6.1 Explanation and induction

At the end of Chapter 3 I suggested that an equilibrium model in which the preferences of individuals were forever changing has some considerable advantages and I therefore proposed an enquiry into the variability of preference orderings. This enquiry has now been completed. We may draw the following conclusions from it. The notion of consistent choice that finds expression in everyday terms such as a disposition, taste, habit, fashion or custom has the form of an ex ante fact. Provided the notion is not illusory, dispositions, tastes, and so on are therefore independent of particular situations and have a certain invariance, though one would expect them to change from time to time. They do not, however, constitute preference fields. The concept of a comprehensive preference field, or a preference ordering that assigns a relative position to every good or combination of goods available to an individual, is based (in the view taken here) on a confusion between ex post and ex ante facts. It fuses into one concept the consistency of choice in an ex ante context and the comprehensiveness of explanation we are used to in our genetic understanding of events. In this view it would be senseless to speak of the consistency or variability of a preference ordering, because there is nothing to vary. It is necessary only to distinguish between consistent choice and choice as an ex post fact.

In this final chapter I shall consider some of the implications of this point of view and thereby also attempt to tie up some of the loose ends left in previous chapters. It will, however, not be possible to deal with these implications in anything but broad outline. This chapter will thus be in the nature of a postscript indicating some possible avenues of further investigation.

A few remarks about what is often called descriptive economics may here serve as an introduction. I have in mind empirical studies of, say, a wage dispute in some industry, the marketing of some agricultural product, the development of a particular mining activity, the protection of an industry, and so on. From an epistemological point of view there is often something peculiarly ambiguous about some (and I do not mean all or most) of these studies. The writer’s intentions are not always clear. Is the writer giving a narrative account of some unique course of events which is satisfying and meaningful in itself, but which has no more significance for the future than the fact that the last hailstorm occurred on a Tuesday, or is the writer indicating the existence of a structure, of an institution, that others could use as a guide to action in the future? When the intention appears to be the latter, it is not always clear why that which is described should be regarded as a manifestation of something regular rather than of something quite fortuitous.

The question also presents itself when one considers the tables of statistics that often accompany such studies, or when a public speaker on an economic subject seems to feel incumbent to read out reams of figures. Perhaps it is supposed that the readers or listeners, on letting a series of figures pass through their mind, will somehow acquire a more substantial understanding of the past than if they were given only an account in terms of what various people tried to do, or in terms of their greed or generosity, their envy, anger, compassion, loyalty, and so on; an account that most people, one suspects, would find a good deal more intelligible. Perhaps the writer or speaker, unable to find much order or regularity in the figures, presents them to the readers or listeners in the hope that they can see some significance in them. Perhaps, again, statistics are provided simply so that they may be stored away on a bookshelf or in the back of the mind as something that may or may not come in useful one day.¹

In short, it is sometimes not at all clear whether the studies under discussion are meant to provide a genetic understanding of the subject matter, or to arrive at ex ante facts by induction. This apparent lack of methodological clarity is perhaps not very surprising since deterministic equilibrium models, based on the view that explanation and prediction are the inverse of each other, cannot be expected to give much theoretical guidance in the matter. It may be worthwhile to consider whether the empirical studies of descriptive economics would be better served by a theoretical framework which does allow choice to be analysed in an ex post context. While it will not be possible to do this here, I shall take a step in that direction by considering briefly the most obvious question to arise out of the conclusions stated in the first paragraph of this section, namely, what difference would it make to micro-economic theory if choice could appear in it either as ex ante or as ex post facts? In section 6.2 I shall consider a method of analysis that has considered choice in an ex post context and in section 6.3 I shall suggest that a generalized form of this method is not incompatible with what has at times been a major preoccupation of economic theory. Section 6.4 will deal with the question of consistent choice and economic institutions.

6.2 The method of Verstehen as a form of the genetic understanding

The genetic understanding, it will be recalled, depends on a set of beliefs. An explanation of the past, or an account of how something in the present here and now has come to be there, is intelligible to us if it conforms to our beliefs. When we interpret the behaviour of others, we most commonly do so in the belief that they act purposefully as we do and that they have emotions and sentiments similar to our own. It is thus an analogy to ourselves that mainly makes such interpretations meaningful to us and creates ex post facts in the form of surmised aims, expectations, sentiments, and so on. Furthermore, the aims, sentiments, and so on that we ascribe to others, being analogous to our own, provide us with an understanding that is intuitive, direct and intimate or, as is often put, with an understanding rather than with a mere explanation. Since most people, at least in a fairly sophisticated milieu, would ascribe aims and sentiments only to human beings (and possibly to some animals) this intuitive understanding is not available to us when we are dealing with purely physical processes.

This conclusion, namely, that human action but not nature may be understood intuitively or directly, has been used to draw a sharp epistemological distinction between the natural sciences on the one hand and history, the social and the ‘cultural’ sciences on the other. The former, it has been argued, have to rely on regularities found by induction whereas the latter can and do explain events as manifestations of the thought, intentions and sentiments of purposefully acting and feeling individuals, without necessarily any recourse to regularities found by induction. The historian or the social scientist can, so to say, re-enact a past event in his or her mind – making use what is often called the method of Verstehen (Verstehen, German for understanding, has in this context the connotation of empathy, of feeling and imagining oneself into the position of another).

A version of this viewpoint was put forward by Vico in the 18th century, but attracted little attention. The view gained acceptance, however, when it was put forward more than a century later as an analytical approach to the philosophy of history, that is, as an alternative to speculative philosophies of history (for example, that history follows some set pattern, or is the unfolding of providential wisdom, or of logical necessity as in Hegel and Marx). The term Verstehen is associated with Wilhelm Dilthey, but similar views were expressed by Benedetto Croce, whom I have mentioned before, the Oxford philosopher R.G. Collingwood and others. The distinction between generalizing and individualizing procedures, which I mentioned in connection with Menger, became in the hands of Rickert and other neo-Kantian philosophers very similar to Dilthey’s, and emphasized that Verstehen was concerned with concrete, unique situations and with the values that individuals attached to them. The method of Verstehen also influenced the Austrian school in economics through von Mises and led to Max Weber’s formulation of a ‘verstehende’ sociology.

The method of Verstehen merely gave a respectable academic status to a very familiar procedure; and this had the advantage of encouraging an investigation of its implications. Previously the idea had been applied as a matter of common sense. Adam Smith, for instance, made ‘changing places in fancy’ a central principle of his Theory of Moral Sentiments (which was based on the same lecture notes from which the Wealth of Nations evolved). ‘As we have no immediate experience of what other men feel’, he pointed out, ‘we can form no idea of the manner in which they are affected, but by conceiving what we ourselves should feel in the like situation’.² In the more formal development of the concept of Verstehen, the emphasis has tended to be on purpose and rationality. Smith integrated purpose fully with the emotive aspects of action and in this he seems merely to have been following common sense. One can certainly understand an action by reconstructing a cool calculation of what constituted adequate means to some immediate end, but in interpreting an action one usually goes beyond this to the sentiments attached to some more ultimate goal. Two observers may be agreed upon an immediate means-ends relation, yet they would understand the action quite differently if one sees it (say) as serving an idealistic struggle for liberty and the other as serving a ruthless grabbing for power. Understood broadly, the method of Verstehen simply consists of linking the visible effects of action with an imagined state of mind. It is of course practised by all and sundry in their daily dealings with other people and in their efforts to make sense of human affairs in general.³

When a study in descriptive economics involves the actions of, say, manufacturers, trade-unionists, government officials and politicians, it is thus natural that the investigator should changes places in fancy with the subjects being studied in order to gain an understanding in the sense of Verstehen. Indeed, the study may seem colourless and off-target if this is not done. Provided the investigator also follows other canons of good historiography, a study in descriptive economics is then really an historical monograph on recent economic events, in principle little different from the monographs produced by Schmoller’s historical school.

Let us suppose that we regard such historical or ideographic (see Chapter 4, section 4.5) monographs as a very worthwhile form of economic analysis. We know from the inconclusiveness of the Methodenstreit that this attitude does not commit us to questioning the value of an abstract and more theoretical economics. However, mindful of the bitterness of the Methodenstreit, and of the attacks by the historical schools on classical political economy, and of various institutionalists on more recent theory, we suspect that the criticisms of theory could not have been entirely vacuous. We would therefore like to have a theory that is as much as possible in harmony with ideographic procedures, or better still, one that can be a theoretical framework for descriptive studies of the actual course of events, as is surely the ideal. One way of achieving this, it seems to me, is to generalize, or abstract from, the method of historical enquiry discussed in this section in such a way that one is left with an abstract form of the genetic understanding.

In the process of abstraction the fullness or richness in which concrete situations can be understood by empathy (or by the method of Verstehen) must necessarily be lost. The great diversity of particular motives and sentiments must necessarily give way to a more abstract concept. The concept of utility, of use value, or of the German Nützlichkeit, as originally used by economists, seems to me to be such a concept. Its scope was not thought to encompass all motives and sentiments, though something more than the desire for wealth in political economy. Exactly what it did encompass, which amounts to the question of how economic action is to be distinguished from other action, has always been surrounded by a lot of haziness. However, the attempt to remove the last ‘vestigial traces of the utility concept’ from economics (Chapter 5, section 5.7) would also remove the last traces of an intuitive Verstehen from economics. One would then be committed to dealing only with consistent choice, or, if one seeks comprehensiveness of explanation, with preference fields.

6.3 The abstract genetic understanding in economics

There are many versions of a generalized or abstract genetic understanding, and some of these rely on an intuitive Verstehen. Hegel’s dialectical unfolding of the mind or spirit is usually held to be the most ambitious of these. Marx’s philosophy of history, which applies the dialectic to the productive relations among men, is not much less ambitious, even if more down to earth. Both, of course, are what are normally called speculative systems. On the purely analytical level the contributions have been more modest. I have already tried to show that Menger intended his exact economics to be an abstract scheme for analysing historical processes. Max Weber’s ideal types can be put to the same use. In short, there is hardly a lack of material for a study of the abstract genetic understanding. However, I shall not even attempt to draw up a list of the available material. I shall confine myself to indicating one possibility for adapting deterministic equilibrium models so that they may yield an abstract genetic understanding which, I shall argue, economists have at times intended their subject to yield.

For these purposes we require an abstract model that retains only the most general features, in non-specific form, of an explanation of how something found in the present has come to be there. In order to see what such a model may look like, it may be useful to look for some parallels. Let us take biological evolution by natural selection as an exemplar of a genetic explanation. We are interested only in its most general features and therefore the fact that it is an example from the natural rather than the social sciences will be seen, I think, not to matter.

Of the assumptions needed for explaining the existence of a particular species, the following are of interest in the context:

1. 1.There are life-forms which reproduce themselves and which are vulnerable to their environment, that is, those ill-adapted to their environment die.
2. 2.Offspring inherit genes which determine and limit their capacity for developing physical characteristics.
3. 3.Gene mutations occur which lead to inheritable variations.
4. 4.The physical, including the climatic, environment changes.

Now, the point that I want to bring out is that the theory of evolution is an explanatory model and not a deterministic model. It can explain how a species has evolved but it cannot predict what species will evolve in the future. The reason for this is plain. Items 3 and 4, mutations and environmental changes, are unpredictable. Within fairly broad limits anything could happen. On the other hand, in order to be an explanatory model the theory must contain some constraints on what could possibly have happened in the past. These constraints are listed as items 1 and 2. For example, characteristics acquired during the life of an organism could not have been transmitted to offspring. Equipped with such an explanatory model, the biologist can then undertake a genetic explanation which he or she can make as detailed as desired and for which the existence of fossil evidence will give certain fixed points. Since items 1 and 2 place certain constraints on the explanation, the biologists will be able to infer the changes that took place under items 3 and 4, and will be able to infer that at some stage in the past a certain inheritable variation must have appeared or that the environment must have changed in a certain way. In other words items 3 and 4 are mere explanatory ‘shells’, so to say, which acquire specific forms only during the course of the explanation. They are empty shells for forming ex post facts.

We may now use evolution by natural selection as an exemplar of a genetic explanation. On that basis we may conclude that an explanatory model in economics need not be a deterministic model and that it must consist of explanatory constraints (without which explanation is impossible) and explanatory ‘shells’ for forming ex post facts. We may also find some fixed points for our explanations in the form of documentary or other evidence from the past. The ideas of explanatory constraints and shells and of fixed points thus form the abstract model of a genetic explanation that we set out to find.

Let us lower the level of generality for a moment. The explanatory constraints may now take on the form of economic institutions, such as the institutions of a market economy or of (a particular stage of) feudalism. On the other hand, the belief that other people have aims, expectations and sentiments, just as we do, provides us with mere explanatory shells. We may believe that other people have aims, expectations, and so on, but we do not know what they are aiming at or what they expect, and so on. This can only be inferred during the course of a constrained explanation, that is, it is only in drawing up an intelligible account of the actions of others that their specific aims, expectations and sentiments are formulated as ex post facts. As is well-known, economic institutions are continually undergoing change and we are therefore at a disadvantage, as compared to the biologist, because our explanatory constraints are not as stable as the biologist’s. Nevertheless we have to be able to formulate some constraints in order to explain the genesis of an event. Even though we can make use of the method of Verstehen, which the biologist cannot, the variety of aims, expectations and sentiments that it would be possible to infer from some event (that is, as having led to the event) would be almost unlimited if no constraint were placed on the explanation.

It may now be seen that an explanatory model in which choice appears as an ex post fact differs from a deterministic model in so far as the direction of explanation is inverted. Instead of starting with an initial or earlier state and determining a final or later state, we have to start with a final or later state and infer an initial or earlier state. The difference is more than just a curiosity. It means that the ways value and price are conceived in deterministic and explanatory models are subtly but significantly different. Let me try to illustrate this first with respect to the concept of efficiency. It is sometimes said that a free market mechanism eliminates inefficient firms. Now, in the context of a deterministic model this statement is tantamount to a prediction that efficient firms will survive. In the context of an explanatory model it means that firms which survive may be called efficient. In the deterministic sense, efficiency is a property which is already inherent in the firms in the initial state, and it will determine their fate in the ensuing course of events. With an explanatory model we have to infer the initial state from the final state, in this case the fact that certain firms have survived. The efficient firms are those that best adapted themselves to the cost and demand conditions which happened to prevail in the preceding period. There can be no other criterion for efficiency than that a firm survived. It may seem that this makes a tautology of efficiency; it is like saying that a race was won by the winner. It is, however, not a tautology. The criterion for efficiency in a market economy is in fact survival. Firm A may do all things according to books on management, but if it goes bankrupt while firm B, which skimps along on very low overheads, survives, it would be said that firm B was the more efficient under the conditions that prevailed. The argument for the market economy is in fact not that efficient firms survive, but that the meaning and value judgement attached to the word efficiency may be applied to surviving firms. In a different economic system it may possibly be said that the concerns that survive are the ones that are favoured by the Party, and we would understand (Verstehen) the matter quite differently. The method of Verstehen retains an influence in genetic explanations but loses it in deterministic models.

There is a similar difference between the way value and price are understood in a deterministic model and in a genetic explanation. In a deterministic model, prices are the result of various behavioural equations or ex ante facts, including the preference fields of individuals. In a genetic explanation, choice is an ex post fact and a tautology may seem to arise again, namely, that the consumer prefers whatever he or she happens to choose. But it is again not a tautology. The original argument was that under the conditions of a market economy with its free choice and free competition, the thing that an individual happens to choose can be understood as something valued, and the term maximum utility (signifying something like the most desirable state of affairs attainable) could be applied to the common outcome of all choices. Verstehen had become an abstract genetic understanding.

Walras explained the rationale of his general equilibrium analysis in terms which made this point. At the end of Chapter 5, section 5.4 I quoted Walras as saying that he had not attempted to predict decisions, but only to express the effects of such decisions. In the same Lesson he also dealt with the ‘importance of a scientific formulation of pure economics’. He had treated free competition as an hypothesis and for this it was unimportant whether one actually ‘observed it in the real world’, as long as one could ‘form a conception of it’. ‘It was in this light that we studied the nature, causes and consequences of free competition. We now know that these consequences may be summed up as the attainment, within certain limits, of maximum utility. Hence free competition becomes a principle or a rule of practical significance.’ He acknowledged that this was the argument for laissez-faire that economists had used for a long time. He differed from them, he thought, in that he had actually proved the argument. ‘Nevertheless, I should like to ask: how could these economists prove that the results of free competition were beneficial and advantageous if they did not know just what these results were?’ He illustrated the advantages of rigorous analysis with the by-now-familiar argument that the idea of maximum utility could not be applied to communally consumed goods nor to goods produced by natural monopolies. It also could say nothing about what is called distributive justice.⁴

Walras, as I have pointed out, usually opted for the ex post context when he had to make a decision. The concept of a preference field did not arise with him. The utility analysis was new, it was more rigorous, but the idea was an old one. When Adam Smith spoke of ‘that general objection which may be made to all the different expedients of the mercantile system; the objection of forcing some part of the industry of the country into a channel less advantageous than that in which it would run of its own accord’, one can well imagine his adding: ‘Whatever that may be.’⁵ If he had lived later he might have said ‘Whatever it may be, it would represent maximum utility.’

6.4 Consistent choice and institutions

So far I have been mainly concerned with choice as an ex post fact, or with the explanatory shells in the alternative to a deterministic model. I now want to consider the explanatory constraints in that model. In the previous section I simply posited that these constraints take the form of economic institutions. An institutional set-up in which free competition takes place was then seen to affect the meaning of the ex post facts of choice.

How have institutions normally been conceived by economic theorists? Are they really accorded the status of constraints on action? The following quotation from Schumpeter seems to me to sum up the traditional attitude, or at least the attitude when the question is paid any attention at all.⁶

The schemata of economic theory derive the institutional frameworks within which they are supposed to function from economic history, which alone can tell us what sort of society it was, or is, to which the theoretical schemata are to apply. [Furthermore] … when we introduce the institution of private property or of free contracting or else a greater or smaller amount of government regulation, we are introducing social facts that are … a sort of generalized or typified or stylized economic history.

There can be no doubt that historical studies can provide one with an appreciation of what are normally called institutions that is richer, more satisfying and more meaningful than any other comprehension one may have of them. To say so is merely to acknowledge the primacy of a genetic understanding, or more particularly of the method of Verstehen. Nevertheless, I would maintain that a genetic understanding of institutions is inappropriate to the needs of economic theory.

The institutional framework for economic theory appears to be an amorphous and ill-specified bundle of things, and a repository for anything not easily handled by theory. However, let us consider briefly what generalizations one may make about institutions. Schumpeter went on to suggest that one could define human behaviour widely enough to include ‘the social institutions that are relevant to economic behaviour such as government, property inheritance, contract, and so on’. If one stretches the meaning of choice a little, it would seem that a great many of the forms of institutions are really cases of consistent choice, in the sense of ends and the choice of means, and constraints on choice which, in the limit of Hobson’s choice, coincide with consistent choice. So, customs and conventions, regulations with legal sanction, norms of good behaviour and so on, can qualify for the appellation of institution. Consistent choice and constraint on choice do not cover all cases; numbers of people and geographical proximity, for instance, seem to have a role in market forms. However, it was my contention in Chapter 2 that anything with the lasting qualities necessary to qualify for the term institution may end up described by an unspecified number of ex ante facts. Consistent choice is merely a special case.

As we have seen, however, when the question of what institutions actually exist does arise among theorists, the information is to be obtained not by inductive methods but from economic history. Understood historically or genetically, institutions are ex post facts. If the theories for which they are to be the framework are deterministic models, then choices in general appear in them as ex ante facts, in the guise of consistent preference fields. Now, this scheme seems to me to be the inverse of what it should be if economics is to have an empirical content. As I have tried to show in Chapters 4 and 5, there appears to be no good reason for supposing that choices in general could ever be formulated as ex ante facts, but that is how they appear in deterministic models.

On the other hand, there appears to be at least a chance that institutions could be formulated as ex ante facts, but we are to understand them as ex post facts. Can any really worthwhile attempt be made to isolate institutional data empirically with such a conceptual scheme? Can one, for instance, look for consistent choice when all choice is meant to be taken care of already by the preference field idea?

Let us consider the famous law of demand. Here, it seems, is a very rare case of a constant conjunction in economics, even if it is somewhat erratic by the standards of the physical sciences. The preference field idea, at least in Hicks’s view (Chapter 5, end of section 5.7), was developed in order to account for the law of demand, though other hypotheses would also have been possible. (It would therefore be naive to ask how one can deny the existence of preference fields and yet accept the law of demand, which is meant to arise out of certain common features of preference fields.) Let us suppose, for the sake of argument, that one were to put forward a rival life-style hypothesis (Chapter 5, end of section 5.6) to account for the same observed regularities. It would proceed on the principle suggested by Duesenberry, namely, that a ‘real understanding of the problem of consumer behaviour must begin with a full recognition of the social character of consumption patterns’.⁷ The details are here not important. The hypothesis may simply be based on a ‘demonstration effect’ or it may be a complex of the common ends, ultimate values, normative orientations, and so on, that arose out of Talcott Parsons’s analysis.⁸ As long as there were certain complementary and commonly held aims and more than one good could serve a particular aim, there would be enough complementary goods and substitutes to ‘explain’ the observed ‘qualitative’ relations between prices and quantities sold.

Anyone who put forward such a hypothesis would no doubt be faced by an army of objectors (if he or she was taken seriously enough). People do not all live up to the same life-style, incomes differ, most individuals have idiosyncrasies, and so on, in comparison, the preference field idea would be extremely neat and clean, especially if preference fields are allowed to vary from time to time. However, would it necessarily be a superior explanatory device? The life-style analysts would be talking about things that everyone knows something about and is therefore able to criticize. The preference field analysts are talking about a catch-all which no one has experienced, and it is in the nature of a catch-all that it can explain everything and nothing.