TRANS Internet-Zeitschrift für Kulturwissenschaften 17. Nr. Februar 2010

Sektion 2.9. Der neoliberale Markt-Diskurs. Zur Kulturgeschichte ökonomischer Theorien im Alltagsdiskurs
Sektionsleiter | Section Chair: Walter Ötsch (Zentrum für Soziale und Interkulturelle Kompetenz und Institut für Volkswirtschaftslehre, Johannes Kepler Universität, Linz)

Dokumentation | Documentation | Documentation

The irrational foundations of economics

Why neoclassical economists choose the wrong theory from physics

Sylvie Geisendorf (Universität Kassel) [BIO]



1. Introduction

Rational choice or as-if methodology in economics is a largely disputed endeavour, although mostly at the margins of the profession. Mainstream neoclassical economists seem to be surer than ever of its relevance and justification, to the extent that they do not even see a need to give good reasons for it any more. Much can be said about the causes and even possible advantages for this venture and some of these arguments will be discussed in this paper. Although being a fictitious assumption, rationality seems to fit nicely into several psychologically proven structures of human thinking, thus helping economists to reason about an otherwise confusing amount of data (Røgeberg and Nordberg 2005). Though this may be true and important, the discussion about the foundation of rational choice theory in economics seems to forget where the assumption originates or rather why it was indispensable. Rationality was a necessary precondition for the employment of optimization mathematics. Economists wanted to fit their theory to the one of celestial mechanics, long before Milton Friedman’s as-if reasoning. But optimization was not the only possibility to describe how physical systems react to gravity. It isn’t even the exact one. The real explanation for the impact of universal gravity is given by the equations of Newton’s law of gravity. The main argument of this paper is that neoclassical economists adopted the wrong theory from physics. The paper tries to examine some of the reasons for this choice and discusses its consequences for the economist’s way of thinking. It finally illuminates the possibilities that an employment of the correct physical theory would offer.


2. The economy as a mechanical system

Neoclassical economics is based on optimization mathematics of classical physics, going back to Newton’s discovery of universal gravity (1687). All physical objects are amenable to the law of gravity and tend to equilibrium positions (falling objects) or stable orbits (planets), if no external force is exerted upon them. In physics, this insight was the beginning of the era of “rational mechanics”. Starting at the end of the 17th century with Galileo, Descartes and particularly Newton, physics began to turn away from the Aristotelian metaphysical and teleological view in favour of a predictable world, guided by the laws of nature. In contrast to former times, natural phenomena were now explainable by causes, measurable and reproducible. In other words: nature became rationally tractable.

Economists, so far governed by Adam Smith’s invisible hand, that was more the hand of god than one of independent natural laws (Weise 1998), were inspired by the clarity and logic of these scientific explanations and wanted to transmit them to economics. They believed to see similarities between the physical and the economic system, justifying such a transfer. In 1871 Jevons wrote about the „mechanics of utility and selfinterest” (Jevons [1871] 1970: 90). In 1874 Walras was convinced that: „It is already perfectly clear that economics, like astronomy and mechanics, is both an empirical and a rational science.” (Walras [1874] 1954: 47) Edgeworth wrote: „As electro-magnetic force tends to a maximum energy, so also pleasure force tends toward a maximum energy” (Edgeworth 1881: 25). Ultimately the economists at the end of the 19th century were fascinated by the beauty and importance of the astronomic universe and wanted to see their area of interest and their science as important as astrophysics: „Thus the system of the economic universe reveals itself, at last, in all its grandeur and complexity: a system at once vast and simple, which, for sheer beauty, resembles the astronomic universe” (Walras [1874] 1954: 374).

A non-negligible advantage of an approach to physics was the mathematics it provided. Started by Lagrange, Hamilton completed the development of a general maximizing equation (known as Hamilton´s principle) in 1834, to determine final states of systems in motion in classical mechanics. Physical systems therefore were not only controlled by natural laws. The impact of the laws could be calculated exactly. Economists strived for the same degree of exactness.

Thus natural sciences became “rational” and economics became “rational” and “mechanist”, the terms in both cases meaning measurable and guided by the laws of nature instead of metaphysically. At first therefore it were not economic agents (let alone physical objects) that were rational, but the science describing them. But of course rationality (or probably some hidden mechanism with the same effect) was always a necessary assumption, if optimization mathematics should work, even if it was hidden behind other terms at first. It was “pleasure force” (Edgeworth 1881: 25) or utility that was maximized – but by what or whom? “The energy generated by pleasure force is the physical concomitant and measure of the conscious feeling of delight” (ibid.). Apparently by some natural law, equivalent to the one guiding planets in their orbit. For early neoclassicists this seemed to be a less important question

Long after this inaugural period of neoclassical economics – and as long ago from contemporary economics – an active debate about the justification of the assumption of rational agents was taking place. Around the 1950th the rationality assumption was largely debated and officially introduced to neoclassical theory. The debate was situated around Milton Friedman’s as-if argument (1953). Rational choice wasn’t a clear-cut postulate of economics but a helpful construction to approach reality: People do not act entirely rational, but market results present themselves as-if they would do (Friedman 1953). By later economists this was basically interpreted as: “assumptions are irrelevant if you predict accurately” (Røgeberg and Nordberg 2005: 547), although economists of the time seemed to take the assumption itself more serious. There are several variations of Friedman’s argument, indicating that the assumption was supposed to have something to do with reality. Alchian (1950) e.g. stated that although not all economic agents behave rationally, the selective forces of the market rule out those who don’t. A similar argument was that rational behaviour could be learned in time.

A major problem of contemporary neoclassical economics might be that such a necessity to justify the as-if methodology seems not to be seen any more. Rational choice is an integral part of neoclassical models. Textbooks introduce optimization mathematics without even mentioning that it necessitates the assumption (thus making it seem as-if it was a deliberate choice). Røgeberg and Nordberg (2005) explain that the unquestioned acceptance of such an obviously “absurd” concept might be founded in the bounded rationality of economists themselves. Human beings have a strong believe in regularities, intentionality and predictability - even against hard evidence. And the use of regularities helps them organize their perceptions and knowledge. The subsuming of huge amounts of otherwise confusing data under a simple mental model enables human beings to think and theorize about some problem at all. Knowing about the importance of celestial mechanics for the whole scientific community of the 19th century, it is at least comprehensible why economists were ready to seek a correspondence between their realm and this fascinating and scientifically established subject. And once the corresponding mathematics had been accepted and deemed true, a fitting of the somewhat indistinct behaviour of economic agents to the exact mathematical method seemed an obvious step.

Trying to understand how economics evolved from a science trying to become rational to one based on an irrational rationality assumption it doesn’t even defend anymore is interesting in itself. But it is even more interesting to see that economists, in their endeavour to make economics as logical as natural sciences, didn’t realize that physics offered two ways to formalize the phenomenon of equilibrium or gravity, and that they choose the wrong one. Economists employ a formalism that isn’t even valid in physics when it comes to explain the causes of a phenomenon.


3. The confusion of cause and effect

Physical phenomena concerned with gravity can be explained from two directions. One is the description of the forces exerted upon an object and its reaction thereupon. The other is a description of the effects this has. In other words, the former is a causal explanation, while the latter is a functional one. Both kinds of equations describe the same phenomenon and are mathematically correct and equivalent under certain conditions, but only the former depicts how an effect is induced.

Imagine a ball you place at the edge of a bowl. It rolls inside and after a while rests at the bottom. It thereby generates the state of lowest energy requirement. The ball is doing so, because it is subject to gravity, not because it knows how to minimize energy needs. All the same it is possible to formalize its behaviour by minimization mathematics, using the form of the bowl as a function to be minimized. Assume the function of a smoothly shaped bowl to be f(x) and friction to be negligible. Determining the final state of the ball than simply amounts to deriving the lowest point of the bowl:

  1. This is a possible explanation of the effect gravity has on the behaviour of a ball. But it only holds under specific conditions. Maximizing (or minimizing) is a crude simplification of what’s actually going on in a physical system. It doesn’t describe how or what forces operate in a system. It only derives its final state, given that the main assumed force – here gravity – is not superimposed by something else. The correct description of the causes of the phenomenon however would necessitate an explicit employment of Newton’slaw of gravity which would imply that all the prevailing forces have to be considered explicitly in the equation:
  2. with f(x) still being the form of the bowl, the inertia force, composed of mass (m) and acceleration, the normal force, dependent on the function f(x) and the force of gravity (mg) and the frictional effect of , with friction b. And this is still a simplification, neglecting the torsional moment of the ball.

Obviously (1) is the easier way to calculate and under simple conditions it is quite alright to use it. The problem however is that even in physics it only holds under specific conditions. As soon as the bowl has no smooth form anymore, but e.g. several minima (some of them being just local ones), it is not straightforward anymore where the ball will end. Optimization fails to offer a solution. Of course it can derive all minima, but it falls short in determining which one the system will actually attain. Only the exact record of the exerted forces and their magnitude in (2) is able to show whether the energy will suffice to drive the ball over the edge of local minima.

And this is the problem of neoclassical economics in a nutshell: Instead of describing the forces operating in economic systems, it describes the effects these forces produce under certain conditions. Such a confusion of cause and effect is dangerous. Not only because it turns logic around in proclaiming effects as reasons (the equilibrium attracts economic agents), but also because it’s all too easy to forget that the solution can only be attained under particular conditions.

And here is where – probably unnoticed by the founders of neoclassical economics – the parallels between economics and physics end. Physicists are and were well aware of the limited reach of optimization – as they are by the way of the limits of the whole of classical mechanics for quite a long time. And they never mixed up cause and effect. Fermat’s principle states that light takes the fastest (not the shortest) path between two points. When light passes from one medium to another – e.g. from air to water – it is reflected at a certain angle. No one assumes that it does so because it consciously attempts to calculate the fastest way to get from A to B. In fact we know that it doesn’t even “want” to get to B. But economists do. As their objects of interest are people, consuming goods, there was no obvious natural law, like the one guiding light in its course, to explain why they should maximize some function (utility), so neoclassical economists had to assume that they do it consciously.

There are three reasons for such a method being tricky. First the original intention to make economics as rational as natural science can not be fulfilled this way. Economics, contrary to natural science, is not a rational science because it has made itself waterproof. Forced to admit that not everybody is literally maximizing monetary utility, the utility concept has been stretched more and more to encompass everything from love over a need for altruism to drug consumption (Becker 1996). Now virtually everything can be explained as the maximization of some sort of utility function. But this seeming and often criticised economic imperialism comes at a high cost for economics itself. The economic concept of maximizing utility is no theory any more. In explaining everything it becomes irrefutable and therefore tautological. According to Rosenberg (1979) this is a problem of all “extremal” theories: „To stop such theories from being able to explain any possible observation we must be able to specify the domain or explanatory variables of these theories independently of what they try to explain.“ Physics and biology did so, economics failed to.

The second reason for the awkwardness of the chosen method is the analytical power of the causal explanation it has to renounce to by using the simpler functional explanation. Although mathematically equivalent, the two theories are quite different in their abilities. Let’s have one more look at the ball in the bowl to illustrate this point. Assume the bowl not being smoothly shaped anymore, but having a bulge somewhere on the way down. Now the ball is not able to just role down to global minimum but might be caught in the local minimum in-between. As stated above, differential calculus is able to work out both minima. But it can not determine under which conditions the ball will reach which one of them. To do so, you would need something like Newton’s equations, enclosing mass, acceleration and friction of the ball. To analyse how a complex system like the economy behaves under more realistic conditions than a smoothly shaped bowl of states we need a more adequate description of initial conditions. In sticking with the wrong side of physics theories, economics deprives itself of this possibility. And it deprives itself of two dimensions of analysis that are particularly relevant in systems changing as fast as the economic system does and being as vast: time and space. Optimization is a method that doesn’t take place in the real world. It’s a simplified abstraction. But time matters. Even if equilibriums were attainable it might take to long to reach them. Relevant parameters might change in between so and force the system to adapt to new conditions, long before it was able to respond properly to the former ones.

A third and last point to mention here is the needlessness of the controversial rationality assumption. Rational agents were needed for optimization but they are unessential if we employ real classical mechanics. If we put the actual forces working in economic systems into our models, we do not need such a simplified assumption. The behaviour of economic agents can be modelled in all its imperfectness and a lot more detail and maximization would get its proper place: as just one special outcome under ideal conditions.


4. An alternative modelling approach

Before giving a brief insight into an alternative way to construct economic models, I would like to emphasize the intent of this paper. It has been written to point out the choice neoclassicists had when building economic theory upon the model of classical physics. This choice was between the causal explanation of Newton and the functional one of maximization calculus, and the paper states that economists followed the wrong path. The current paper is not concerned with the question whether classical physics is a useful analogy for economic theory at all. Realizing how much change and evolution is going on in economic systems it is probably not. But that’s not what we shall deal with here primarily. Although the proposed alternative sort of jumps several steps in scientific development and refers to approaches closer to the theory of evolution than to classical physics, it is still an alternative showing what can be done when we want to construct a causal economic theory.

Interesting enough some early neoclassicists tried to do so. Marshall (1890) attempted to include the evolution of the economic system over time into his equations. But contemporary neoclassical economists followed Jevons (1871) narrower focus on equilibrium (Wolfson 1995: 281). Admittedly, at the time, it would have been far more complicated to consider more of the actual forces working in economic systems. Equations encompassing such aspects become quickly unsolvable. But today we have other opportunities and we should try to fit economic theory more to reality. We have powerful computers and are able to write simulation models that can solve equations numerically and even include any desired verbal or logic statements. A closer depiction of the actual behaviour of economic agents is technically no problem any more and it is done in abundance in evolutionary economics, referring to the concept of bounded rationality. This is not the place to discuss the details or advantages and problems of such models. It has been done elsewhere and by others in abundance (e.g. Brenner 1999, Van den Berg et al. 2000, Dopfer 2005, Geisendorf 2007, or the Journal of Evolutionary Economics). Concluding, I only want to point out that such an explicit modelling of the motives and capabilities of the agents we are concerned with is what is required when we want to make economic forces explicit. Those interested in the analogy to physics however, have to be aware that such models are no literal translation of Newton’s equations. The analogy goes as far as depicting actual forces and attempting causal instead of functional explanations, but there are no literal counterparts for mass, acceleration or gravity.


5. Summary

Neoclassical economics is based on classical mechanics. But instead of adopting Newton’s laws of motion it chooses to employ Hamilton’s principle of general maximization. Whereas the motion laws explicitly name the forces acting upon a system, Hamilton’s principle just deduces the final outcome. Although mathematically equivalent for some cases, the two approaches are therefore quite dissimilar. The main difference being that, even in physics, optimal solutions, in the sense of global minima or maxima, are only possible under specific conditions. As soon as the problem considered has no unique solution, only an explicit introduction of the prevailing forces would reveal the actual behaviour of the observed system. But the problems induced by the use of the wrong mathematical approach go even deeper than limiting economic analysis to very specific cases. Contrary to the laws of motion, optimization does not take place in time and space – two dimension of analysis economic theory therefore has to forgo. And optimization mathematics forced economists to make very ambitious assumptions about the intellectual capacity of its agents – the controversial assumption of perfect rationality. Both problems would not have occurred, had the economists chosen the correct physical approach. The paper discussed these problems, some reasons and a possible solution. A different form of economic modelling, considering initial conditions and therefore the causes of economic phenomena and the bounded rationality of economic agents, is possible today by simulation models.



2.9. Der neoliberale Markt-Diskurs. Zur Kulturgeschichte ökonomischer Theorien im Alltagsdiskurs

Sektionsgruppen| Section Groups| Groupes de sections

TRANS   Inhalt | Table of Contents | Contenu  17 Nr.

For quotation purposes:
Sylvie Geisendorf: The irrational foundations of economics. Why neoclassical economists choose the wrong theory from physics - In: TRANS. Internet-Zeitschrift für Kulturwissenschaften. No. 17/2008. WWW:

Webmeister: Branko Andric     last change: 2010-03-04