a system has for different outcomes or states of affairs. For each of these efficiency concepts, the test only takes account of the preferences of the specific set of participants to which it has been applied, without paying any attention to parties that are not defined to be included.
In particular, for a change that would bring new parties into existence (individuals or firms), rather than simply affecting the participants who are part of the system whether or not the change takes place, no account can be taken of these parties other than, perhaps, through the preferences that the existing participants have towards the formers' existence.
Similarly, no account can be taken of the preferences of parties that would be removed from the system as a result of the change (i.e. current participants in the system who are eliminated but would not have been eliminated in the counterfactual, or future participants who would enter the system in the counterfactual but are prevented from entering the system under some of the options under consideration): it would be futile to try to conceive of whether a party prefers one state of affairs to another when that party only exists in one of those states.
This impairs the legitimacy of the use of Pareto or Kaldor-Hicks criteria for social decision-making, since these criteria exclude consideration of seemingly relevant factors such as the interests of individuals or firms whose very existence is affected by the outcome of the decision-making process.
The fact that the existence of a Pareto or Kaldor-Hicks improvement relates only to a specific population assumed to be independent of the decisions or options under consideration also leads to further imperfections in the ordering relationship provided by these economic efficiency criteria. For example, it is conceivable that a policy action may be a Pareto improvement in the eyes of the population before the change, but that once the action is implemented, knock-on effects to (i) the composition of the population or (ii) the preferences of the population (which will depend, of course, on information gained through experience) may mean that reversing the action becomes a Pareto improvement in the eyes of the "new" population. If this effect was foreseeable, which it might well have been, then was the initial assessment that the action was a Pareto improvement correct? It seems that recognising the effect of decisions on the composition and/or preferences of the population means that even the Pareto criterion does not guarantee logical consistency. (This issue is additional to the very partial nature of the Pareto ordering and the non-transitivity of the Kaldor-Hicks criterion outlined above.)
A common approach to policy, programme or project assessment, appraisal, evaluation or impact assessment is to use a Kaldor-Hicks efficiency criterion as the basis on which "costs" and "benefits" of various options can be combined into an overall assessment of which of the options ought to be chosen.
In many cases, this is a relatively straightforward process, at least at the conceptual level. For example, a sectoral regulator responsible for approving changes to an infrastructure access agreement and/or charging structure may reasonably require that the changes be in all parties' interests (Pareto criterion), or at least that any deviation from the Pareto test should be limited to Kaldor-Hicks improvements which cannot be transformed into Pareto improvements only because of a failure to arrange for "compensatory" lump sum transfers (e.g. on transaction cost grounds).
This reasoning explains the focus of regulators on such circumstances to ensure that costs are recovered in full from the parties that cause them to be incurred; that companies that make (productive) efficiency improvements should retain at least some of the financial benefits; and that compensation be paid for service quality failures or effects such as pollution, especially (e.g. in the case of monopoly network operators or "externalities") if no competitive pressures can be expected to ensure that those who can influence such failures take proper account of the impact of failures on users.
Problems with economic efficiency criteria in policy development and implementation
The remainder of this page explores the problems faced by attempts to rely on economic efficiency criteria or cost-benefit analysis to support policy, administrative or legal decisions. Having concluded that systematic reliance a Kaldor-Hicks efficiency concept would be incorrect, the discussion then turns to the extent to which such methods might be justified in some cases.
Efficiency criteria may lead to an unnecessary diversion
Economics regulators were used above in an example of a proper use of Pareto or Kaldor-Hicks efficiency criteria to inform administrative decisions.
But it should be recognised that regulators often have to deliver other purposes than the Pareto or Kaldor-Hicks one of ensuring that all mutually beneficial economic exchanges (or all exchanges that would be mutually beneficial if compensation transfer could be made without frictions) occur between industry participants.
For example, regulators of essential infrastructure industries such as water, transport or energy networks are often given duties to set prices that will promote the interests of users, subject to the constraints of ensuring that investors have an opportunity to receive a reasonable return on their investment and that the company can raise any additional financing that the service requires. Whilst it may be possible to use cost-benefit analysis to "explain" such objectives or constraints on the basis that, say, a breach of trust to investors would increase the cost of capital in the future and in other regulated sectors, to the detriment of customers, we would submit that it is simpler, more honest and more effective to accept these duties as arising from statute (e.g. the regulator's statutory duties) and legal precedent and doctrine, and to implement them without the diversion of cost-benefit analysis or impact assessment.
The situation of competition authorities is often similar. Their duty is to detect and remedy breaches of the law (e.g. Article 81 or Article 82) or to assess transactions or markets against criteria set in law (e.g. the EC Merger Regulation or the Enterprise Act 2002). Perhaps these criteria were partially informed by cost-benefit analysis at the level of rule-making — e.g. a possible interpretation of Article 81(3) is that it restricts anti-competitive agreements to those that improve the productive efficiency of an economic sector to the benefit of the relevant customers. Perhaps they were derived from other objectives — e.g. a benefit of Article 82 may be thought to be the protection of the freedom of every entrepreneur to compete without being excluded by the exercise of market power by dominant incumbents, and the protection of customers against certain forms of exploitation of a dominant position which were considered undesirable by the powers that be in the same way that slavery might be considered undesirable.
Irrespective of the political or constitutional rationale behind the rules that regulators are responsible for implementing, it is quite possible, and appropriate for an independent regulator, to understand and enforce the rules on the basis that they have now acquired a life of their own through years of practice and legal precedents.
Thus a potential danger of cost-benefit analysis is that it may end up being a diversion from the proper purpose of Government bodies. This might be a pure intellectual diversion — why worry about complicated and unnecessary things when your duty is to meet objectives which are, in fact, simpler and clearer? In some cases, it might be even more harmful, if it leads supposedly independent and non-political judicial or administrative authorities into an unnecessary analysis of the political underpinnings of the policies or laws that they are meant to implement, thus compromising their legitimacy.
Efficiency criteria are inapplicable in some cases
The other key potential risk associated with excessive reliance on cost-benefit analysis and efficiency criteria relates to cases where the underlying assumption of the Pareto or Kaldor-Hicks are not satisfied, to the extent that an analysis that relies on them becomes meaningless.
This may arise for two categories of reasons.
- Information-related reasons. If the effect of the decision under consideration is not a change in the level or distribution of the consumption (and production) of reasonably familiar goods or services, but instead triggers the emergence of completely new goods and activities — say, preventing or permitting the universal adoption of a technological inter-operability standard — then it is not possible to assess willingness to pay or willingness to accept on the basis of revealed or stated preferences, since the counterfactual is difficult to imagine (this might be seen as an example of Hayek's "pretence of knowledge" argument). In addition to this issue, the use of the Kaldor-Hicks criterion for changes that are not marginal is impaired by issues such as the Scitovsky paradox: using the Kaldor-Hicks criterion on a range of options may lead to logically inconsistent findings.
- Population-related reasons. If the effect of the decision under consideration affects the identity of the participants included in the population, for example by preventing entry by a firm and thereby leading to the non-existence of that firm, or by affecting social or economic conditions in such a way that children who would otherwise have been born are not actually conceived, then neither Pareto nor Kaldor-Hicks criteria can be applied to the population as a whole. It is absurd to ask whether a person would prefer to exist or not to exist, or how much consumption of other goods they would be willing to forgo in exchange for being created or born! (It would of course remain possible to assess whether a change is a Pareto or Kaldor-Hicks improvement in the eyes of a specific set of people whose existence is not affected by the decision.)
These risks manifest themselves across a range of areas of law and policy. They are particularly significant in the context of policies or rules which affect the dynamics of competitive market processes, since such interventions may affect the ability of different players to enter markets, whether new products are invented or not, and otherwise affect the identity of the players and the nature of the "economic game" that they play as much as it affects the particular outcomes.
In such a case, it is simply absurd to ask whether the individual or companies prefer one or another outcome, and how much they would be prepared to pay for them: these companies might not have been formed in the counterfactual, or might have a completely different structure; and individual people might have completely different skills and experiences — or indeed they might have died, or never been born.
Efficiency criteria are only relevant insofar as they can be inferred from the relevant law or specific policy objectives
The above argument, taken on its own would lead to a total rejection of cost-benefit analysis and efficiency criteria in all cases. For almost any decision worth worrying about, it is bound to be the case that any policy or legal decision may affect someone's life somewhere in a dramatic way that makes reliance on the preferences of that person (or non-person if he or she would not exist in the counterfactual) conceptually impossible in all cases.
Whilst this line of thought leads, correctly, to the utter rejection of cost-benefit analysis as an analytical technique for social decision-making if it has to stand on its own merits and logic alone, it does not in fact invalidate all uses of cost-benefit analysis and economic efficiency criteria in practice.
However, it does reveal the need for a justification of the use of such techniques in each case, based on a legal or policy (or moral rights) analysis in the particular circumstances under consideration. Economic cost-benefit analysis cannot justify itself.
The proper justification will depend on the circumstances:
- In the context of negligence (e.g. in torts law), the relevant duty of (reasonable) care can be analysed as an obligation of the person owing the duty to incur effort and cost to the benefit of the person to whom the duty is owed up to the point where the "willingness to pay" of the beneficiary for more effort would be less than the "willingness to accept" of the person owing the duty for that effort. This is a Kaldor-Hicks efficiency concept in which the participants are the person(s) owing the duty and the person(s) to whom the duty is owed — effects on other people are not to be taken into account (unless they are involved in another duty of care relationship).
- In a regulated industry context, the regulator will often design or authorise particular structures of access contracts or other industry arrangements which are subject to regulatory approval on the basis of efficiency tests which are in effect based on a Pareto or Kaldor-Hicks criterion for which the scope is the industry, or the industry and its customers. This can be inferred, for example, from a statutory duty to promote efficiency and economy in the management of the sector, or to promote the interests of users. Again, the analysis makes no pretence to take account of all effects on affected parties: indeed consideration of the interest of users is properly limited to their interests as users, disregarding other interests that the same individuals may well have, e.g. as workers or investors.
- Corresponding arguments can be developed for pollution control, management of common resources, and so on. The key point is that the exact scope and nature of the cost-benefit analysis is particular to each case, and derived from an understanding of the rights and wrongs or of the relevant policy objectives in that case.
In these example, the analysis may share many "mechanical" features with cost-benefit analysis as defined in economics textbooks (or with the forms of analysis advocated in Government appraisal guidance documents listed below) but there is a crucial difference: the scope of the economic analysis is defined by a proper understanding of the specific rationale for using cost-benefit analysis in each specific case. Thus, technical economic analysis no longer pretends to stand on its own as a guide to social choice, but instead is seen as a set of techniques that is subservient to the legal, moral or political purposes that have been found to be relevant for the particular social choice that has to be made.
Links
Some annotated links to related online content are provided on the Cost-benefit analysis links page.
This page is managed by Reckon LLP. Users are encouraged to contribute relevant views and links. Use the "Edit this page" or "Add a comment" button above to edit the text or to add comments and notes. Advice on making contributions (including trackback) is in the Guidelines.
Last changed by 63.123.252.5 at 11:47 PM on Wednesday 4 February 2009.