8. The ‘Mindless Society’

In the last chapter I introduced the concept of the social welfare function – that is, a function describing what variables contribute to the welfare of society as a whole. But the British Prime Minister in the 1980s, the late Baroness Thatcher, is famously misreported as having said that there is no such thing as ‘society’. What she actually said is ‘Who is society? There is no such thing! There are individual men and women and there are families…’¹ It is not clear what exactly she had in mind. One possibility is that she was simply referring to the view expressed by Jeremy Bentham in 1789 that ‘the community is a fictitious body, composed of the individual persons who are considered as constituting as it were its members’. ² Alternatively, she may have had in mind Pareto’s view that it makes more sense to talk about the welfare for society, rather than the welfare of society.³ Simon Blackburn has suggested that she was reflecting ‘the ideology of the self-interested agent in eternal competition with others’.⁴ Another – and more charitable – interpretation is that she was making a philosophical objection to the metaphysical notion that one can attribute states of mind to ‘society’ of the same kind that one can to individuals. It is not for me to attempt to adjudicate between these various interpretations and the reader must judge for himself.

But whatever the correct interpretation of the late Baroness Thatcher’s view there is no doubt that it is linked to a serious aggregation problem. This is the problem of how to go from individual preferences to the preferences of society. It is a well-known problem in welfare economics and in political philosophy which we shall discuss in more detail in the next section. The problem arises because the whole edifice of welfare economics is built up on a fairly clearly defined theory of individual choice, some components of which – such as the way consumers respond to changes in prices – has been empirically confirmed. But even if one could assume that these choices correspond to their welfares it is not clear how far one can go from this to a theory of what constitutes the welfare of a group of individuals. How does one weigh together the welfare of individuals to arrive at the welfare of the group?

Another problem, which is less obvious and far less important for practical purposes, follows from the fact that our theory of whether the utility of an individual has risen or fallen is in terms of whether the individual is on a higher or lower ‘indifference’ curve (or ‘behaviour line’ as it is sometimes called). Hence, strictly speaking, one cannot compare the welfare of societies comprising different people with different indifference curves. This gives rise to what is known as the ‘constituency principle’, namely that comparisons of welfare are only valid if they refer to identical groups – that is, a ‘constituency’ of people. But this would even mean that if the GDP of some country were much higher today than yesterday one could not say that its economic welfare had risen. For some people would have died during the previous 24 hours and others would have been born. So we would be comparing societies with different populations, which, strictly speaking, is not possible within the framework of economic theory.

But all such limitations on the possibility of passing judgement on the economic welfare of different societies are nonsense, of course. For example, strictly speaking, the constituency principle would exclude our saying that welfare is higher in a country where people are free, well-fed, healthy and happy than in some other country where people are oppressed, hungry and miserable. But we know that this is nonsense.

So what do we mean when we talk about the aggregate welfare of ‘society’? It is intuitively plausible to assume that society does not have a ‘mind’ in the ordinary sense of the word so that one cannot attribute to ‘society’ the sort of mental thought processes that correspond exactly to those that an individual goes through when making choices. This means that, for example, society does not follow thought processes such as those that correspond, ideally, to the rational choice process that an individual would follow. There is no such thing as a ‘society’ that has a mind that could say to itself, ‘Oh dear, silly me! Since I preferred X to Y and Y to Z how stupid of me to prefer Z to X’. This is particularly inconvenient for economists – who are the guardians of ‘rational choice’ and who like to equate rational choice with the maximisation of something, such as a utility function, or a social welfare function, that respects some criteria of rationality.

One implication of this is that whereas one can make empirically verifiable positive propositions about the preference rankings of any individual, it is not possible to make similar propositions about the collective preference rankings of any collection of individuals. For example, if we assert that Jones thinks he prefers situation 1 to situation 2, we are making a positive proposition that could be tested by asking Jones. If he says, ‘Yes, I think I prefer 1 to 2’, our proposition has been confirmed, unless we have some reason to suspect he is lying. But we cannot make a similar testable proposition about the combined preferences of a number of individuals. We can, of course, add up how many prefer one or other of the two options, like in a referendum. But that would simply amount to defining society’s preference as being what the majority want.

Or, as in a market economy, we could check the amounts of money that the agents were prepared to pay for the two alternatives and add them up. But to conclude from these that ‘society’ prefers one situation rather than another merely means that we define the statement that ‘society prefers one situation rather than another’ in terms of survey results or people’s expenditure patterns.⁵

Nevertheless, decisions have to be made all the time about various policies or projects, at all levels of public life. The classic examples are public goods such as national defence, or law and order. These have certain well-known characteristics, namely ‘non-excludability’ or ‘non-rivalry’. Non-excludability means that if the service is provided for one or more persons it is technically impossible, or very difficult, to prevent other people from using it. If we have armed forces to protect the British Royal Family from invasion by the French it is impossible to exclude everybody else in the country from this invasion, however much some of them may have welcomed it. Non-rivalry means that once the service is provided to some people then, within limits, it costs no more to provide it to others.

For example, if one person admires the flowers in a park he does not prevent others from doing so, as he would do if he ate them instead. So his admiration for the flowers imposes no cost on anybody else. In addition, there are services, such as education or health, where there are good reasons for believing that positive externalities or asymmetrical information requires some public provision. And, even where the technical properties of the goods in question may not require public provision there are often cases for some sort of public intervention in the market – for example, to cater for the well-known weak links in the choice-to-welfare chain such as those discussed in Chapter 5.

This raises the problem of collective – or social – choice theory, to which major contributions have been made since Ken Arrow’s seminal work in 1951 set out to answer the question of whether ‘…it is formally possible to construct a procedure for passing from a set of known individual tastes to a pattern of social decision-making…it being assumed, on the way, that individuals’ choices are rational’.⁶

Arrow tackled the question of whether it was possible to find a ‘constitution’ – that is, a set of rules – by which society could rank various social states and that satisfied some apparently appealing ethical axioms, such as no dictatorship, or the weak version of the Pareto principle to the effect that, if everybody prefers situation A to B then A is better than B, or the requirement that rational social choice should respect the same transitivity axiom required of rational choice for an individual.⁷ What Arrow did was to prove that no possible rules could be found for reaching such an aggregate ranking in the light of individual rankings without violating one or other of his proposed axioms, all of which have great intuitive appeal. To prove such an ‘impossibility’ theorem is, of course, a very disconcerting and fundamental result. Although one may question some of the assumptions required to reach this result (such as no interpersonal comparisons of utilities) it was nevertheless an extraordinary achievement to prove that, given these assumptions, ‘…attempts to form social judgments by aggregating individual expressed preferences always lead to the possibility of paradoxes. Thus there cannot be a completely consistent meaning to collective rationality’.⁸

For example, suppose that there is some unpleasant localised air pollution from factories, such as smoke or sulphur dioxide. Consider three methods of dealing with it (apart from just ignoring it). One of them is simple regulation. The other two both recognise the scope for a market in ‘property rights’. But these could take the form either of rights to clean unpolluted air that could be given to the potential victims of the pollution, or rights to pollute the air that could be given to the owners of the factories that cause the pollution. There are thus three possibilities, namely:

X = use the market by giving property rights in clean air to the poor people who are the ones usually most affected by the pollution. They would then negotiate to sell some of their rights to clean air to the factory owners at a mutually acceptable price and quantity.

Y = use bureaucratic regulation.

Z = use the market by giving a certain amount of ‘air pollution rights’ either to the rich owners of the factories who might be willing to reduce their use of clean air in return for suitable compensation, or to the victims of the pollution who might be willing to sell some of their rights to clean air to the factory owners. Both sides could then negotiate a mutually acceptable deal.

The flavour of the argument can, perhaps, be illustrated in the following simplified example, in which the local authorities set up a small sub-committee of only three members, namely:

Mr LWF-M has left-wing tendencies but is in favour of free markets, up to a point, so wants to promote his distributional judgement that one ought to make income distribution more equal.

Mrs RWF-M, who has right-wing tendencies and also favours free markets, thinks that the pollution rights ought to be given to the factory owners.

Miss BB is a bossy bureaucrat so naturally prefers regulation since it gives her more power and responsibility.

Their preference orderings are likely to be as follows:

Suppose they vote on the choice between X and Y. Two of them rank X > Y. And if they also vote on the choice between Y and Z, two of them will prefer Y to Z. So it seems that a majority vote would show that X > Z. But if the committee then vote on the choice between X and Z they will find that two of them rank Z > X. So their ranking is X > Y > Z > X!!!! In other words, the democratic procedure of majority voting in this instance does not satisfy the transitivity axiom of rational choice that we met Chapter 4 in connection with individuals’ choices. This paradox, to which Arrow referred, is known as the ‘Condorcet paradox’ in honour of the Marquis de Condorcet who drew attention to it over 200 years ago.

A big industry on social choice theory has developed, of course, since Arrow’s pioneering contribution, some of which disputes the significance of Arrow’s work. One of the most fundamental critiques was made back in 1954 by Nobel Laureate James Buchanan.⁹ He argued that one should not expect a ‘social decision rule’ to satisfy the same criteria of rationality – notably transitivity – as for a single mind (such as an individual’s mind). He argued that ‘rationality or irrationality as an attribute of the social group implies the imputation to that group of an organic existence apart from that of its individual components.…’

In other words, one might expect that rationality would be the criterion of the preference ordering of a single rational mind. But we cannot go from that to expecting transitivity to be the characteristic of collective preference orderings. Thus Buchanan was not worried by the implications of the result that any decision rule can lead to apparently intransitive orderings, for he regarded this as a necessary feature of democratic decision-making. He argued that ‘Correctly speaking, majority decision must be viewed primarily as a device for breaking a stalemate and for allowing some collective action to be taken’.

That is all very well and may even set one’s mind at rest. But most of us would have liked to have believed that, if only one could find it, there must surely be some way in which a group could reach decisions that would satisfy our conception of rationality. But Arrow showed that – disconcerting though it may be – this is simply not the case. As he put it ‘…there are well-known fundamental dilemmas in any concept of social good…for reasons that I think are intrinsic to the logic of the subject. The root facts here are the incommensurability and incomplete communicability of human wants and values’.¹⁰

Thus we have lost our innocence and we now have to abandon our belief in some ‘rational’ collective choice. According to Partha Dasgupta, ‘It seems to me we simply have to live with Arrow’s theorem and do the best we can…Just as circles can’t be squared, ideal voting rules don’t exist, ideal markets are a pleasant myth, and ideal governments can’t be conjured up because governments are run by people’.¹¹ One cannot refuse to do anything at all on account of some known ambiguities in the notion of an aggregate, or collective, ‘social’ welfare. So what do we do? Presumably we eschew referenda on every important decision in the domain of economic policy. Instead we have to fall back on the best approach we have, namely old-fashioned welfare economics, while bearing in mind its limitations.

The welfare economics approach, as exemplified, for example, in a CBA, is far less ambitious. It is limited to trying to promote the state of affairs that would have been achieved in a free market in the absence of the usual market imperfections, such as monopolistic conditions, taxes, externalities and so on. In this way it might even be able to take some account of the intensity of people’s preferences, which is, of course, absent in the social choice approach. As explained in Chapter 5, we know that people’s preferences do not always accurately reflect what is good for their welfare. But let us assume, for present purposes, that we are dealing with situations in which this divergence between preference satisfaction and welfare is not very important. How then would conventional welfare economic analysis handle the aforementioned problem of dealing with local air pollution?

In the previous example, two out of the three members of the committee might have preferred solution Y to solution X, but the intensity of the preference of the third person for solution X may have been very great, whereas the intensity of preference of the two members of the committee for solution Y may have been very slight. In ordinary market transactions the intensity of people’s preferences is represented by the prices people are prepared to pay for the options open to them.

For example if a vote were to be taken in some office about whether smoking were to be permitted in common areas, there may well be a majority for smoking to be permitted. However, it is possible that the non-smokers would have been prepared to pay a much larger sum for smoking to be prohibited than the smokers would have been prepared to pay for it to have been permitted. In that case, there would have been scope for a mutually satisfactory deal between the smokers and the non-smokers in which some agreed compensation would have been paid by the non-smokers to the smokers that would have left both parties better off. A CBA would have indicated the scope for a Pareto-optimising move.

The welfare economics approach focuses our attention on the causes of market failure in any particular instance and possible ways of rectifying them. One criticism of the cost-benefit approach is that – as mentioned earlier – allowance for the intensity of preferences by taking account of how much they are prepared to pay for different things means giving greater weight to wealthier people’s preferences. But a good CBA could take some account of the extent to which costs and benefits are influenced by the distribution of incomes, as well as by other institutional features of society that had a major influence on relative prices.¹²

Thus in the aforementioned example, welfare economics would start by identifying whether there was an externality which established a prima facie case for some sort of public intervention. It would then try to establish what would be the optimal amount of the pollution permitted, taking account of the social costs of reducing it and the social damage done by the pollution. (This could make some allowance for the extent to which people’s preferences might be biased by incorrect information, etc.) One obvious method for reducing pollution to the optimal amount would be to impose a tax on the pollution equal to the marginal cost or marginal social damage at the point where these two magnitudes are equal.¹³ Another method would be to mimic a market system by giving pollution rights to either the victims of the pollution or the factory owners, along the lines indicated in Section 3, and then let them negotiate a price with each other.

The next step would be to go beyond the basic theory of Pareto optimality and to take account of distributional considerations. For example, the victims of the air pollution in the previous example might be mainly poor people, since on the whole, polluting factories tend to be located in poorer neighbourhoods. On the other hand, even if the factories were not located in particularly poor areas they might have employed mainly poor people whose livelihood would be threatened if the pollution charges made their factories uncompetitive. In either case the distributional judgements represented in the social welfare function of the decision makers would come into play and should influence the outcome.

But if the domain of welfare economics is extended to incorporate aspects of processes, such as respect for ‘rights’, into the assessment of the outcome of any policy, it raises a whole host of other value judgements. In the aforementioned example, judgements would have to include the intrinsic value of ‘rights’ to the clean air that was being polluted, since this would determine who is to pay and who is to receive the proceeds. For example, if the local residents had lived in the area before the factories had arrived, it might be argued that they enjoyed the property rights in the clean air and so should receive compensation from the factory owners. But if the factories had arrived first their owners might argue that there had been no pollution – in the sense that anybody was actually harmed – before the local residents came to live there. So, it could be argued, it was the arrival of the poor residents who are to blame for causing the pollution. In the words of the old song:

It’s the same the ‘ole world over

It’s the poor wot gets the blame

It’s the rich wot gets the pleasure

Ain’t it all a bloomin shame!

Thus, one way or another, although the welfare economics approach avoids the logical problems of arriving at a decision that can be said to represent some ‘collective’ rational preference ranking, it raises numerous difficult choices between the intrinsic value one attaches to aspects – consequences or processes – of the choices in question. But both welfare economics and collective decisions have crucial roles to play. It is widely accepted that the unfettered market mechanism is unlikely to provide a perfect mechanism for the promotion of society’s economic welfare. Hence, there may be room for some form of social intervention. The problem then is what form this should take starting from individuals’ preferences. There is no simple algorithm that can tell us how to reach the decisions. Meanwhile, decisions have to be made about policies and projects that will affect the whole of some society.

It was asserted in the Introduction to this book that, in most practical policy decisions, the basic value judgements implied in welfare economics may often not play as a significant role in decision-making as does the knowledge about the facts. This, it was argued, was because one is never starting from an optimal initial situation. Instead, one is always concerned with relatively small moves from whatever initial situation we happen to have started. But there are often cases where basic value judgements are important.

One well-known example is the memorandum written in 1991 by Larry Summers, who was the chief economist at the World Bank at the time. The memorandum that he wrote to some colleagues soon became famous.¹⁴ This is because he raised the question of whether the export of polluting industries to less-developed countries (LDCs) ought to be encouraged. He outlined the conventional economic case for doing so which was based on certain propositions, notably:

His famous memorandum has, inevitably, been heavily criticised and scrutinised, and it would be out of place here to rehearse again many of the points made.¹⁵ But it should be pointed out that in the final paragraph of his memorandum Summers did indicate that various objections could be made to this proposal, including objections on moral grounds. And he pointed out that similar objections could be made to many other World Bank proposals for trade liberalisation. He was, therefore, raising a question about consistency in Bank policy rather than making a specific proposal about the export of polluting activities. (It is not his fault that he sometimes raises questions that most people would prefer to leave unasked, presumably because they know that they would not like the answers.)

But it does serve as a particularly good example of five of the weaknesses in Pareto optimality that have been identified earlier (and of which Summers was, of course, perfectly aware, as the final paragraph of his memo indicated). They would be:

The way that the value judgements implicit in welfare economics enter into the decision-making process is illustrated in Fig.​ 3.​1 in Chapter 3. Some specific value judgements have already been mentioned – such as the intrinsic value of consumers’ freedom to make their own mistakes, or of equality of incomes, or of economic prosperity – and others will be discussed in more detail in later chapters.

It might already appear at this point that welfare economics cannot take us very far in making rational choices among different policy options. But this would be a mistake. Welfare economics and its application in cost-benefit analysis provide a valuable framework and organising principle for taking account of the effect of any economic policy. It clarifies criteria that help to evaluate the relative economic merits of alternative policies and projects. This conceptual framework enables one to identify the value judgements involved so that one can take explicit account of them, rather than to leave them to dwell unacknowledged in some murky background.

This does not mean that decisions become easy to make. For there are probably no clear-cut ethical theories that are in harmony with all our ethical intuitions and attitudes. Insofar as we hold plural values, some of which often conflict, one cannot expect to find an over-riding ethical system that reconciles the different preferences of different people.

On the other hand, as pointed out previously, most policy choices are about marginal changes in the allocation of resources, starting from a ‘second-best’ point. In practice, therefore, what is usually required will be factual information, and there will be little point in wringing one’s hands over the normative significance of the starting point. There are many situations where some numerical evaluation of alternative courses of action along cost-benefit lines that take account of distributional considerations will make a useful – if not decisive – contribution to rational decision-making that would be preferable to total reliance on other decision-making processes. An example of this in the UK context is given at the end of Chapter 19, and no doubt innumerable others could be found, not to mention the way some decisions are made in other parts of the world. In most situations the ethical limitations on CBA and its welfare economics underpinning it are small beer compared to the role of uncertain facts, administrative failures, pressure groups and empire-building officials, not to mention downright corruption, so that to jib at the former while accepting the latter is a clear case of swallowing a camel while straining at a gnat.

But, because some of the value judgements implicit in welfare economics are usually only in the background, the extent to which they qualify the whole structure of standard welfare economics is not always adequately appreciated. As a result, the valuable achievements of welfare economics can be put to the wrong use, as is sometimes the case, for example, with bad applications of CBA. The function of normative economics is to help improve the allocation of resources in the economy in the pursuit of its objectives. But this should not take the form of providing ready-made formulae that can be applied mechanically. As far as possible it should take account of the deviations between market prices and costs, on the one hand, and ‘social’ prices and costs, on the other. It must also take account of historical experience, of psychological motivations that are poorly understood, of political and social circumstances, and of ethical values that are usually only dimly recognised. Otherwise it becomes a sterile – and possibly harmful – activity.