The Formal Ontological Critique of Neoclassical Economics

Throughout this book and especially in the above chapter, I have been highly critical of the positive image of science that still pervades economics and the associated Newtonian metaphor. Similarly, in drawing attention to the historical emergence of the capitalist system and its institutions, I have been questioning the ahistorical nature of economic analysis and the ahistorical human nature it claims to investigate. Additionally, I have been critical of the narrow mathematical formalism of the neoclassical program and the fictional assumptions it employs to maintain the mathematical tractability of its analysis. In this, I am joining distinguished economic company, as we saw earlier in this chapter. Despite such objections, the use of mathematics has become the unifying feature of the neoclassical program and that modelling is often employed for its own sake—rather than for the light it is believed to throw on economic phenomena.[68] This formalism results from the unfounded belief that mathematical methods are an essential part of the scientific method—a belief derived from classical physics and ultimately from Pythagoras via Plato and the Enlightenment. It is also an essential part of the illusion that neoclassical economics is a positive rather than a normative discipline. It is simply assumed that these methods can be applied irrespective of the nature of the domain being studied. As we saw earlier, however, the use of this method cannot be equated with ‘science’ in general.

Lawson, a critical realist, makes the same point in the language of ontology—that is: the study of the nature and structure of being, of the nature of what exists and, in the particular case, of the nature of social reality. To transfer mathematical deduction and the Newtonian metaphor successfully from classical physics to economics requires both domains to share underlying ontological characteristics—that is, that they are special cases of the same general thing. It is this sharing of characteristics that can allow parallel forms of analysis. Clark expands this point by suggesting that there are two types of analogy that involve shared characteristics that are used in this transfer. In the first, the formal structure of the model must correspond in some way with the phenomena being explained. In the second, the shared characteristics involve a material analogy—as in the assertion of a uniform invariant human nature.

Human economic behaviour does not, however, share characteristics with the physical entities of classical physics. They are very different things, not special cases of the same general thing. Lawson calls their confusion the abductionist fallacy. As Clark points out: ‘Societies are not natural phenomena ruled by natural laws, nor created by natural forces. They are the creation of humans, as are their institutions, culture and history.’[69] Consequently, mathematical deduction is not a tool that is appropriate for the use to which it is being put.

To reiterate, there is a mismatch in most cases between the nature of social and economic phenomena and the nature and structure of reality presupposed by mathematical deductive modelling and its world-view. Furthermore, Lawson points out that the occurrence of event regularities of the sort that can be analysed scientifically by mathematical modelling are rare even in the natural realm—being restricted mostly to well-controlled experiments. Mathematical deductive modelling as an explanatory form or structure requires a closed, self-contained, atomistic, deterministic system that allows the deduction of consequences or predictions. This follows from Hume’s conception of causality as constant conjunctions of brute, atomistic events in which ‘same cause, same effect’ applies everywhere.[70] In this view, causal laws are empirical regularities that are reducible to sequences of events and those events to experiences.[71] These are essential features of the particular explanatory form. Such atomistic entities are required to have separate, independent and invariable properties. This is not true of human beings, who are active agents in an open, complex world. In this world, there are no constant conjunctions among events and we cannot rely on empirical generalisations as law-like statements.[72]

Adolph Lowe drew attention to this fundamental objection to neoclassical modelling as long ago as 1935, when he argued that neoclassical economics rested on a highly questionable conception of mechanistic rationality and the constancy and uniformity of individual behaviour.[73] Interestingly, Lowe points out that even the laws of supply and demand involve adjustments of a much more complex nature than proposed in the mechanics of a pendulum, which do no more than maintain a preordained equilibrium.

He therefore rejected the idea of uniform economic laws, including the so-called laws of supply and demand. As he claimed, ‘[T]here are today no reliable laws of economic behaviour on which prediction can be based.’[74] Similarly, Knight—also in 1935—criticised the mechanical analogy in economics, and in particular the idea that the motive causing an economic action was understood as a force, similar to that idea in classical mechanics. That idea in mechanics—as well as having been discredited—was also criticised by March and Heinrich Hertz as being metaphysical. It acquired respectability only because in mechanics it was experimentally reproducible, not because it provided any ultimate explanation of mechanical behaviour. That experimental reproducibility is not true, however, for economic preferences.

Lowe, in his 1965 book On Economic Knowledge, develops his early critique into a formal argument with similarities to Knight’s and Lawson’s. Like them, he questions whether the logical structure of economic systems can be defined properly in mechanical terms. Lowe finds three features that are cornerstones of classical mechanics, which do not match those of real economic systems: the atomistic hypothesis; the mode of behaviour of these ‘atoms’—what he calls the extremum principle, the economic equivalent of a force, the universal action directives operating in economic affairs, maximisation and minimisation; and a conservation hypothesis in which market processes are understood in terms of the conservation of energy in a closed system. As in Newtonian mechanics, the atomised entities of neoclassical economics are entirely independent and entirely reactive in a predictable manner to the forces acting on them with no independent freedom of movement. This simply denies human agency; and this feature of mathematical deduction requires economists to specify their theories in terms of atomised entities that are isolated so that only a few factors bear on phenomena, and to produce constant and invariable responses to given conditions so that event regularities are guaranteed. In such a closed system, event regularities occur in a causal sequence.

Economic ‘idealisation’ is, then, falsely supposed to produce a situation analogous to the experimental situations in the natural sciences, which attempt to close the system and to prevent any interference with the operation of the mechanism under study. Explanation, however, requires a move from phenomena at one level to underlying causal conditions. In experimental conditions in the natural sciences, most event regularities are restricted by experimental control so that the workings of specific intrinsically stable causal mechanisms are isolated from the effects of countervailing factors. The purpose of experimentation in a controlled environment is to identify empirically an underlying causal mechanism—not to produce event regularities for its own sake. It must also be assumed that the powers attributed to agents must always be exercised in non-experimental situations and in the given ways, regardless of the real outcome. This is so in the experimental situations used in the natural sciences: the underlying causal relationships identified by any experiment also operate in a predictable way in non-experimental situations and the event regularities produced relate to an underlying empirical causal relationship.

Such isolated event regularities in the social disciplines are rare because aspects of social systems are not static, but are open and intrinsically dynamic, involving a multitude of shifting causes. Such social systems exist in a constant state of becoming, exhibiting emergent properties and causal power dependent on human agency and practice but not reducible to that agency and practice, and dependent on internal relationships with other social entities. We are all involved in a very large number of different and changing, relatively stable social positions or roles that are independent of the individuals occupying them and which influence what we can do. These roles involve rights and obligations with normative force only and are related in turn to other roles occupied by other individuals with their different rights and obligations. Our reactions to social situations are related internally and are highly context specific, with primary importance probably attaching to the relations between roles rather than between people. Nevertheless, human beings possess intentionality or agency: they are not just passive reactors to the environment but are forward looking and make variable choices. Lawson argues that individuals form their longer-term goals in terms of the enduring, highly abstract aspects of society with the intention of adapting those plans to the specific contexts they encounter in action through life. They are therefore involved in the emergence of a sense of identity of the individual—the autographical narrative of the individual. Lawson warns us also against treating the features of social reality that are rather abstract as though they are concrete, and of mistaking the particular for the general. As we saw in Chapter 2, such social behaviour involves a skilled performance that is not reducible to analytical rationality involving the simple application of rules, but involves the transcendence of rationality by intuitive, experience-based, situational behaviour. In short, social entities are not invariable in behaviour, they are not atomistic and they are not reducible to a representative agent. Indeed, because of the way in which the social structure and human agency depend on each other—but are not reducible to each other—methodological individualism is not a tenable position in social theorising: ‘Mainstream economics continually falls back on states of affairs, etc, that could not possibly come about.’[75]

In any event, in economics, the fictions assumed are not employed as a means of isolating the influence of other phenomena so as to identify an underlying causal relationship that explains how the surface phenomena was produced, but as essential preconditions to the mathematical deductive form of analysis and to the generation of the results. At the same time, abstractions such as representative agents and preferences are treated as if they are concrete and can successfully isolate the real phenomena of interest from other influences. Because of the above interdependencies, it is highly unlikely that social behaviour can be manipulated in any useful way by the experimental researcher. In short, Weber’s qualified justification for his idealisations is misconceived.

To summarise, the closed atomistic and deterministic system of deductive mathematical modelling—this modern manifestation of a priori reasoning in economics—cannot, in general, fit the open, non-deterministic and non-atomistic social world. This lack of fit renders mathematics an inappropriate tool in the study of most economic and social phenomena; and this is the reason why this sort of modelling has been unsuccessful. It simply lacks the ability to illuminate most of the social realm. This is what Aristotle effectively told us 2,300 years ago. This is the fundamental reason why it is illegitimate to use this form of modelling and the associated Pareto-optimality decision rule in policy analysis. In addition, Lawson sees these methods as a barrier to true progress in economics—preventing us from uncovering any real causal relationships in economic phenomena. Although not excluding entirely the possibility of using mathematical deductive methods, Lawson is pessimistic about the prospects of their successful use. Consistent with the views expressed in Chapter 6, he calls for a pluralistic and interdisciplinary approach to economic analysis, but one involving a critical reflection on the ontological assumptions being used.