The element of time … is the centre of the chief difficulty of almost every economic problem.
Alfred Marshall (1961, p. vii)
In the two previous chapters we presented an overview of the methodological and substantive importance of subjectivism. In this chapter we shall extend and deepen dynamic subjectivism by applying it to a critical concept in economic theory: time. The nonsubjectivist method of neoclassical economics has not been able to handle time in a satisfactory manner. The core of orthodox theory is beset with temporal paradoxes and inconsistencies. As long ago as 1934, Kaldor saw that in equilibrium theory “the formation of prices must precede the process of exchange and not be the result of it” (1934, p. 127). Recently, Bausor more precisely diagnosed the problem when he concluded that in general equilibrium theory there is a “logical simultaneity between current decisions and current prices” (1981, p. 6; emphasis added). Thus, cause and effect, present and future are all laid out instantaneously. Even models that purport to take time seriously fail. Intertemporal general equilibrium theory has also effectively annulled time. As Hahn admitted, “The assumption that all intertemporal and all contingent markets exist has the effect of collapsing the future into the present” (1980, p. 132). Decisions are all made in a single primordial instant: the future is merely the unfolding of a tapestry that exists now.
In recent years, a number of articles have emphasized both the neglect of time by neoclassical economics and the importance of incorporating it into our models (e.g., Hicks, 1976). Nevertheless, much of this critical literature fails to pinpoint the exact source of the problem and, more importantly, to indicate the character of a possible solution. In this chapter we attempt to remedy that situation by, first, critically analyzing the orthodox treatment of time and, second, presenting an alternative, thoroughly subjectivist, conception of time.
The chapter is divided into five sections. The first is an analysis of Newtonian time and its implications for economic theory. The second is a contrasting study of the implications of real or subjectivist time. In the third section we clarify the relationship among real time, planning and action. There follows a development of the consequences of real time for economic processes and the characterization of uncertainty. Finally, we examine the interrelation between Newtonian and real time and specify their respective analytical roles.
Newtonian time
The Newtonian conception of time is spatialized; that is, its passage is represented or symbolized by “movements” along a line. Different dates are then portrayed as a succession of line segments (discrete time) or points (continuous time). In either case, time is fully analogized to space, and what is true of the latter becomes true of the former. Neoclassical theory has uncritically adopted this idea and used it in almost every systematic treatment of the temporal aspects of economics. In fact, almost all work on the economics of time is plagued by an excessive dependence on this analogy. Theorists treat a mere analogy as if it were the reality to which all models had to be faithful.
While the Newtonian view of time has doubtless been useful for many purposes, it nevertheless abstracts from some of the most important problems with which contemporary economics must deal. In order to support this conclusion, we shall begin by analyzing the central features of Newtonian time and the particular ways in which they manifest themselves.
Central features of Newtonian time
Newtonian time can be characterized in many ways, but for our purposes there are three aspects that are especially important: (1) homogeneity, (2) mathematical continuity and (3) causal inertness (Capek, 1961, p. 36–48).
Homogeneity In the geometry of lines each point is identical to all others, except for its position. In fact, position is really all that constitutes a point. Analogously, spatialized time is merely temporal position. It is an empty point or “container” that can (but need not) be filled with changes. Just as matter occupies empty space, changes occupy points or instants of time. Since a point is, by construction, empty, time must in principle be independent of its contents. In other words, Newtonian time can elapse without anything happening. The Newtonian conception thus transforms time into a static category.
One implication of this is that time can pass without agents learning. In the Arrow-Debreu model, for example, all decisions are made on some initial day. Although time passes and agents eventually find out which of the various contingent states of the world occur, they never learn anything that would cause them to want to change the decisions made on day one (Radner, 1970, pp. 480–4). Agents are completely constrained to the view of the world held at the beginning of time.
Mathematical continuity This is not continuity in the sense of an interrelation between successive instants but merely continuous divisibility. Just as a line can be divided and subdivided without end, the intervals of Newtonian time can be made arbitrarily small. In addition, no matter how finely we divide time and how close the resultant points are to each other, there is always some space between them. Each instant of time is isolated or, in principle, independent of the others, for points on a line can never touch.
Gunnar Myrdal (1939) saw the significance of mathematical continuity. If economic adjustments occur at points or durationless instants, all dynamic problems are evaded and hence left unsolved. In temporary equilibrium models, for example, adjustments occur at “the timeless points dividing the periods from each other” (p. 44). This, then, must neglect both the learning that accompanies processes and, more specifically, the order of each element in a process (since they are modeled all to occur at once). Furthermore, adjustments must have an infinite velocity and resources must be infinitely mobile for a process to take place at a mere instant. Here, of course, is the heart of the Newtonian paradox: if adjustment is instantaneous, why is change ever necessary? Things should be “right” from the beginning. Indeed, Myrdal put it more strongly, “The occurrence of change is contradictory to the idea of a timeless point” (1939, p. 44). This difficulty has been noted in the literature on price adjustments. Any stable disequilibrium adjustment process must occur instantaneously. Stability of the adjustment functions implies the predictability of the entire process. Consequently, if the process were to take time, pure profit opportunities would emerge. Entrepreneurial response to these opportunities would ensure that prices immediately adjusted (Gordon and Hynes, 1970, p. 372; White, 1978, p. 9).
Nevertheless, Newtonians have attempted to represent change “as a series of states, each of which is homogeneous … and consequently does not [itself] change” (Bergson, 1911, p. 163). Hence, any movement must emanate from outside the system; that is, it must be exogenous. A Newtonian system is merely a stringing together of static states and cannot endogenously generate change. Each period (or point) is thus isolated. Consequently, either we have the mere continuation of a period (no change) or we have change without the ability to show how it could be generated by the previous period. This is the fundamental problem with the comparative dynamics method that derives from Lindahl’s temporary equilibrium theory (Hicks, 1965, pp. 68–9). Since each period is in temporary equilibrium, we can examine change only by asking what the “new” equilibrium would be if an alternative set of data were exogenously imposed on the system.
Causal inertness That the mere elapse of time does not produce or cause anything is obviously just another way of looking at homogeneity: the independence of time from its contents. More importantly, however, if there are any changes, then they must be determined from the beginning. The initial state of the system must contain within it all that is necessary to produce “change”; time adds literally nothing.
The causal inertness of time is especially evident in the deterministic treatment of learning in neoclassical models. The passage of time does not produce change in the method of processing exogenous messages. In equilibrium an agent’s “theory” is “independent of the date t” (Hahn, 1973, p. 19). Therefore, given an exogenous message, we definitely know what the agent learns from it. More specifically, given his present information, we know what his expectations are. Neoclassical learning is merely a series of cross-sections over a surface or function depicting the existing state of knowledge.
The illusion of change
Change or the succession of events is not “genuine” in Newtonian time. Just as the points on a line coexist or are simultaneously given, so too must the future that the Newtonian model is intended to portray. In our previous example of neoclassical learning, the entire learning function exists “now” or, more exactly, is outside of time. Thus, there is no analytical difference between the various cross-sections being spread out over what we call successive instants and those same cross-sections being spread out over different regions at one instant in time (Shackle, 1969, pp. 16–17). It is perfectly possible for t1, …, tn to be compressed into a single conscious present.
The eradication of time
As we stated in the introductory section of this chapter, the explicit compression of change into a single instant is an important feature of neoclassical economics. No doubt this often results in greater mathematical tractability and the elimination of such problems as false trading (since equilibrium prices and exchanges are simultaneous), but, as we shall see, it also negates the very problem the economist ostensibly has set out to solve. This is the inevitable outcome of the causal inertness of Newtonian time and its consequent determinism. If all of the causes are present at t1, why must we wait until t2 for the results? In this section we shall examine three cases of the explicit elimination of time and the resultant undermining of the essential problem at hand.
Capital theory The classic Clark-Knight theory of capital is an important illustration of the neoclassical treatment of time. Throughout the 1930s and 1940s this theory was contrasted with the Austrian approach to capital. The latter placed heavy emphasis on variations in the time structure of production. Knight, however, fundamentally disagreed with that emphasis, for “under equilibrium conditions production and consumption are simultaneous” (1946, p. 387). There is no time structure in Knight’s theory because he fully adopted the Newtonian conception of time. The Clark-Knight view collapses the time structure of production into a cross-section of the economy’s “stages.” Once again, the future is spread out over instantaneous space.
Intertemporal general equilibrium The purpose of the Arrow-Debreu construction was to refine and extend the simple Walras-Pareto model of general equilibrium. The latter had been criticized for neglecting intertemporal relations and uncertainty. A simple and ingenious analytical device incorporating contingent futures markets was used to remedy these defects. But even though commodities are made available on different dates over an indefinite future, the analysis is still essentially static. There is no real difference between a given physical commodity at two dates and at two geographic locations. The further complication of adding a contingent state of the world (e.g., “if it is cold”) to the physical description, location and date changes little. All we have done is expand the number of commodities over which the individual maximizes.
Change, rather than mere uncertainty, is the true effect of time. Suppose, for example, that all of the “future” commodities were spread out in space at a single instant. Then we could view contingent markets as merely the uncertain prospect of what now is actually in a given container (perhaps ice cream, perhaps hot soup, etc.). The incompleteness of knowledge is not the sole element of a genuinely temporal model.
Tight prior equilibrium The most dramatic case of the abolition of time is the research program described by Reder (1982, esp. pp. 11–13) and ostensibly practiced by Chicago school economists. The core of this approach is the maintained hypothesis that “in the absence of sufficient evidence to the contrary, one may treat observed prices and quantities as good approximations to their long-run competitive equilibrium values” (p. 12). Despite a considerable amount of confusion as to which long-run model Reder is referring to, the main thrust of this approach is clear. Agents are assumed to “treat long-run equilibrium values of all endogenous variables as parameters” (p. 18). Thus, permanent rather than transitory income explains variations in consumption; security prices fully incorporate all relevant information at (almost) every instant in time; agents “rationally” expect the distribution of prices generated by the long-run stochastic structure of the model; and all decisions can be effectively executed, that is, all decisions are immediately pre-reconciled.
This position is the most thoroughgoing and consistent application of the Newtonian conception of time in contemporary economics. The “tight prior equilibrium” framework thus reduces all economic theory to a set of comparative static propositions (Reder, 1982, p. 12) that are assumed to hold instantaneously. The question of time as change does not arise, much less is it answered, in this world.
The measurement of time
The eradication of succession is intimately related to the way in which Newtonian time is measured. Strictly speaking, there is no such thing as the succession of Newtonian moments. Time is measured by the “counting of simultaneities” (Bergson, 1911, p. 338) and not of successive events. Suppose, for example, that T is a mobile that moves along its trajectory. Then the dating of a movement (“change”) of T is by reference to the simultaneous position of another mobile, T′. Thus, if T moves by one unit simultaneously with T′ moving by one-half a unit of distance, then T can be said to have moved (or “changed”) by one unit in one-half a unit period of time (as measured by the movement of the second mobile). Now, compress the psychological duration separating the trajectory positions, T1 and T2, etc., into one instant, and nothing has really happened. The change of T is dated by comparing simultaneous positions of T′. If mobiles T1, T2, T3 … all coexist, they can still correspond to the positions of other mobiles T1′, T2′, T3′ … which also coexist. Hence, even in the measurement of Newtonian time, succession, as opposed to juxtaposition, is inessential.
Hood (1948, p. 462) gives a simple example of the instantaneous measurement of time. Consider three continuous and differentiable functions: y = f(t), x = g(t) and y = F(x), where t is the trajectory of the movements in the mobiles X and Y. Time can be measured in terms of either X or Y. Suppose we choose to measure it by movements in X. Let’s assume that dy/dt > dx/dt. This then implies (since the relationship y = F(x) was established) that dy/dx > 1. Now if dy/dx = 2, we can conclude the Y changes by 1 unit in a half unit measure of time (which is analogous to our result in the previous paragraph). This conclusion, however, is really just a convention or stipulation. A system based on differential equations is actually concerned with present or instantaneous rates of change: neither the past nor the future is involved. Newtonian time is thus measured by simultaneous correspondences with a static moment.
As a final illustration, consider the adjustment model discussed by Hood (1948, p. 463) in a somewhat different connection. Let the rate of price change, dp/dt, be a function of excess demand:
where K = constant; D′, S′ are the slopes of the demand and supply curves respectively; P0 is the equilibrium price; and Pt is the current price. Time is here measured in terms of the absolute value of the changes in excess demand. The simultaneities counted then are the changes in the quantities supplied and demanded and the changes in price. Thus, the above equation determines the instantaneous (present) rates of change in prices for various present changes in excess demand. All of these changes literally refer to a single moment. The formal model effectively collapses diachronic and synchronic changes; that is, the temporal cross-section distinction is obliterated.
Real time
Along with other economists, Austrians have stressed the importance of real time. Often, however, the main features of this alternative concept have not been made precise. From our discussion of the Newtonian framework, it ought to be evident what real time is not. In this section we shall go beyond the previous discussion and present the central features and consequences of non-Newtonian time.
Following the work of the philosopher Henri Bergson (1910), we distinguish between spatialized time and the subjective experience of the passage of time. Bergson called the latter concept “la durée” or duration, but we shall generally use terms more congenial to economists: “real” or “subjective” time. Time, in this sense, is not the static subjectivist concept used in planning or reflection. Instead, it is a dynamically continuous flow of novel experiences. This flow is not in time, as would be the case from a Newtonian perspective; rather, it is or constitutes time. We cannot experience the passage of time except as a flow: something new must happen, or real time will cease to be.
Features of real time
There are three interrelated features of our alternative view of time: (1) dynamic, rather than mathematical, continuity; (2) heterogeneity; and (3) causal efficacy (Capek, 1971, pp. 90–1).
Dynamic continuity This form of continuity can most easily be grasped by an aural analogy. Hearing only one note of a melody, for example, is insufficient to capture the experience of music. This is because our perception involves memory of the just-elapsed phases (or notes) and anticipation of those yet to come. The actual experience is thus more than a mathematical instant; it is impossible to subdivide continuously a piece of music without fundamentally altering or negating the experience. The dynamic structure of real time consists, then, of two aspects: memory and expectation (Shackle, 1958, p. 16). On this view, the present is in principle linked with other periods through the perceptions of the individual. Memory and expectation are the structural components of real time accounting for its dynamic continuity. Although the physical or mathematical time that a given experience takes can be continuously subdivided, these durationless instants are not, from a subjectivist viewpoint, independent of or isolated from one another. Real time thus implies the very linkages from which Newtonian time abstracts.
All purportedly dynamic models postulate some kind of connection between successive periods. The existence of linkages per se, however, does not mean that Newtonian time has been abandoned. The exact character of the intertemporal relationship is very important. For example, Hicks (1965, p. 32) links periods in a dynamic analysis by the “stock of physical capital that is handed on from one single period to its successor” (emphasis added). Similarly, Hey (1981, p. 52) discusses a dynamic optimization model in which the present is linked to the past through the “inherited” value of wealth; but, given that value, the individual’s optimal strategy is independent of time. Thus in both the Hicks and Hey models, “the future always looks the same irrespective of from where it is viewed” (Hey, 1981, p. 52). The central difference between this form of intertemporal connection and the dynamic continuity of real time is that the latter necessarily involves a change in perspective. The future must look different depending on from where it is viewed. In a somewhat different context, Hahn (1952, p. 805) has made a similar point: “The experience of the first situation must always enter as a new parameter into the second situation.”
Heterogeneity If memory is the component of our experience that links the past to the present, it is also the factor responsible for the continuous differentiation of each successive moment. As time passes, the individual’s memory is continually enriched and thus the subjective standpoint from which he experiences the world undergoes change. As a consequence, each phase of real time is novel precisely because it is linked to previous periods by memory (Capek, 1971, p. 127). The dynamic continuity and heterogeneity of real time are thus not separable features, but are merely two aspects of the same phenomenon.
The heterogeneity of time sheds considerable light on the limitations of agents’ forecasting abilities. Suppose that an individual tries to predict an event. Paradoxically, even if it occurs “exactly” as predicted, it will not be experienced exactly as predicted. The simple reason is that before he made the forecast his standpoint was different. Afterwards, his memory incorporated the forecast and this changed his perspective (Schutz and Luckmann, 1973, pp. 240–1). This is a limiting case of the general phenomenon of a prediction interacting with and changing the object of prediction. The focus of this interaction, however, is on the subjective standpoint (state of knowledge) of the agents and not on a physical event. Thus, where the predicted event is dependent on a subjective state of affairs, such as the expectations of individuals, the event itself may be altered by predictions. If an oligopolist, for example, decides to change his price, this decision will in part depend on his expectation of the prices to be charged by competitive firms. But the prices that they will charge depend on their expectations of the price he will charge. Therefore, the competitors must determine his expectations about their expectations of his expectations. If he were to make a “wrong” prediction, then their predictions and hence actions would in fact be different. This phenomenon is well known in the literature of oligopoly models. In the next chapter we discuss in greater detail the particular form of uncertainty engendered by a similar example.
Causal efficacy This follows immediately from heterogeneity. The mere elapse of time, as distinct from physical change, is a source of novelty. As we saw in the previous section, the swelling of memory alone changes the perspective from which the world is seen. Hence time is causally potent and creative. This implies that all economic processes must involve the transmission and growth of knowledge. In this framework, for example, competition is no longer merely the name given to a certain equilibrium state. Instead, as we demonstrate later in this book, the process of competition is literally a discovery procedure. The growth of knowledge is the endogenous force that endlessly propels the system.
General consequences of real time
There are two important consequences of adopting our non-Newtonian concept of time. First, real time is irreversible. There can be no return to a previous period. Thus, “movements” along supply and demand curves do not mirror real temporal changes. Strictly speaking, as soon as we move away from a given point on such a curve there is no going back to it. Second, the passage of time involves “creative evolution”; that is, processes produce unpredictable change. A process is not a mere rearrangement of given factors, as it is portrayed in deterministic models of “change.” If change is real, it cannot be completely deterministic: there must be scope for surprise.
Real time, planning and action
The concept of time directly incorporated in plans is Newtonian. The planner can imagine units of time that are isolated, empty and as small as convenient. These homogeneous units can then be filled with specific activities. If completed plans were the only important aspects of economics, then Newtonian time would be quite sufficient as a tool of analysis. However, the process of planning must take place in real time. As we contemplate a course of action and project its consequences, we continually refine and refocus our tentative plans. Knowledge must be gained in the process of projecting. If this were not so, we could never go from a state of indecision to one of decision. Something must have changed during the process in order to account for the individual’s eventual ability to decide (Bergson, 1910, p. 171).
In a plan, the individual projects a completed act (e.g., the ditch as already dug) (Schutz, 1967, p. 58), and so, as we have seen, the plan itself is static and hence compatible with spatialized time. In the process of acting (e.g., digging), however, the individual experiences things. These experiences are novel if only because he approaches the world from subjective standpoints continually changed by the memory of what has been occurring. Moreover, exogenous factors are also affecting the system and upsetting the individual’s prior decision framework.
Neoclassical economics conflates the plan with the process of planning, the completed act with the process of acting. Furthermore, in its perfect-foresight version, the plan and completed act are conflated. This is equivalent to the assumption that all plans can be successfully executed and that there are no coordination problems. In such a world, there is no room for real time.
Novel experiences acquired in planning or acting are significant only to the extent that they engender plan revisions or alterations in the course of action. Thus these experiences must be connected with changes in the stock of knowledge. Knowledge, unlike pure experience, has applicability beyond the individual case that gave rise to it and thus can affect the future. In order to show that the growth of experience can be significant for our purposes, we must show its interrelation with the growth of knowledge. This can be demonstrated in alternative ways. From an inductivist perspective, agents can infer universal statements from statements of particular experiences. If all past increases in the money supply were ultimately associated with higher interest rates, then agents might infer the universality of this relationship. Therefore, as they “collect” more particular experiences they strengthen (or perhaps change) their theoretical understanding. Even from the now-dominant hypothetico-deductive perspective, it is conceded that experiences “suggest” certain hypotheses or interpretative frameworks. In addition, experiences can overturn or refute existing hypotheses. As a consequence, any change in (addition to) the stock of experience will give rise to new conjectural frameworks. Necessarily, then, it will also generate revised anticipations of the future because these are based on the current tentative framework. Hence growth in the stock of experience leads, via growth in the stock of knowledge, to alterations in both memory and expectations (the intertemporal links).
These alterations can never be deterministic. The connection between experience and new knowledge is not logically airtight. In the first instance, induction is a logically insufficient basis for any generalization. The fact that past expansions in the money supply were associated with higher interest rates does not justify a law covering all future occasions (Popper, 1964, pp. 27–30). Second, since many hypotheses are consistent with the same data (Friedman, 1953, p. 9), any given experience will not point uniquely to a particular generalization. Finally, even the elimination of hypotheses is not absolute. A particular observation may be incorrect, or further data may reinstate a previously refuted hypothesis (Popper, 1964, p. 50).
Real time is important because in the course of planning and acting the individual acquires new experiences. These new experiences then give rise, in a nondeterministic way, to new knowledge. On the basis of this new knowledge, the individual changes his future plans and actions. Thus the economic system is propelled by purely endogenous forces. The “natural” state of an economy in time is change and not rest, for “as soon as we permit time to elapse we must permit knowledge to change” (Lachmann, 1959, p. 73).
Economic processes and uncertainty
The most important implications of real time are for the modeling of adjustment processes and the characterization of uncertainty. No process can fit comfortably within a purely Newtonian construct. As we have seen, a world of instantaneous adjustment is really a world without process. If, on the other hand, adjustment takes time, then knowledge can grow and novel outcomes can emerge. Real time also affects the characterization of uncertainty. Risk analysis, whether objective or subjective, is essentially a weighting of possibilities already known. Genuine uncertainty, however, allows for the unpredictable growth of these possibilities and thus for “gaps” in agents’ probability distributions.
Reality of succession and indeterminacy of processes
Since the essence of real time includes novelty and causal efficacy, the future cannot be logically derived from the present because the former has not yet been created (Capek, 1971, pp. 106–11). Recognition of the creative aspect of time reinforces the position we adopted in Chapter 2 that decision-making is not deterministic. Conventional utility maximization, or what Kirzner calls “Robbinsian maximizing” (1973, p. 38), is a mere rearrangement or computation of things already known. To model processes in these terms is to render the analysis completely mechanical. In real time the situation must be different. Although the stages of a choice-theoretic process must be dynamically continuous (memory and expectation link the periods), these stages cannot ineluctably give rise to one another. The relationship between stages will be understandable even though each stage will not be completely predictable given its predecessor.
Examples In the Arrow-Debreu construction an agent knows with certainty his response to every possible contingency. Although the future states of the world are uncertain, nevertheless, given any such state, the agent views his response as certain. Pye (1978) developed a model embodying at least one aspect of real time: a decision-maker can predict only stochastically his response to hypothetical future situations. This is a reasonable feature for models in which the individual can be expected to learn during the course of time. Such a feature implies indeterminism in any sequence of decisions. Even if the consequences of the first decision and the exogenous state of nature at the moment of the second decision were perfectly predictable, we still could not predict exactly that second decision. Hence, the second decision does not follow inevitably from the first and the connection between stages in the process is stochastic. Winter (1971, pp. 254–6) depicts the stochastic evolution of an industry, and Radner (1975, pp. 198–205) discusses a stochastic process of cost reduction. And a general theoretical analysis of probabilistic processes is examined in substantial detail by Howard (1971).
In Schotter (1981, pp. 13–14), a possible framework is described for processes that are both nondeterministic and nonstochastic. There, a game is constructed for which two stable institutional arrangements can be construed as solutions. Because each arrangement is really a class of possible specific solutions, the theory delimits classes rather than exact imputations. This type of outcome is typical of many game-theoretical analyses. “The complete answer to any specific problem consists not in finding a solution, but in determining the set of all solutions” (von Neumann and Morgenstern, 1947, p. 44).
This view has been more recently echoed by Hayek (1967b) in his idea of pattern predictions. Theories of complex phenomena can be expected to predict only the overall pattern of outcomes (e.g., the kind or type of institution) rather than the exact outcome (e.g., the particular variant of the institutional type). If we model a social process as a series of such games, that process will be indeterminate. Even if we know the outcome of the first game and the nature of the second, we cannot say what the particular outcome of the second will be. We can only delimit a class of outcomes given the first outcome.
Genuine uncertainty
We shall postpone the main discussion of uncertainty to the next chapter. However, it is important to understand that real time implies a characterization of uncertainty that is fundamentally different from that prevalent in neoclassical economics.
Ordinary treatments of uncertainty depict it as a weighted arrangement of already-known possibilities. This is simply a given framework under static uncertainty (Langlois, 1982b). As we have seen, genuine uncertainty involves an open-ended set of possibilities. At the moment of choice, the individual will have conceived of a certain number or range of possibilities. Nevertheless, he is fully aware that in a world of change something might happen that he could not list beforehand. So he perceives his choice set as, in principle, unbounded in at least certain respects.
Genuine uncertainty is inherently ineradicable in the sense that additional knowledge may not enable the individual to overcome it (Dahrendorf, 1968, p. 238). Recall the forecasting example where the forecast altered the predicted event itself. In more general terms, since action takes place in real time, any activity designed to deal with uncertainty may merely transform that uncertainty. The source of uncertainty is thus endogenous in a world in real time.
Newtonian and real time: the interrelation
Until now, we have merely presented two concepts of time and examined the different implications of each. The impression thus given is of alternative and irreconcilable perspectives. Newtonian and real time, however, are each special cases in a more general temporal analysis. While it is true that these extreme cases are quite distinct, some types of human activity tend more in the one direction and others in the other direction (Bergson, 1911, p. 200). The deciding characteristic is the strength of the mnemic and exceptional link between time periods. As we shall see, creative decision-making (e.g., the setting up of frameworks of analysis) tends in the real-time direction. Ordinary maximizing behavior, on the other hand, tends in the Newtonian direction. Therefore, the actual content of the thought or decision-making process determines the appropriateness of the particular concept of time.
Individual entrepreneurship
What is commonly called “creative” activity or “insight” involves solving a problem or seeing a solution in a single leap (Bergson, 1920, p. 20). After this undivided insight is gained, an analyst may reconstruct the solution in a series of steps that others, at least in principle, are capable of following. The original leap can then be portrayed as the condensation of the reconstructed steps into a single, undivided, one. From the perspective of the less creative activity that follows these explicit steps in time, the more creative activity involves the compression of the past preliminary stages into the present final stage (the problem’s solution), and hence a very wide mnemic link.
This is precisely Schumpeter’s concept of entrepreneurship. For Schumpeter entrepreneurial success depends on “the capacity of seeing things in a way which afterwards proves to be true, even though it cannot be established at the moment” (1934, p. 85). A creative leap cannot, by definition, be conclusively “established” because it literally leaps over the requisite logical steps. Through this intuition the entrepreneur may be able to discover better technologies, new products and new resources. Similarly, Kirzner (1979a, pp. 158–92) has analyzed entrepreneurship at the individual level in which the central task is to formulate the “given” means-ends framework. This framework is logically prior to ordinary maximizing behavior.1 It is the result of a creative insight or relatively condensed activity.
This analysis has important implications for the agent’s perception of the rapidity of time change. In entrepreneurial or creative activity the preliminary stages in a problem solution are seen as part of the very recent past or, in the limit, as an aspect of the subjective present moment. In contrast, the less creative the activity under study, the more distended those stages become, or, equivalently, the narrower the mnemic link between them. Each stage becomes relatively more isolated. Reduction in the degree of creativity is thus associated with a relegation of the stages to the more remote past. Increasing the degree of creativity and the consequent widening of the mnemic link results in a subjective quickening of time. For any given interval of clock time, more is happening relative to the less creative state. Thus, the entrepreneur will perceive clock time as passing relatively more quickly (Capek, 1971, p. 200).
Maximizing behavior
Less creative activity takes each step at a time so that each phase more nearly resembles a Newtonian time period. The steps are relatively isolated from one another and thus the mnemic link is attenuated. This “Newtonization” of time is greater the more the agent’s actual solution to a problem is broken down into explicit steps. Solving a problem is made fully explicit when the steps that lead up to the solution are logically sufficient, i.e., when the framework yields determinate implications. In this form of activity there is no place for “creative leaps.” Pure Robbinsian maximizing is an example of an explicit step-by-step and determinate technique applied to decision-making. Thus, maximization analysis is compatible with the essential features of Newtonian time.
Since the mnemic link for explicit maximization has a relatively narrow span, the perceived present will encompass a smaller range of activity. This in turn means that, ceteris paribus, maximizers perceive clock intervals as passing relatively less quickly. In other words, there is a subjective lengthening of time (Capek, 1971, p. 200).
Rules of thumb
Routine activity, such as following rules of thumb, moves us still closer to the Newtonian end of our continuum. It is of the essence of a rule of thumb that the individual will perform certain actions without seeing how they fit into an overall picture. Thus, even more than with maximizing behavior, each step in what may appear to the observer a coherent process is relatively isolated and fragmented from the agent’s perspective. Thus the mnemic span is further narrowed and subjective time is further slowed.
None of the foregoing discussion of the degrees of real time makes any sense from the perspective that views all decision-making as if it were the explicit maximization of some objective function. To the extent that we reconstruct decisions in terms of that framework, we implicitly adopt the Newtonian conception of time. Only by paying attention to the actual content of the typical decision process can we determine the degree to which the activity is best explained in terms of real or Newtonian time.
