Response to “ Worrying Trends in Econophysics ”

This article is a response to the recent “Worrying Trends in Econophysics” critique written by four respected theoretical economists [1]. Two of the four have written books and papers that provide very useful critical analyses of the shortcomings of the standard textbook economic model, neo-classical economic theory [2,3] and have even endorsed my book [4]. Largely, their new paper reflects criticism that I have long made [4,5,6,7,] and that our group as a whole has more recently made [8]. But I differ with the authors on some of their criticism, and partly with their proposed remedy. 1. Money, conservation laws, and neo-classical economics Our concerns are fourfold. First, a lack of awareness of work which has been done within economics itself. Second, resistance to more rigorous and robust statistical methodology. Third, the belief that universal empirical regularities can be found in many areas of economic activity. Fourth, the theoretical models which are being used to explain empirical phenomena. The latter point is of particular concern. Essentially, the models are based upon models of statistical physics in which energy is conserved in exchange processes. Gallegati, Keen, Lux, and Ormerod [1] Since the authors of “Worrying” [1] begin with an attack on assumptions of conservation of money and analogs of conservation of energy in economic modelling, let me state from the outset that it would generally be quite useless to assume conserved quantities in economics and finance (see [4], Ch. 1). There is no reliable analog of energy in economics and there are very good reasons why no meaningful thermodynamic analogy can be constructed (see [4, Ch. 7]). In particular, money is not conserved. Money is created and destroyed rapidly via credit. Leveraging leads to big changes in ‘money’. Are there any conservation laws at all in real markets, and if so do they have any significance for deducing market dynamics? Conservation laws follow from invariance principles [8], so one should not expect conservation laws in socioeconomic ‘motions’ like financial transactions or production and consumption [4,5]. We have identified exactly one invariance principle in finance: no arbitrage is equivalent to a discrete version of rotational invariance of the price distribution [see [4], Ch. 7). For inviolable mathematical laws of motion rather then era-dependent mathematical modelling we know from Wigner’s explanation why mathematics has been so ‘unreasonably effective’ in physics [8] that we would need the equivalent of all four space-time invariance principles: translational invariance, rotational invariance, time translational invariance, and Galilean invariance. Those local invariance principles are the foundation for the discovery of mathematical law in classical mechanics, quantum mechanics, and general relativity, as Wigner has explained. Wigner pointed out that without those underlying invariance principles mathematical law might well exist, in principle, but the world would be too complicated for us to discover mathematical law in practice. That is exactly where we stand in trying to make a science of socio-economic phenomena [7], but even in the absence of inviolable mathematical laws of trading behavior we are still able to generate finance market statistics dynamically accurately [4] in our era. In the definition of the neo-classical economic model (‘general equilibrium theory’ [9]) one global conservation law is assumed at the start. However, for the global integrability presumed for the existence of equilibrium there actually must be n global conservation laws, where n is the number of different goods exchanged by agents. The budget constraint combined with price dynamics, dp/dt=D(p)-S(p) (where D is total demand at price p and S is the total supply), yields Walras’ law, a global conservation law that puts the motion in price space (phase space) on an n-sphere [4,9]. The budget constraint is conservation of money, and that constraint is badly violated in the real world, where money is created and destroyed with the tap of a computer key via credit. In the beginning of our current era of credit and deregulation, which we can date from the early 1970s, the supply-demand curves predicted by neo-classical economics [9] were falsified by Osborne [10], who simultaneously predicted the observable microeconomic supply and demand curves. In a strict gold-standard economy money would be conserved if mining production were ignored. Credit is required for the expansion of both industry and consumption, which is why we’re no longer on any sort of gold standard. Nixon took the Dollar off the (Rooseveltian) gold standard in 1972 due to the inflation created by the Vietnam War. Option pricing was deregulated in 1973, accidentally parallel with the revolutionary Black-Scholes paper. OPEC correctly understood the Dollar devaluation and raised the price of oil dramatically to compensate for the rapid inflation that would follow the degradation of the dollar. The deregulation of financial markets that followed is described in Liar’s poker [11]. In contrast with the main concern of “Worries” [1], the work done in Fribourg, Boston, Alessandria, UCLA, Palermo, Oldenburg, and Houston, to name a few centers of econophysics research, do not assume conservation of money or any analog of conservation of energy. Searches for conservation laws can be found in the neo-classical economics literature [12,13], but those efforts bore no fruit at all for lack of a firm basis in empirical data. There is one econophysics model assuming conservation of money that is interesting within the historical context of neo-classical economic theory. Radner had observed in 1968 [14] that money/liquidity cannot appear in neoclassical economics: with perfect foresight, perfect planning into the infinite future, there can be no liquidity demand (in real markets we have, in contrast, largely total ignorance, market as pure nonGaussian noise). Radner speculated that money/liquidity arises from uncertainty and also computational limitations, although he did not say what he meant by “computation” and apparently had no idea at all of a Turing machine [15]. Conservation of money is assumed in the simple trading model made by Bak, Nørrelyke and Shubik [16]. There, expected utility was maximized (so the paper is neoclasssical) and then noise was added to see if “money” would emerge from the dynamics of trading. The attempt failed, the model is empirically irrelevant (as is every model of optimizing behavior) but can be understood as an honorable effort to address the mathematical problem posed by Radner’s discussion. I have no evidence that Per Bak was aware of Radner’s paper, but maybe Shubik was aware of it. The Bak et al model is asymptotically a pure barter model because statistical equilibrium is reached, and in general the authors of “Worries” are right that making further barter models is generally (both empirically and theoretically seen) useless. In our era, conservation of money is a silly assumption. The paper by Bak et al tried to bridge the gap between standard economic theory and econophysics but was too ambitious, the gap is a chasm. 2. Should econophysicists rely on the economics literature for guidance? Our concerns about developments within econophysics arise