RATIONALITY AND PLANNING
Individuals Are Complex Decision-Makers: We have noted some of the problems
which the assumption of rational man encounters from the fact of the variability
of men and women, the fact that they are neither the same nor stay such as they are.
Beyond this critical point, however, there is still another of equal import. Individuals
make decisions on the basis of a complex pattern of personal needs, desires, skills,
knowledge and, crucially, ability to reconcile and make coherent their aims. Various
aspects of this criticism have already been made by a number of commentators on traditional
economics, each noting from a different perspective the excessive simplicity of 'rational
man'. These critics have included Kroeber-Riel (1971), Halldin (1974) and even Parsons
and Smelser:
Above all, in the field of economic behaviour, as well as in other fields, motivation must be treated as complex. It seems to us that the central fallacy in much of economic thought is to postulate some single motivational entity as an explanation of all economic behaviour. Usually this entity is some version of the "profit motive" regarded as an inborn propensity of human nature. But when complex phenomena are subsumed under the operation of one variable, the variable de-generates into a name for the class of phenomena in question. This is true of the "profit motive," the "instinct of acquisition," and the "rational pursuit of self interest," and others. (1956, p. 183)
This is plainly enough stated, from a position not so distant from neo-classical economics, yet the point has been missed or ignored. To be sure, fully working out its implications would require major revisions in economic practice and theory. The same point has also been made neatly (interestingly by someone trained in economics) with regard to social exchange theory:
The wit and good humour sonotable in Blau and Bailey is achieved by immediate abstraction: by the assumption that the principles on which the exchanges they consider are made, are the same ones which guide and inform a professor's wife in a supermarket. Similarly, the force and power of Barth comes from his selection of a single type of exchange as the motive force: there is such a firm sense of direction, the analysis does not fail to be mind-capturing. (Davis, 1973, p. 163)
Theory, in other words, has taken too easy an out. I shall not belabour the point either by repeating it or adding further citations of critics. Rather, I shall simply make note of five sources of complexity inevitable in processes of social decision-making and point out elements of their salience for any theory attempting to consider concerted action in a social context:
1. Individuals operate with a multiplicity of 'ends' which may fluctuate an importance, and which are not necessarily rationable. The multiplicity of ends is of course not contradictory to rationalist assumptions, but a further assumption that the ends are rationable is usually added. In practice, choice theorists tend to assume the existence of a single underlying end of which all others are but diverse reflections: profit or self-interest. Making ends rationable is the impact of Arrow's conditions of consistency and transitivity, and can be shown to inhere in his independence condition as well. There is some dispute over the ways in which preferences are to be seen as rationable, but not generally over the notion that they are:
In the theory of social choice as it has developed after Arrow, one can roughly distinguish two approaches to the problem of finding good social choice functions: firstly, the positionalist view, which supports the Borda function and similar voting functions, where the positions of alternatives in the preferences orders play a crucial role for social choice; secondly, the non-positionalist view, where the social ordering is determined mainly from binary comparisons between the alternatives. Typical non-positionalist voting functions are those based on majority decisions. (Gärdenfors, 1973, p. 1)
Despite the commonality of the assumption that there 'must' be an answer, the problem of the insufficiency of formal choice theory (that is, its arbitrariness) does emerge consistently from discussions over Arrow's independence condition. For example, take this exchange between Fishburn and Hansson:
In the 1968 United States Presidential election I preferred Abraham Lincoln to every one of the feasible candidates: should that preference have been taken into account by the social choice process? (Fishburn, 1970, p. 933)
It is perfectly all right that preferences concerning Lincoln are not considered in the 1968 election, but the independence condition requires a lot more: it requires the choice between Nixon and Humphrey to be independent of preferences between them and Wallace, the choice between Nixon and Wallace to be independent of Humphrey and the choice between Humphrey and Wallace to be independent of Nixon, all at the same time. And that is what makes the independence condition too strong. The problem is of course that there can be no formal way of singling out the truly feasible ones from the set of all conceivable alternatives. (Hansson, 1973, p. 38)
In other words, preferences interact on each other in addition to being ranked (or binarily compared); voters maximize ends other than the positive choice implied by the ideology of the voting situation. This is only partially handled by the condition of non-negative response, which was part of Pareto's attempt to formulate an equilibrium model, especially one tending towards a desired stasis.
Hicks (1946) raised the problem of integrability to deal with somewhat parallel problems in individual choice situations, but then dismissed it as having no economic importance, since he did not construct demand from any observed (i.e., personally coherent) individuals, but only from a aggregate of individual choices. Arrow, and other writers following him have attempted to reintroduce the problem, and, for example, have attempted to show that, contrary to Hicks' assertion, we can make meaningful comparisons of marginal utilities at different bundles (different sets of related choice-objects). Mayston suggests that
We are . . . able to compare the marginal utilities of any two goods, at any two bundles, on the same indifference curve. (1974, p. 21)
He accomplishes his 'proof', however, only by assuming, not by demonstrating the absolute comparability of goods. His supposedly 'meaningful' statement in effect reduces to the simply reflexive one: (x) marginal utility = (x) marginal utility, having never confronted the problem of measurement of dissimilar benefits. In effect, Mayston is constructing a Marshallian theory of demand over again, arguing that equal utility is demonstrated by equal demand; a construction of the individual on the basis of the market which is precisely what Hicks did not do. In other words, Mayston repeats the minimum theoretical tautology with the (untrue) assertion that it is empirically meaningful. As Coleman formulates this:
In the tautological theoretical framework itself, there exist formal relations between elements of the theory; in a theory with the concept of utility and the action principle of maximizing utility, an individual actor always and invariably takes that action which maximizes his utility subject to the constraints he faces. (1974, p. 36)
Unfortunately, although Coleman does note the fact of the actor's variable rationality, he does not take up the circumstance of irrationable multiple ends.
Game theorists from von Neuman and Morgenstern (1955) to Coleman (op. cit.) have approached this issue from the limits of logical action and the construction of essentially minimax strategies (although leaving open the security vs. high gain-potential choice, where applicable) in closed micro-analytic situations. In particular their situations have been those where 'winning' the game, or benefiting, is defined as a value on a single scale. Demand theorists (esp. 'revealed-preference' theorists) from the outset of utilitarianism have approached the issue as a reconstruction from market situations. (It must be realized that a preference poll is nothing but a verbal approximation of a market situation.) They establish the maximization of a clear preference ordering (one need not assume utility, although some writers have attempted to deduce utility from market behaviour) to which they impute some underlying character of commonality on which items may be compared (inasmuch as they construct indifference functions, which are usually seen as having to do with utility). These two approaches thus have in common the (often implicit but still) necessary asssumption of a single scale of evaluation. This scale of valuation is further assumed to be apparent to the rational actor. A multiplicity of ends thus cannot exist for either approach, since they recognize only 'benefit', of which there is merely more or less. The empirical unreality of this is hidden by a second order assumption which says that what individuals do in any given (and discrete) situation is not only directly, but completely representative of their desired end states, that forced choices, which are artifacts of games, questionnaires and markets (or rather of single decisions in markets) are in fact the form which predicates all social action. But in fact, individuals act to maximize varying proportions of conflicting ends, and sometimes even act to avoid decisions which are too equally counterposed.
2. Individuals make choices with regard to different time-spans of planning. For most people, with regard to most choices, this resolves into something of an averaging process: I maximize a certain amount of immediate preference over long-term potential preference, but not so much that all potential preference is lost to immediate pleasures. At the very least, what individuals want most they need not want all the time. In other words, I am neither all work nor all play. The problem for me is that I must decide when to work and when to play. The problem for the social analyst is that the decision I reach (and the proportionality it implies) is likely to be different from that reached by others. The difference is, further, likely to vary systematically in accord with a whole range of categorizations. This, for example, is the key point in Weber's discussion of the Protestant ethic (1904-5), a point which many 'neo-Weberians' have forgotten in their studies of social organizations. We shall consider the significance of time spans and planning in general in the next part of this chapter, here the point is only that each individual must choose to attempt to maximize his ends over varying durations, etc. His resolution of this complexity of potential temporal choices is a sociological variable.
3. Not only do people choose to plan for results over different spans of time, but their decisions affect one another over time. Decisions are not discrete: means interact with other means, ends with other ends. In the latter case, we may simply remark that the pleasure of travelling and the pleasures of staying at home are mutually exclusive at any one moment, but can to some extent be balanced over time. Thus the outcome of any given decision may be influences by the outcome of a previous decision, even if the two are in no way materially related (i.e., if you didn't spend all your money travelling the last time). In the case of means, most desired ends admit of a number of possible courses of action to achieve them, with a varying range of costs and probabilities. The choice of means for any given end may be different, however, if it is considered in isolation from all other means (actions) the individual is currently engaged in. Thus the most efficient means of driving to work may be to use a Mini, while the most efficient means of impressing your neighbours is to own a Jaguar (a Rolls-Royce being beyond the point of diminishing marginal utility). It does not follow that it is most efficient for you to own a Jaguar which you leave sitting in your driveway while you drive a Mini to work. Efficiency and choice patterns both change as one changes the scope of decisions which one considers to be part of a single series.
4. Individuals are also complex decision makers to analyze because they vary in their efficacy in relating means to ends. This is another way of raising the point that ends maybe defined at differing levels, as well as that their means are not implicit. As Durkheim noted,
Every means is from another point of view, itself an end; for in order to put it into operation, it must be willed quite as much as the end whose realization it prepares. (1895, p. 48)
I am willing to believe that in 1968 many people sought to achieve the end of improving American government by electing Richard Nixon, and thus defined that as end. I think their success in the latter end is unfortunate, precisely because it was a poor means to the former end: what score do we give these people on an efficiency based scale of rationality? In many, indeed probably in most significant instances of action, all we achieve are "those secondary and subordinate ends which we call means" (Durkheim, ibid.). We may achieve a little of the result we desired as a rationale for our choice, but most likely we also achieved a large number of unintended consequences (cf. Merton, 1936, and discussion below). In this context Bateson puts forward a cogent argument a to why ends cannot justify means. Essentially, he suggests that the unpredictability of end-results requires that means be subject to immediate moral scrutiny (1942). This issue may be one of the reasons why historical necessity is ideologically important to revolutionary socialism. The revolution can generally only be justified by its ends or results, and it is necessary to show that the ends will indeed be brought about, and are not subject to intervening events which would render the revolution unjustified.
Simon has suggested that formal organizational structure is simply the hierarchical organization of means and ends (1957). Each administrative level is seen as allocating responsibility for achieving certain ends to the level below it, which must choose the means. Thus in the hierarchy of administration each level is assigned ends from above, for which it provides means by allocating new ends to the level below it, down to the level of actual production in the act of labour. Central organizational issues are thus seen as the ability to analyze problems (i.e., turns an end into means) and communication (conveying the appropriate end accurately to subordinates). For this analysis unintended consequences become simply the results of inadequate analysis or communication and are seen as, in principle, completely avoidable. This of course also assumes perfect first-round communication so that correctives from below are not necessary. The question again arises as to whether everything may be subject to administration, or whether some things can (like Nirvana) only be obtained by indirection. The answer is the latter. Simon's theory offers little analysis of (a) motivation: why should subordinates (and super-ordinates) continue converting ends to means; (b) choice: it assumes rational know-ability of appropriate means to all levels and treats as non-problematic the initial specification of goals at the apex of the hierarchy; (c) change: it allows only for change of goal and internal adjustments; (d) social relations among members of the organization: it treats these only as more or less adequate communication.
Simon's theory does, to give it its due, note that the ideal of total rationality of organization can never be achieved, and that the need for action limits the search for 'best' alternatives, which then tends towards a search for the merely satisfactory (cf. also march and Simon, 1958, pp. 140-1, 169). This state of affairs is treated as a matter of the inadequacy of individuals. In the first place complex choices are seen as beyond the limits of individual rationality, so the organization limits the scope of decision of each member. In the second place, the administrator, being only human (i.e., a stunted computer with inadequate information), and under time constraints, chooses and option which will 'work;, even if it is not the best possible. Little attention is given to the definition of 'work', which is presumably an unintended function of the organizational structure. The criteria of satisfactory means tends to be treated as though they were material, that is, somehow inherent in the problem, and not inherent in the social organization. Further, no analysis is given of the determination of how far and how long one must search before assuming that he has obtained satisfactory information regarding the decision at hand. This last is an important variable, and there is no rational solution to its question. One may demarcate the bounds of consideration anywhere; the choice is rationally arbitrary, though socially determined.
Part of the problem with this approach comes from treating people only as either objects or agents of organization. This is largely a function of the fact that March and Simon look only at 'formal' organizations, which are conceived as rational in contradistinction to those which arise 'naturally'. The distinction, however, is untenable, for it is only through a process of reification that it turns a difference of degree into a category of kind. Interestingly, people are treated as agents to the extent to which the scope of their decisions makes their propensity to err significant. At the lowest level, production workers are seen only as abstracted labour, and curiously described in terms not dissimilar from Marx's. As far as workers at the lowest levels of the organization are concerned, the only relevant rationality is the matter of rational use.
The worker is seen as a factor of production, which has a cost and a return, and which has to be combined with the other factors of production in order to obtain some product or another. The problem of the rational use of the worker consists, therefore, in (1) determining the optimum rate of wages and of employment which will maximize the profits of the enterprise, allowing for the productivity of this factor, and (2) determining the factors that affect the productivity of the worker, and influencing these factors. (Godelier, 1966, p. 35)
'These factors' may include both psychological and sociological variables which are to be duly researched and used as means of manipulation of worker performance in a system which sees workers as rational only to the extent that they make choices which benefit the organization. A curious contradiction comes into utilitarian discussions here, for the organization tends to appear in this context as a reification of the actions of its owner and/or senior administrators. Social relations thus become the tool not the substance of the organization. When formal definitions of organizations are given, however, by other writers in the same general tradition, they tend to emphasize the opposite characteristics. As Olson has it:
One purpose that is nonetheless characteristic of most organizations, and surely of practically all organizations with an important economic aspect, is the furtherance of the interests of their members. That would seem obvious, at least from the economist's perspective. To be sure, some organizations may out of ignorance fail to further their members' interests, and others may be enticed into serving only the ends of the leadership. (1965, p. 5)
Blau and Scott suggest looking directly at the issue of who benefits as a key consideration in organizational analysis. They propose four categories: members, owners and managers, clients or 'public-in-contact', and public-at-large (which tends in modern society to be, interestingly, the state) (1963, pp. 42-5). Simon uses categories similar to the first three of these, but as we have shown, his analysis of operation, in particular his treatment of workers, assumes a model of a capitalist organization or bureaucracy with an exclusive locus of control (1957, pp. 16-17 and passim). Blau, following a definition of organizations as centrally being groups of people, makes this distinction explicit:
The collective efforts of men become formally organized either because all of them have some common interests or because a sub-group has furnished inducements to the rest to work in behalf of its interests. (1968, p. 298)
As Olsen has observed (among others), common interests by no means guarantee organization, since they do not necessarily exceed the benefits to be derived from individual-based action unless there is coercion present to ensure that all relevant actors participate in the organization. The coercion need not be the result of a single agency, however. One can readily see that the long-term benefits of property exchange through bridewealth and kin links could not exceed the benefits of individual action unless the whole population, or at least a closed population participated. Such coercion as there is comes from moral bonds (supported by various levels of concrete social relations).
In sum, then, the problem discussed in the last few pages is that economic theories of rational man tend to deal chiefly with instances in which means and ends are very closely and directly related. They either assume on this basis -- without demonstration -- that they are equally competant to deal with more distant relations, or else ignore the existence of such distant relations.
5. Perhaps most centrally it is a complex matter to assess any generality in individual decisions because they vary according to the identification of relevant and conceivable alternatives. This has been mentioned at various points earlier in the discussion particularly in the considerations of Arrow's independence condition and Simon's 'satisfaction' analysis of administrative action. I shall merely represent it schematically here. It is important to bear variations over time in mind, remembering that an option an individual defines as irrelevant at one point in time may be regarded as valid at another. Revolution, suicide, and marriage are three common examples. Either the individual or external social forces may act as the limiting agent; the individual may be aware or unaware of the proscription. If he is conscious of it, he may see it as the result of choice or of positive or negative coercion. If he is unconscious of the proscription, we may assess either coercion or ignorance, bearing in mind that 'ignorance' is socially conditioned. We treat at present only the proscription of options, not their prescription. The variable which is behind either is general for us: the limitation or proliferation of options.
Figure 4: The Limitation of Options
These factors of decisional complexity combine both to make individual decision processes refractory to external analysis and generalization, and to make them inadequate as mechanisms of intentional control over environments and situated events. We give this latter aspect more explicit attention in the next section.
Problems of Utilitarian Rationalism - Individuals are Variable Actors - Individuals are Complex Decision-Makers - Individual Actions Have Unintended Consequences - Time Spans and Plans - The Insufficiency of Rational Social Organization