Chaos theory versus Heisenberg's uncertainty: risk, uncertainty and economic theory.

Introduction

There has been a rising chorus of heterodox economists who advocate the irrelevance of the notion of equilibrium in a world characterized by evolutionary change (see Nelson, 1995). These economists - coming from diverse traditions such as Veblenian (Hodgson, 1993), Schumpeterian (Nelson & Winter, 1982), Austrian (Loasby, 1991; Witt, 1992), and Keynesian (Robinson, 1979) - aim at constructing a view of change which dispenses with the static idea of equilibrium. On the other hand, neoclassical economists have sharpened their tools to provide endogenous accounts of the change of social institutions (e.g., Schotter, 1981), cooperative behavior (Frank, 1987), and vengeance (Guth & Yaari in Witt, 1992).The theoretical propositions of the evolutionary approach are clearly incompatible with the propositions of the equilibrium approach. However, the incompatibilty does not necessarily entail that the two approaches are competing alternatives. They can be so only if they are discussing the same phenomenon.

The central motivation behind the paper is to show that while the equilibrium approach and the evolutionary paradigm are irreconcilable, they do not deal with the same aspect of the phenomenon and, hence, their exclusive opposition is unwarranted. While the notions of equilibrium and evolution are incompatible, they are not alternatives. The paper conjectures that economic change expresses a creative aspect which originates developmental processes as well as an optimizing aspect which generates equilibrium adjustments. Each aspect should not be absorbed by the other. Economists, heterodox as well as orthodox, try to offer an over-generalized theory which accounts for both. The heterodox economists who champion the evolutionary approach base their claim on the idea that the future is indetermined and, hence, action can never entail pre-given equilibrium states.

In this way, evolutionary economists generally confuse between two kinds of indeterminism, one arising from the knower's limited skill of computation and the other from the phenomenon's inherent uncertainty. The former kind of indeterminism, characterizing market equilibrium dynamics, is heuristically captured by chaos theory and, in economics, by Frank Knight's notion of risk. The latter kind of indeterminism, expressing innovativeness, is analogous to the laws of quantum mechanics and, in economics, Knight's notion of uncertainty. The confusion is evident when authors use chaos theory metaphors to buttress the critique of the equilibrium notion (e.g., Buchanan & Vanberg, 1991).

Section one commences with quantum mechanics and chaos theory. Section two relates them, at least metaphorically, to Knight's uncertainty/risk distinction. Section three shows that the two kinds of indeterminism shed light on the difference between developmental change and fluctuational change. Section four illustrates the two kinds of change by differentiating between developmental auto-feedbacks characterizing the evolution of technology/institutions and dynamical auto-feedbacks characterizing market prices.

1. Two Kinds of Indeterminism

There are many statistical laws in physics. But the two which have captured the attention of economists the most are the quantum mechanics and chaos theory. Although quantum mechanics has a longer tradition than chaos theory, economists seem to be more interested in the indeterminism a Ia chaos theory (Anderson et al., 1988). The reason lies probably in the difficulty of interpreting quantum theory.

The mainstream view of quantum mechanics, known as the Copenhagen school, rests on an epistemological interpretation. The uncertainly is attributed to the interaction between the subject (the experimenter) and the object (the particle). That is, the subject's tools or perspective changes the course of the objects behavior and, hence, one cannot specify objectively the state of the object. This interpretation is not followed here (see Khalil, 1989). It is sufficient to state that the epistemological road leads us to the much discussed new theory of knowledge, spearheaded by Thomas Kuhn, which questions logical positivism and Karl Popper's falsificationism (Khalil, 1987). While the issue of whether scientific theory can be value-free is fruitful, it is not useful for our purpose here. And even if one finds such a methodological debate pertinent to every subject, one still has to answer why only the behavior of particles is sensitive to the experimenter's tools. In other words, one has to investigate the nature of the quantum phenomenon because uncertainly does not arise from the experimenter's encounter with other, non-quantum kinds of phenomenon.

Aside from the different interpretations of the quantum phenomenon (passim Hiley & Peat, 1987), there are undisputed experimental facts. Such facts have been confirmed in a series of experiments which undermine the charge that quantum mechanics is an incomplete theory leveled by Albert Einstein and his collaborators.(1) In the attempt to refute the radical implication of quantum mechanics, A Einstein, B Podolsky, and N. Rosen (1935) proposed a mental experiment which they thought should demonstrate that physical reality is given independently of the tools of the observer. Let us take two electrons who form a pair revolving around the atom's nucleus. Such a couple are identical in terms of all physical characteristics (such as energy shell level and angular momentum) except for the quantum spin. The quantum spin value, which can be either + 1/2 or -1/2, is co-determined: Once the quantum spin value of one electron is given, the spin of the other can be deduced. According to the mental experiment, if an observer severs the two electrons from each other and measures the quantum spin of one electron, the observer should be able to know the quantum spin of the other. If this is the case, one can know the quantum spin value of the sister-electron without measurement and, hence, the reality of the quantum spin exists independently of the observer.

Aside from the logical consistency of the mental experiment (see Albert, 1992, ch. 24), Alain Aspect and his collaborators in Paris subjected it to a rigorous experiment in 1983 (Aspect et al., 1984; Grangier et al., 1986). Although they used photons rather than electrons, the difference should not affect the conclusion. Aspect and his co-researchers showed that when one photon is observed, the sister-photon, even if hundreds of miles away, is disturbed instantaneously and, hence, its value cannot exist independently of the observer as suggested by Einstein and his collaborators. The finding clearly defies the principle of "locality," which states that particles can only influence each other at speeds less-than-the-speed of light.

In specific, Aspect et al., showed that when a couple of photons, which were once related, pass through separate filters, they act, even if millions of miles apart, in a coordinated manner not allowed by the locality principle. Such a non-local interaction entails that correlated particles or photons act, even when they are physically separate, as nonseparable entities unaffected by distance (d'Espagnat, 1979, 1983). The mere act of finding the spin value of one electron disrupts in an uncontrollable way the spin value of the other - as long as they were recently united as a pair. As explained by David Albert (1992, 1994), physicists have failed to find any specific trait of the electron which makes its spin value disrupted by the measurement of the other. So, laws concerning the changes of spins by measurement fail to be deterministic. The theorist can put this kind of indeterminism, called here "uncertainty," in probabilistic terms only with reference to a pool of electrons. But he cannot formulate a deterministic law of the values of the two spins of a particular electron.

Such indeterminism is formulated in the well-known Heisenberg's Uncertainty Principle, after Werner Heisenberg (1958; see Horgan, 1992). The Uncertainty Principle states that it is impossible to specify simultaneously a particle's location and momentum. As soon as the experimenter finds out the particle's location, the experimenter's tool influences the particle's momentum in an unpredictable way, and vice versa. The Principle is not about the impossibility of specification of certain reality because of shortcomings of human tools, but rather because reality does not consist of physical entities, each occupying a certain location and characterized with certain traits. The entities acquire traits as a result of further interaction with other entities, one of which is the observer's tool.

Erwin Schrodinger summarizes quantum uncertainty with the famous metaphor of a cat placed in a box with a radioactive substance which can trigger at any moment the release of a lethal poison. In a Newtonian indeterministic world, one can state with certainty the chance (i.e., risk distribution) of whether, upon opening the box in two hours the cat can be found dead. That is, in two hours, the cat can be either alive or dead with a certain probability distribution. In a quantum indeterministic world, however, the cat can be in the potential state of being alive and dead. The uncertain state can be determined only through experience, one of which is the act of opening the box. This finding has lead to the subjectivist Copenhagen interpretation that the act of opening the box affect in an uncontrollable manner whether the cat will be found alive or dead.

Alternatively, Heisenberg's view entails that the state of the cat is a "potentiality." The idea entails that the particle (the cat) is a superimposed reality which does not occupy a concrete locality in space. David Bohm (1978, 1980) proposes a similar notion of potentiality. For him, matter moves constantly from the realm of abstract space to the concrete as the fuzziness of its state dissipates.(2) Quantum phenomena do not obey the notion of locality familiar to everyday human perception. If a quantum particle is taken to exist independently of the observer, it must be conceived as existing as a coherent potential which occupies a locality in an uncertain way.

Physicists are able to express the uncertainty of the potential states in probability distribution, called Schrodinger's wave function, only when they consider a large number of particles. The uncertainty is transformed into certitude expressed in risk distribution (i.e., chance) only when physicists give up the idea of predicting the behavior of a unique particle and focus instead on the representative particle. The representative particle is a fictional entity extracted from the averaging of the behavior of multitudes.

The resultant statistical description of the representative particle, however, radically differs from the statistical depiction of the states arising from the tossing of a coin. There is no need to resort to the "representative" coin in order to arrive at the probability distribution of its states. The probability distribution of the coin's states does not arise from its being an individual as is the case with the quantum particle. Rather, the probability distribution stems from the shortcomings of the observer. In principle, physicists can determine with definitiveness (100% assurance) whether a specific toss will result in a head or a tail. The result depends on the force and direction of the toss, air friction, the surface upon which the coin lands, and so on. The only reason, as P.S. de Laplace (1951) has stated, physicists cannot determine definitely the result is because of ignorance of all relevant information. Such ignorance may not be avoidable given that the determinants are too numerous and too intricate for any observer to collect and keep track of. If the observer is a "Laplacian devil" with perfect information-collection capability, he can determine in a non-probabilistic way the outcome of each coin toss. The two outcomes of the coin tossing are not superimposed possibilities which collapse in one way or the other in an uncontrollable way. Rather, each outcome is determined according to tractable causes. It is only because it is impractical or too costly for humans to attain the perfect information possessed by the Laplacian devil that the states of the coin are expressed statistically. In contrast, the Laplacian devil cannot determine even statistically the way the wave function of a specific particle will collapse. The devil can arrive at such a statistical estimation only by observing a large number of particles.

The core of the indeterminism of chaos theory, called interchangeably here "chance" or "risk," is essentially not different from the indeterminism of coin tossing.(3) Edward Lorenz (1963a, 1963b, 1964; Ruelle, 1992), who pioneered modern chaos .theory, is a meteorologist. The weather was a most suitable starting point since its fluctuation has defied predictability for centuries. Lorenz argued the the impossibility of non-probabilistic prediction arises from "sensitivity to initial conditions" which are minute enough to escape human detection. A slight change of an initial datum can engender a radically different outcome. The sensitivity to initial conditions arises when the parameters of recursive feedbacks are within a certain range. Lorenz half-jokingly stated that it is theoretically possible for the flap of a butterfly's wings in Brazil to set off, through auto-feedbacks, a tornado in Texas. If one knows the initial conditions in their greatest detail a la Laplacian devil, the exact behavior of a storm structure can be predicted in a non-probabilistic certitude. But humans are not Laplacian devils that can keep track of all initial and extraneous variables. Humans have to resort to a probabilistic form of prediction because of the astronomical cost of perfect information.

All in all, chaos indeterminism (risk) radically differs from quantum indeterminism (uncertainty). Risk or chance arises from imperfect information because of the shortcomings of the observer. In contrast, uncertainty stems from incomplete knowledge because of the nature of the object as a potential entity whose traits are not fully elucidated (Albert, 1992). While chaos chance probability emerges because of the enormity of the facts pertaining to a phenomenon, quantum uncertainty probability occurs because the particle is not a certain, localized fact to start with.

2. Knight's Uncertainty/Risk Distinction

There is an interesting parallel, at least heuristically, between the two kinds of indeterminism in physics and Frank Knight's (1972, pp. 219-220) well known distinction between uncertainty and risk. Knight draws a sharp distinction between the "real" doctrine of probability, on one hand, and "ignorance" theory of probability, on the other.

The real doctrine expresses "uncertainty" in the sense that future states are not given facts because actors art innovative or creative. To draw the implication of Knight's sense of uncertainty, the creative act makes the agent uncertain about the magnitude of "self-ability." As the agent tries to define the magnitude of self-ability, the ability undergoes developmental change. Such a self-defining process is the basis of what Herbert Simon (1976) names "procedural rationality" as opposed to the neoclassical "substantive rationality," or what Shaun Hargreaves Heap (1989) calls "expressive rationality" as opposed to the neoclassical "instrumental rationality" (see also Stewart, 1995).(4)

In contrast, the ignorance theory denotes "risk" in the sense that future states are facts in a world of certainty - such as the weather fluctuation or environmental disruption caused by human production, technological innovation, and product innovation. However, such facts are not perfectly available to humans. They express limited information which makes humans formulate only chance probability (risk) about their occurrence. Herbert Simon (1957) discusses the ignorance theory under the term "bounded rationality." Bounded rationality makes rule-following behavior more efficient on average than a case-by-case extensive investigation (Heiner, 1983).

To caution, the phenomenon of business confidence or optimistic/pessimistic expectation, pertinent to macroeconomic models of the business cycle, falls short of representing Knight's notion of uncertainty. Business confidence, aside from the particular term one chooses, expresses the agent's belief concerning the beliefs of other agents in general about future aggregate variables. Such confidence also differs from strategic expectation of the action of particular agents in oligopolistic markets or strategic games. Business confidence and macro-level expectation, unlike strategic expectation, explains positive feedbacks in macro fluctuation - what is commonly called "macro externality." Such feedbacks can be of the non-linear kind as Richard Day and G Piangiani (1991) show. Paul Davidson (1995) and other post Keynesians (e.g., Lawson, 1988/89) are correct in relating such macro externality to John Maynard Keynes' notion of probability. But they err when they present Keynes's notion of probability as radically different from the neoclassical notion of risk which informs market-clearing models. Keynes (1937) contends that the future cannot be presented as a "measurable risk" - a contention on which G.L.S. Shackle (1952) expands. The contention, however, might be the result of the limits of our faculties to collect and process all pertinent facts; it might not be the result of the nature of the fact of human creativity with which Knight deals. And insofar as one claims, as Davidson (1996) does, that Keynes's probability expresses "ontological" non-knowable reality, we are no longer studying macro fluctuations, but rather entrepreneurship (Khalil, 1997).

As defined here, Knight's uncertainty is about the uncertainty of self-ability because ability undergoes creative development as the agent tries to define it. In the Keynesian world, risk, which Keynesians also call "uncertainty," arises from the contingency of one's expectation on the expectation of others, which simply engenders an infinite regression. The risk distribution is not the outcome of the development of human capital, institutions, and technology. As Keynes (1973, p. 114) himself puts his notion of probability succinctly: "Knowing that our own individual judgement is worthless ... we endeavor to conform with the behavior of the majority or the average." So, Keynes' "uncertainty" is about chance where the outcome depends on others' judgement of the chance probability. It explains, at best, mob behavior, stock market bubbles, and cycles and, hence, ultimately can be subsumed under the notion of risk.(5)

Furthermore, uncertainty degenerates into certain risk when it is treated, as Knight (1971, p. 225) himself slips into, as a subjective/non-measurable estimate of a unique situation where there is no classifying instance. Subjective probability theory in the tradition of Leonard Savage (1954) has successfully shown that Knight's subjective risk estimate can be transformed into objective risk. The transformation is possible through Bayesian learning which classifies the unique situation into a broader class.(6) Therefore, one needs to keep the notion of uncertainty, insofar as different from risk, separate from the estimation of unique events such as the likelihood of the collapse of a newly designed building or the failure of a new telephone technology.

To keep the lexicon clear, Figure 1 opposes the word "uncertainty" to "certainty", and restricts the the word "certainty," to situations of "perfect certainty" and "imperfect certainty" (i.e., risk). There are, in turn, two kinds of imperfect certainty. First there is "incomplete information" which describes "subjective risk" assessment based on ignorance. Subjective risk assessment does not arise because of lack of information. Rather, it arises because the possible marginal gain of further search is lower than the rising search cost. Second, there is "complete information" which describes "objective risk" assessment based on a priori statistical distribution. Objective risk assessment, such as lotteries, does not arise because of search cost - which is zero. Rather, it arises because of the stochastic character of events which have a priori known likelihood of taking place.

3. Two Kinds of Change

The difference between risk (chance) and uncertainty is probably at the root of two kinds of change identified by Nicholas Georgescu-Roegen. As shown in Figure 2, Georgescu-Roegen (1971, p. 197) divides change into two broad categories, "reversible" and non-reversible motion. Reversible motion is exhibited in pendulums in zero-friction containers and billiard balls on zero-friction surfaces. In such movements, objects "follow the same course phase by phase in the reverse order" (ibid.). In contrast, non-reversible motion consists, in fact, of two radically different kinds, "irreversible" and "irrevocable."

The two kinds of indeterminism, chance and uncertainty, can act as metaphors in the effort to delineate between "irreversible" and "irrevocable" change (Khalil, 1990a, 1995, 1996a, 1996b). Irreversible change amounts to cyclical motion where objects "can return to any previously attained phase" but not by following the same course phase by phase in the reverse order (Georgescu-Roegen, 1971, p. 197). Georgescu-Roegen illustrates irreversibility with the flow of vehicles in a traffic circle. As each agent chooses a path which complies with optimization rationality, traffic moves in an irreversible flow. But each object can return to its previous position after making a full circle. Also driving on the right can be switched to driving on the left, and so forth - if one disregards the transaction cost. In addition, the motion of each particle along the shortest distance makes heat transfer from hot to cold areas and the equalization of pressure irreversible. But each particle can theoretically return to its previous position and, hence, heat can move from cold to hot areas in isolated systems. The probability of such a state, however, is very low (Khalil, 1990b). Such an unlikelihood does not mean it is theoretically impossible. It rather expresses the unlikeliness of chance events of sequencing in such order. If the interaction of factors involved is simple, it is possible for storms to reverse direction. With such a simple picture, the cost of perfect information is manageable and, hence, there would be no need for chaos theory.

In contrast, irrevocable change involve "processes that cannot pass through a given state more than once" because of the nature of the phenomenon (Georgescu-Roegen, 1971, p. 197). Such change includes the aging process which characterizes all cells and multicellular organisms.(7) Although some trees shed their leaves seasonally, and most organisms experience the sleep pattern, the subjects also undergo a developmental process which is non-cyclical. Such a unidirectional development also characterizes the successive progression of personality, knowledge, and technology change. Even when knowledge becomes extinct - as when a tribe or an old man dies without passing the knowledge on to other individuals - its "reinvention" by others does not amount to revocability. We have at hand here two different, separate developments. Furthermore, the fact that particular techniques are eclipsed only to reappear later because of change of fashion or input prices (Stiglitz, 1973) does not mean revocability. Once a technique is acquired and stored in one's inventory of "latent capital" - i.e., the tacit knowledge of the developing agent (Polanyi, 1958, 1966) - one does not need to undergo the same conceptual leap to find it. The extent of latent capital attests to one's flexibility in the face of changing economic circumstances. So, the fact that the agent may die without passing the knowledge or shelve the knowledge into latent capital does not affect the irrevocable character of development of knowledge.

Irrevocability entails the unfolding, in the sense of detailing, of one's knowledge as one strives towards a purpose. Once the agent reaches a goal, a second attempt to reach the same goal poses a lesser challenge. The second attempt cannot repeat the same steps because subject undergoes a learning process where what was an uncertain potential becomes a more certain actuality. Such a process of self-development is non-reversible because of the nature of the process, not because of the likelihood of sequencing of chance events a la chaos theory. Development entails opportunities which cannot be repeated since the person can only grow older. The sunk cost expended in light of one's assessment of self-capacity cannot be recovered.

Thus, the process of development involves irrevocable rather than irreversible change. While irreversible change cannot be reversed because of statistical improbability, irrevocable change cannot be reversed because of theoretical impossibility. A hurricane or a traffic flow represents an irreversible change only because it cannot be reversed for practical reasons, while it can be reversed theoretically once these practical hurdles are eliminated. In contrast, the development of technology is an irrevocable change because it cannot be reversed even theoretically. Once a person attains a particular know-how, he cannot repeat the same steps in attaining it again. The person is not the same after the first experience.(8)

The two kinds of change highlight the difference between the analytical and historical modes of explanation.(9) In the analytical mode, one's capacity can be assumed statistically and outcomes can be predicted according to the differently probable states of the world. A Laplacian devil can know the equilibrium point according to a perfect, non-statistical certitude. Traffic order or market equilibrium emerges because agents employ optimization rationality, where ends are assumed as given. The previous state of traffic flow or equilibrium prices can theoretically be returned to. The unlikeliness of such a reversal stems from a statistical likelihood that events would be repeated in an exactly reversed order. Even when the variables are minimal, as expressed in the logistic functions of chaos theory, it is extremely unlikely that events would flow in a particular sequence.

In the historical mode, however, the estimation of one's capacity to undergo development can only be guessed. One can never know the accuracy of the guess because the act of finding out can only occur through experience which changes the subject irrevocably. So, the future course of development is engulfed with uncertainty since it is unidirectional at the theoretical level. The unidirectionality of development can be expressed in the notion that the future is not like a position in space to which one can return.

4. Two Kinds of Nonergodic Feedbacks

The two kinds of change can help us understand the difference between two kinds of nonergodic feedbacks. To state that the change of a system or an organization is path-dependent is insufficient to specify the change at hand. The two kinds of feedbacks are usually conflated in evolutionary economics irrespective of the theoretical orientation (e.g., passim Anderson et al., 1988; Witt, 1992; Day & Chen, 1993; Boulding, 1981; Radzicki, 1990; Rosser, 1991; Hodgson, 1993).

The first kind of feedbacks, called "nonergodic development," characterizes technological/institutional evolutional trajectories. It is exemplified in the works of Brian Arthur (1989), Douglass North (1990), Paul David (1985), Richard Nelson and Sidney Winter (1982), and Paul Romer (1990; see passim Dosi et al., 1988). Such works show how an initial investment in a particular, and maybe less efficient, technology or institution becomes entrenched through positive feedbacks. The positive feedbacks are premised on learning-by-doing. These works came to epitomize the evolutionary approach.

As Ulrich Witt (1992) characterizes it, the evolutionary approach can model technological change as an endogenous variable, stemming from learning-by-doing, which is in contrast to its treatment in traditional growth neoclassical models as an exogenous variable. The classic starting point of evolutionary economics is Armen Alchian's (1950) seminal paper. Aside from its flirtation with Darwinian non-intentional natural selection (which cannot ultimately endogenize technological change (Khalil, 1993)), the paper invokes the Lamarckian process of the process of the intentional adoption of routines in light of their ex-post survival benefit. According to Alchian, the agent does not seek the optimum routine but rather, in light of uncertainty, the one which affords viability. The successful routine becomes entrenched as it is further developed through learning.

The second kind of feedbacks, named "nonergodic dynamics," typifies chaotic market cumulative causation (Kurdas, 1988). It is best elaborated by Allyn Young, Nicholas Kaldor, and lately Paul Krugman (1991). Krugman cites the work of Arthur on technological development, rather than appeal to business cycle theory, to lend support to his auto-feedbacks story about geographical bifurcation. Such a citation exemplifies the confusion of the two kinds of nonergodic feedbacks. One reason behind the confusion is that Krugman's model centers on the notion of space (geography) rather than time as is the case in business cycle theory. Spatial bifurcation (core/periphery) deceptively appears closer to Arthur-like technological development, than to temporal bifurcation (boom/bust) typifying the business cycle, because the geographical core also intensifies the pace of technological development.

To dissipate the fog, one has to keep in mind the deciding criterion. Namely, nonergodic development does not involve cycling (i.e., bifurcation) around a center or a trend as is the case with nonergodic dynamics. Such a cycling in nonergodic dynamics takes place temporally (the business cycle - which in a boom also intensifies the pace of technological development) and spatially (uneven growth pattern - which at the core also intensifies the pace of technological development). Such a cycling or bifurcation does not take place in Arthur's story. In Arthur's model, the development of one technology does not cause another technology to deteriorate. It only entails that the other remains absolutely frozen or undeveloped. In Krugman's model, in comparison, the immigration of a firm to a region entails a loss for another region. That is, the increase of activity in one region causes, ceteris paribus, activity of other regions to decline. The same applies with regard to the business cycle: The overheating of the upward phase lays the ground for a longer recession.

It is beyond the scope of the present paper to review the diverse theories of the business cycle. However, the theories which rely on endogenous factors hypothesize that the failure of the fulfillment of one of the conditions of equilibrium theory is at the origin of the business cycle. General equilibrium arises when we have complete markets with products about which agents have perfect information with regard to quality. As well-known (e.g., Hahn, 1984), actual markets hardly approximate such an idealized set up. Actual markets are riddled with agents who have imperfect information about the products. Imperfect information leads these optimizing agents to rely on the expectation of others who, in turn, follow what they think others will think. While such a behavior is rational at the individual level, it occasions positive dynamics at the macro level which prevents the actual clearing of markets. However, insofar as one is concerned with the business cycle, one must, first, start with the idealized set up and, second, investigate how the violation of one or more of the assumptions of the set up entails cumulative causation which engenders the business cycle.

According to one story of cumulative causation, optimistic prediction creates its own conditions of fulfillment, while pessimistic expectation engenders a slowdown of economic activity. Positive expectation, or what social psychologists (e.g., Bandura, 1982) call "high self-efficacy," is the basis of nonergodic dynamics. Nonergodic dynamics is about disequilibrating mechanisms depicting booms/busts through time or core/periphery bifurcation through space. Such dynamics can be analyzed only in light of the assumptions of general equilibrium theory and how, for instance, the violation of the assumption of perfect knowledge leads to far-from-equilibrium states.

The treatment of nonergodic dynamics as nonergodic development by neoclassical and heterodox economists alike (e.g., Boulding, 1981; passim Anderson et al., 1988) amounts to the confusion of disequilibrium with evolutionary development. For one thing, while disequilibrium is premised on equilibrium, it is not the case with development. For instance, Arthur's explanation of the development of technology is not mainly about the self-feeding character of expectations in the macroexternality sense. It rather focuses on the unfolding of an organization of production in light of its underpinning regime of technology and corresponding institutions. It is about the secular evolution of production which is as predictable as one can predict his own ability in the face of new challenges. The secular trend exhibits - after allowing for "underdogs" or heterodox technological practices lingering at the margin - a single unfolding of the potentiality of the underpinning scheme. The unfolding might be short- or long-lived depending on the viability of the uncertain capacity. The single unfolding entails that the betterment of one group or region does not cause the impoverishment of another group or region. Nonergodic development involves a positive-sum game (in the case of economic prosperity) or a negative-sum game (in the case of economic decline). Nonergodic development does not involve a zero-sum game.

In contrast, nonergodic dynamics entails a zero-sum game since the fluctuations around the secular trend are supposed to offset each other. Fluctuations usually exhibit a dual track dynamics or bifurcation. The dual track means, that after the abstraction from fundamental development, the rise of one region or income class is more-or-less undertaken by impoverishing another region or income class, or the booming phase of the business cycle is normally offset by the dip of another phase. So, the phases do not relate to each other asymmetrically as is the case with primary paradigm development and underdog alternatives. The occurrence entails that the bifurcation feedback is by definition double-track, symmetrical dynamics. It basically amounts to a zero-sum game.(10)

It is important to distinguish the two kinds of auto-feedbacks. The modeling of development as dynamical change implies that economic prosperity is guaranteed once positive expectations are adopted. Such a view depicts the course of development as the outcome of self-fulfilling prophecy: Agents can attain any goal and achieve any level of developmental growth as long as they adopt optimistic, positive outlook. Such a conception reduces assessment of self-ability to wishfull thinking. The subscription to Keynesian or to a more standard theory of expectations is no alternative to a theory of nonergodic development.

5. Conclusion

The paper argued for a fundamental difference between two kinds of change, developmental processes and dynamical fluctuations. The former is expressed in the evolution of division of labor, technology, and institutions, while the latter is epitomized in market oscillations arising form auto-feeding expectations. Both kinds of change can be indetermined. So, the thesis put forward by heterodox economists - such as Austrian-subjectivists (Shackle, 1952, 1970; Loasby, 1991; Lachmann, 1977) - that the future is unpredictable is, at best, under-specified. The indeterminism of market auto-feedback arises from the limited computational ability of the scientist - similar to why the scientist uses chaos theory to model weather patterns. In contrast, the indeterminism of creative development stems from the inherent uncertainty concerning one's capacity which can be known only through experience - analogous to the principle of uncertainty in quantum mechanics. Since we have two kinds of change, the heterodox view of the economy as an evolving organism cannot be a substitute to the orthodox view of the economy as an equilibrating structure. While the two views are incompatible, they are not alternatives. Each view simply deals with a different facet of the phenomenon.

To caution, the difference between development and dynamics should not suggest that the two kinds of change are of symmetrical importance. Without being able to give a proper defense here, development is more fundamental than dynamics. Dynamics can take place only after one's capacity and goals have been specified and pursued. The agent's action engenders dynamics only if he has a goal to pursue. The economy as a dynamical structure cannot exist if there is no economy whose organization of division of labor and knowledge is developing.

The difference between the theories of the economy-as-dynamics and economy-as-development is metaphorically illustrated by the difference, respectively, between the laws of thermodynamic structures and laws of atomic organization. In the past three decades, theoretical physicists have been earnestly trying to formulate a unified field theory. The diverse formulations, such as super-strings theories, try to show that different forces of nature are ultimately derived from a single force (passim Davies, 1989). The single unified force is presumed to be evident at high energy states such as the one which existed at the birth of the universe according to the Big Bang theory. There has been undisputed success in unifying electrical and magnetic forces at the atom level and relating them to the weak and strong forces at the nucleus level. (But so far there has been much less success in integrating the gravitational field force into the framework).

What is interesting about the search for the single unified force is that it does not include the laws of thermodynamics. The laws are simply not regarded as fundamental ones. The first law of thermodynamics (law of conservation of mass/energy) is valid only within certain domains of energy/matter (Feynman, 1965). The second law (law of entropy) describes the equilibrium, end-state of a structure which arises from the chaotic interaction of the molecules which make up the structure. The second law does not apply to a single molecule - similar to how the laws of market equilibrium do not apply to Robinson Crusoe. The third law depicts the greater hardship of lowering temperature as one gets closer to absolute zero. The law is also a property of the collective behavior of molecules.

Thus, the laws of thermodynamics are not fundamental laws of nature. They rather describe collective behavior once nature and its fundamental laws are in operation. Likewise, the laws concerning the equilibrating dynamics of markets across space and time are the outcome of the collective activities of agents. They are not fundamental laws about the economic organization of labor and technology and their development. It is true that market dynamics may pose serious problems when they ensue a disequilibrium which goes uncorrected for one reason or another. Nonetheless, organizational development has a theoretical priority over structural dynamics for a simple reason. Structural dynamics are ultimately regulated by the development process, and not vice versa. To wit, economic theorists should dedicate less attention to structural dynamics and equilibrium/disequilibrium analysis for a practical reason as well. Namely, the long term prosperity of a nation depends on economic development and its secular trend rather than on dynamics. Structural dynamics, at best, influences short-term fluctuations of well-being.

All in all, the evolutionary process does not hinge on fluctuations occasioned by non-clearing markets. After all the dust settles, equilibrium relative prices will influence, ceteris paribus, the technological developments of Robinson Crusoe, market economy, and socialist economy in the same manner. The evolution of technology and division of labor was on the march before the rise of the modern market institution. In this light, the challenge is to steer the main pilot of economic theory away from the study of equilibrium and disequilibrium dynamics, and to direct it more toward the investigation of developmental change.

Notes

1. While it is widely known that Einstein's theory of relativity challenges Newtonian physics, it is less known that Einstein was aloof towards quantum mechanics, which is at the origin of his well-known statement, "God does not play with dice." Relativity theory views gravitational forces as waves, which is contrary to the classical/Newtonian view of gravitational force as action-at-a-distance between self-contained solids. However, it is quantum mechanics which carries further the break with Newtonian perspective. Beside postulating the matter character of light waves, quantum mechanics stresses the wave character of matter. That is, the particle is "ghostly" similar to light, and light is massive similar to the particle.

2. According to Bohm, the movement of matter towards concreteness defies the traditional notion of efficient causality, and rather supports the idea of finding final causality in abiotic nature (Bohm in Hiley & Peat, 1987, p. 440). Bohm, in fact, welcomes Robert Rosen's (in Hiley & Peat, 1987) effort to trace the purposefulness of organisms to the conjecture that final causality underpins the behavior of particles and atoms.

3. One can also state that the core of the indeterminism of Ilya Prigogine's (1978,1980) work on far-from-equilibrium thermodynamics is essentially not different from the indeterminism of coin tossing. But there is an important difference between chaos theory and Prigogine's work. Prigogine's work focuses on feedbacks insofar as they explain spontaneously emerging structures stimulated by energy input from the outside. There is no center of gravity or attractor in what Prigogine calls "dissipative structures." In contrast, the emerging science of chaos explicitly incorporates the idea of internal resistance or center of gravity which in some structures can become what is called "strange attractor." There is no need in chaos theory to appeal to energy influx.

4. The concern with ability and its development is also expressed, although for somewhat different reasons, by Amartya Sen (1985) and John Davis (1995). Sen is concerned with welfare assessment in light of diverse capabilities. Davis is concerned with the personal identity of the economic agent which, he maintains, cannot be perceived through the preferences of the agent, but rather through the agent's active will or through self-creating ability. To note, Davis seems to be side-tracked when he presses his thesis to criticize the methodological individualism of neoclassical economics. He assumes that starting with preferences necessarily entails methodological individualism, while starting with ability does not. However, one may start with preferences while recognizing the social determination of preferences (e.g., Becker, 1996); and one may start with ability as if it is a Robinson Crusoe world (as this paper assumes). Put generally, the question of methodological individualism is separate from the question of the fruitfulness of starting with ability as opposed to preferences.

5. Even Keynes's notion of "animal spirits," which comes closest to express entrepreneurial activity, is far from being the motive force behind the entrepreneur. It is rather about the erratic expectations or subjectivism:

Even apart from the instability due to speculation, there is the instability due to the characteristic of human nature... Most, probably, of our decisions to do something positive, the full consequences of which will be drawn out over many days to come, can only be taken as a result of animal spirits - of a spontaneous urge to action rather than inaction, and not as the outcome of a weighted average of quantitative benefits multiplies by quantitative probabilities (Keynes, 1964, p. 161; see Matthews in Meeks, 1991).

It is true that according to Keynes (1964, p. 150), investment depends "on a sufficient supply of individuals of sanguine temperament and constructive impulses who embarked on business as a way of life, not really relying on a precise calculation of prospective profit." But this explains the bunching or timing of investments, and not the investment itself.

6. There would be no classifying instance if an agent in a new city does not search for the Tourist Information Center because he never heard of such a service in his life. In contrast, another informed agent may decide it is not worth the trouble to search for the Tourist Information Center since he is only visiting the city for one day. While the two situations differ, they do not belong to different classes as claimed by Sanford Ikeda (1990). Ikeda would call the former state "radical ignorance" while the latter state "rational ignorance." The two categories are not that different at a closer scrutiny. Radical ignorance can be transformed into rational ignorance once we expand the classifying instance as, e.g., to include the decision not to bother and ask people on the street how to get from point A to point B. The tourist does not have to know about all the particular services which the town offers. With a re-definition, he can know about them along more general categories.

7. The idea of irrevocable change should not mean that Georgescu-Roegen's (1971) main thesis - viz., the entropic degradation of energy is irrevocable - is valid (Khalil, 1990b).

8. In light of the role of experience, Marx's concept of abstract labor is untenable. The producer does not stand as an external, abstract ability vis-a-vis natural resources. The ability to harness nature is formulated through concrete experiences and, hence, labor-time cannot be easily added together across agents (see Khalil, 1992).

9. Philosophers of science call the explanations which focus on historical factors "etiological." In contrast, the neoclassical models of general equilibrium which focus on analytic time somewhat resemble Carl Hempel's view of scientific explanation a la covering-law model, or what has been called recently by philosophers of science "dispositional" explanations. While the former theories are interested in the historical development of entities, the latter ones are interested in the logical interconnection among entities. Sandra Mitchell (1993) correctly argues that both modes of explanation need not be alternatives. They can co-exist as independent modes because each deals with a different problem area.

10. Nash equilibria of incentive-compatible outcomes, such as driving rules, also arise from nonergodic dynamics. The dominance of one standard, such as the English over the metric system, may not reflect efficiency but rather historical accidents diffused because nearby others are using the inferior standard. But economists (e.g., Young, 1996, p. 106 n.3) Generally confuse positive feedbacks which originate the spread of standards with learning-by-doing which characterizes nonergodic development of technology. A country may end up with standard A rather than B for stochastic shocks, especially if both standards are equally efficient. Also, for stochastic reasons, there is a chance for a country's standard to flip. So, the dominant standard is ultimately independent of initial conditions (ergodic dynamics). The reason given by Young is that the memory of previous practices goes back to only a few incidents or years (bounded rationality). However, given that the marginal return from practicing a standard, such as stopping at a red light, falls to zero after a few practices, a more reasonable explanation of the flip is the absence of learning-by-doing phenomenon. Such a phenomenon sets the practice of standards apart from locked-in technological learning where the marginal return from learning stays positive for a long period of time.

References

Albert, David Z. Quantum Mechanics and Experience. Cambridge, MA: Harvard University Press, 1992.

-----. "Bohm's Alternative to Quantum Mechanics." Scientific American, May 1994, 270:5, pp. 58-67.

Alchian, Armen A. "Uncertainty, Evolution, and Economic Theory." Journal of Political Economy, June 1950, 58:3, pp. 211-221.

Anderson, Philip W., Kenneth J. Arrow, and David Pines (eds.). The Economy as an Evolving Complex System. Redwood City, CA: Addison-Wesley, 1988.

Arthur, W. Brian. "Competing Technologies, Increasing Returns, and Lock-in by Historical Events." Economic Journal, March 1989, 99, pp. 116-131.

Aspect, A., J. Dalibard, P. Grangier, and G. Roger. "Quantum Beats in Continuously Exited Atomic Cascades." Opt Communication, 15 April 1984, 49:6, pp. 429-434.

Bandura, Albert. "Self-Efficacy Mechanism in Human Agency." American Psychologist, 1982, 37, pp. 122-147.

Bohm, David. "The Implicate Order: A New Order for Physics." Process Studies, Summer 1978, 8:2, pp. 73-102.

-----. Wholeness and the Implicate Order. London: Routledge & Kegan Paul, 1980.

Becker, Gary S. Accounting for Tastes. Cambridge, MA: Harvard University Press, 1996.

Boulding, Kenneth E. Evolutionary Economics. Beverly Hills, CA: Sage, 1981.

Buchanan, James M. and Viktor J. Vanberg. "The Market as a Creative Process." Economics and Philosophy, October 1991, 7:2, pp. 167-186.

David, Paul. "Clio and the Economics of QWERTY." American Economic Review, May 1985, 75:2, pp. 332-337.

Davidson, Paul. Post Keynesian Macroeconomic Theory. Aldershot, UK: Edward Elgar, 1995.

-----. "Reality and Economic Theory." Journal of Post Keynesian Economics, Summer 1996, 18:4, pp 479-508.

Davies, P.C.W. (ed) The New Physics. Cambridge: Cambridge University Press, 1989.

Davis, John. "Personal Identity and Standard Economic Theory." Journal of Economic Methodology, June 1995, 2:1, pp 35-52.

Day, Richard H. and G. Piangiani. "Statistical Dynamics and Economics." Journal of Economic Behavior and Organization, 1991, 16, pp 3-36.

-----. and Pin Chert (eds.). Nonlinear Dynamics and Evolutionary Economics. New York: Oxford University Press, 1993.

d'Espagnat, Bernard. "The Quantum Theory and Reality." Scientific American, November 1979, 241:5, pp. 158-181.

-----. In Search of Reality. New York: Springer-Verlag, 1983.

Dosi, Giovanni, Christopher Freman, Richard Nelson, Gerald Silverberg, and Luc Soete (eds.). Technical Change and Economic Theory. London & New York: Pinter, 1988.

Einstein, Albert, B. Podolski, and N. Rosen. "Can Quantum-Mechanical Description of Physical Reality be Considered Complete?" Physical Review, set. 2, 1935, 47, pp. 777-780 (Reprinted in J. Wheeler and W. Zurek (eds.) Quantum Theory and Measurement. Princeton, NJ: Princeton University Press, 1983.)

Feynman, R. The Character of Physical Law. Cambridge, MA: MIT Press, 1965.

Frank, Robert H. "If Homo Economicus Could Choose His Own Utility Function, Would He Want One with a Conscience?" American Economic Review, September 1987, 77:4, pp. 593-604.

Georgescu-Roegen, Nicholas. The Entropy Law and the Economic Process. Cambridge, MA: Harvard University Press, 1971.

Grangier, P., G. Roger, and A. Aspect. "Experimental Evidence for a Photon Anticorrelation Effect on a Beam Splitter: A New Light on Single-Photon Interferences." Europhysics Letters, 15 February 1986, 1:4, pp. 173-179.

Hahn, Frank H. Equilibrium and Microeconomics. Cambridge, MA: The MIT Press, 1984.

Hargreaves Heap, Shaun. Rationality in Economics. Oxford: Basil Blackwell, 1989.

Heiner, Ronald A. "The Origins of Predictable Behavior." American Economic Review, September 1983, 83:4, pp. 550-595.

Heisenberg, Werner. Physics and Philosophy: The Revolution in Modern Science, intro. by F.S.C. Northrup. New York: Harper & Row, 1958.

Hiley, B.J. and E David Peat (eds.). Quantum Implications: Essays in Honour of David Bohm. London: Routledge & Kegan Paul, 1987.

Hodgson, Geoffrey M. Economics and Evolution: Bringing Back Life into Economics. Oxford: Polity Press, 1993.

Horgan, John. "Quantum Philosophy." Scientific American, July 1992, 267:1, pp. 94-104.

Ikeda, Sanford. "Market-Process Theory and 'Dynamic' Theories of the Market." Southern Economic Journal, 1990, 57:1.

Keynes, John Maynard. "The General Theory of Employment." Quarterly Journal of Economics, February 1937, 52:1, pp. 209-223.

-----. The General Theory of Employment, Interest, and Money. New York: Harcourt, Brace & World, 1964.

-----. "A Treatise on Probability." The Collected Writings of John Maynard Keynes, vol 8. London: Macmillian, 1973.

Khalil, Elias L. A Review of Quantum Implications: Essays in Honour of David Bohm, edited by B.J. Hiley and F. David Peat. Journal of Social and Biological Structures, October 1989, 12:4, pp. 391-395.

-----. "Natural Complex vs. Natural System." Journal of Social and Biological Structures, February 1990a, 13:1, pp. 11-31.

-----. "Entropy Law and Exhaustion of Natural Resources: Is Nicholas Georgescu-Roegen's Paradigm Defensible?" Ecological Economics, 1990b, 2:2, pp. 163-178.

-----. "Nature and Abstract Labor In Marx." Social Concept, June 1992, 6:2, pp. 91-117.

------. "Neo-classical Economics and Neo-Darwinism: Clearing the Way for Historical Thinking." In Ron Blackwell, Jaspal Chatha, and Edward J. Nell (eds.) Economics as Worldly Philosophy. London: Macmillian, 1993, pp. 22-72.

-----. "Nonlinear Thermodynamics and Social Science Modeling: Fad Cycles, Cultural Development and Identificational Slips," American Journal of Economics and Sociology, October 1995, 54:4, pp 423-438.

-----. "Non-Linear Dynamics versus Development Processes: Two Kinds of Change" The Manchester School, September 1996a 64:3, pp. 309-322

-----. "Kenneth Boulding: Ecodynamicist or Evolutionary Economist?" Journal of Post Keynesian Economics, Fall 1996b, 19:1, pp. 83-100.

-----. "Buridan's Ass, Risk, Uncertainty, and Self-Competition: A Theory of Entrepreneurship." Kyklos, June 1997, 50:2, in press.

Knight, Frank H. Risk Uncertainty and Profit, intro. by George J. Stigler. Chicago: University of Chicago Press, (1921) 1971.

Krugman, Paul. Geography and Trade. Belgium: Leuven University Press jointly with Cambridge, MA: The MIT Press, 1991.

Kurdas, Cigdem. "The 'Whig Historian' on Adam Smith: Paul Samuelson's Canonical Classical Model." History of Economic Society Bulletin Spring 1988, 10:1, pp. 13-23.

Lachmann, Ludwig M. Capital, Expectations and the Market Process. Kansas City MO: Sheed Andrews, 1977.

Laplace, P.S. de. A Philosophical Essay on Probability. Dover: New York, (1820) 1951.

Lawson, Tony. "Probability and Uncertainty in Economic Analysis." Post Keynesian Journal of Economics, 1988/89. 11, pp. 38-65.

Loasby, Brian J. Equilibrium and Evolution: An Exploration of Connecting Principles in Economics. Manchester: Manchester University Press, 1991.

Lorenz, Edward N. "Deterministic Nonperiodic Flow." Journal of the Atmospheric Sciences, 1963a, 20, pp. 130-141.

-----. "The Mechanics of Vacillation." Journal of the Atmospheric Sciences, 1963b, 20, pp. 448-464.

-----. "The Problem of Deducing the Climate from the Governing Equations." Tellus, 1964, 16, pp. 1-11.

Meeks, Gay. (ed.) Thoughtful Economic Man: Essays on Rationality, Moral Rules and Benevolence. Cambridge: Cambridge University Press, 1991.

Mitchell, Sandra D. "Dispositions or Etiologies? A comment on Bigelow and Pargetter," The Journal of Philosophy. May 1993, 90:5, pp. 249-259.

Nelson, Richard R. "Recent Evolutionary Theorizing About Economic Change," Journal of Economic Literature, March 1995, 33:1, pp. 48-90.

------. and Sidney G. Winter. An Evolutionary Theory of Economic Change. Cambridge, MA: Harvard University press, 1982.

North, Douglass C. Institutions, Institutional Change and Economic Performance. Cambridge University Press, 1990.

Polyani, Michael. Personal Knowledge: Towards a Post-Critical Philosophy. New York: Harper & Row, 1958.

-----. The Tacit Dimension. Garden City, NY: Doubleday, 1966.

Popper, Karl. "Indeterminism in Quantum Physics and in Classical Physics." British Journal for the Philosophy of Science, 1950, 1, pp. 117-133.

Prigogine, Ilya. "Time, Structure, And Fluctuations." Science, 1 September 1978, 201:4358, pp. 777-785.

-----. From Being to Becoming: Time and Complexity in the Physical Sciences. San Francisco: W.H. Freeman, 1980.

Radzicki, M.J. "Institutional Dynamics, Deterministic Chaos, and Self-Organizing Systems." Journal of Economic Issues, March 1990, 24:1, 57-102.

Robinson, Joan. "History Versus Equilibrium," In Collected Economic Papers, vol 5, Oxford: Oxford University Press, 1979.

Romer, Paul M. "Endogenous Technological Change." Journal of Political Economy, 1990, 98, pp. S71-S102.

Rosser, J. Barkley. From Catastrophe to Chaos: A General Theory of Economic Discontinuities. Dordrecht, Netherlands: Kluwer, 1991.

Ruelle, David Chance and Chaos. Princeton: Princeton University Press, 1992.

Savage, Leonard J. The Foundation of Statistics. New York: Wiley, 1954.

Schotter, Andrew. The Economic Theory of Social Institutions. Cambridge: Cambridge University Press, 1981.

Sen, Amartya. Commodities and Capabilities. Amsterdam: North-Holland, 1985.

Shackle, G.L.S. Expectations in Economics. Cambridge: Cambridge University Press, 1952.

-----. Expectations, Enterprise and Profit. London: Allen and Unwin, 1970.

Simon, Herbert A. Models of Man. New York: John Wiley, 1957.

-----. "From Substantive to Procedural Rationality." In Spiro J. Latsis (ed.) Method and Appraisal in Economics. Cambridge: Cambridge University Press, 1976, pp. 129-148.

Stewart, Hamish. "A Critique of Instrumental Reason in Economics." Economics and Philosophy, April 1995, 11:2, pp. 57-83.

Stiglitz, Joseph E. "Recurrence of Techniques in a Dynamic Economy." In J. Mirrlees and N.H. Stem (eds.) I.E.A. Models of Growth. New York: Wiley, 1973, pp. 138-161.

Witt, Ulrich (ed.). Explaining Process and Change: Approaches to Evolutionary Economics. Ann Arbor: University of Michigan Press, 1992.

Young, H. Peyton. "The Economics of Convention." Journal of Economic Perspectives, Spring 1996, 10:2, pp. 105-122.