Complexity and Economics
Victor A. Beker
Victor A. Beker is Professor of Economics at the University of Belgrano and the University of Buenos Aires, Argentina.
Please cite the paper as:
Victor A. Beker, (2017), Complexity and Economics, World Economics Association (WEA) Conferences, No. 2 2017, Economic Philosophy, Complexities in Economics
Abstract
Nobody will discuss that the economy constitutes a very complex system. The traditional approach to understanding it has been to reduce complexities to simple rules and behaviors, abstracting of many features of the real economy.
An alternative to reductionism consists of studying economic systems with a complexity approach. The complexity approach ́s point of departure is that the behavior of the whole is much more complex than the behavior of the parts.
Complexity economics has focused on economic phenomena like business cycle, crises and other out of equilibrium behavior. Its use of non-linear models offers the advantage that the same model allows us to describe stable as well as unstable and even chaotic behaviors.
The use of non-linear models in finance as well as the possibility of finding chaotic behavior in economics are discussed. Models of interacting agents in economics and finance are mentioned as another promising line of research in complexity applied to economics. Finally, whether the complexity approach is another twist of orthodoxy or constitutes a heterodox paradigm is another issue discussed in the paper.
The abstract sounds very promising; I am looking forward to read your whole text, when technical access is made possible.
Dear Prof. Victor A. Beker ! You are right to point to the methodological barriers of economic complexity research. Since the~ 5000 years of our civilization, we are used (habit is a despot) to linear economic accounting models (mainly for private wealth), which are all post mortems, and data science is about Gaussian distributions. To communicate with economic complexity, e.g. via central banking, is mainly an informed guess as many excellent experts have confessed. It is like the art of chess-playing; the brilliant player can anticipate one-to-none game operations. The decisive question from your article is, for me: How can human economic agents communicate with complexity? This methodical research will be mainly about the right assessment strategies (statistical learning) of the systems feedbacks (response analysis). Our worst economic response to complexity is the control freak panic of increased planning agency, which is not (!) only typical for a closed command economy, like the Soviet system was. So, *congrats* for pointing with your research paper into the right direction !
Dear Prof. Beker ! Your methodological caution is justified; we are used to about ~ 5000 years of linear calculation and accounting, and this mainly for book-keeping techniques of private property and credit. How can human economic agents communicate with complexity ? If we take a look at central banking, we can see the art of informed guesses, with some statistical learning strategy (non-Gaussian). Our methodical behavior is similar to the brilliant chess-player, who can foresee none-to-one game operation. Our methodological wisdom must be expanded in a manner that enables us to communicate with complexity; this comes usually with the methodical ‘interpretation’ of feedback loops, we can call that methodical response analysis (4th order cybernetics,…). In addition, we should never forget that our human economy is part of the physical earth, despite all technical ingenuity or modelling, i.e. the body economic is physically part of matter/energy. In any case, congrats for this forward thinking research paper.
complexities are bound to remain as long as we focus on behaviour and to analyse it make rules (say rationality) about it. Behaviour is a free will related entity and thus beyond the pale of rules. the trick to simplifying economics lies is analysing the limits, beyond which the free will does not operate. man is mortal, mortality–limited lifespan– is a limit beyond which no human being can act. the barrier of limited life space (of which limited amount of possible work, each individual can perform) is the key to begin from. if an individual has to do good for himself–enjoy the higher level of goods and services–he has only two ways: one, improve the productivity of his production (related to whatever he wishes to enjoy) or two, to utilise power (defined ability to influence the other individuals in a way that it does not leave any option for the other except to pay) to to get some work time from other individuals. macroeconomy is built from the productivity of production of each individual production and power in each exchange. economics thus has only two variables to deal with: productivity of production and power and thus enters simplicity in economics that is beyond recognition. Reference: Malhotra S. K.(2008) Next economics: Simple, Assumption less, Falsifiable. Shakuntala Devi Memorial Trust E-162 Kalkaji, New Delhi, 110019, India.
Comment on Victor Beker’s paper: Complexity and economics
Yoshinori Shiozawa
2017.10.11
Basically I have the same stance with respect to mainstream economics and the same prospects for complexity economics. However, I feel two kinds of uneasiness when I read Beker’s paper.
The first kind of uneasiness is the details of criticism against mainstream or neoclassical economics. His criticism is very rough and not very exact. There are many young economists who are still thinking within the neoclassical framework but feeling uneasiness on various parts of their economics. Beker’s criticism has risk to push them back to mainstream economics, because they feel that the criticism is not well based and not accurate enough.
For example, let me list up some of such statements:
(1) Nobody will discuss that the economy constitutes a very complex system. (p.2)
(2) For instance, the neoclassical general equilibrium model is concerned with the static, timeless allocation of resources. (p.2)
(3) Convexity was a necessary assumption to warrant uniqueness of (p.3)
(4) If the equilibrium is unique, history does not matter: sooner or later the system will arrive at that unique equilibrium.(p.3)
There might be no need to explain why these propositions are only roughly correct. Beker himself must know it. However, we should be careful for this kind of general account of neoclassical economics.
When Beker wrote statement (1), “nobody” must have meant “nobody among neoclassical economists.” Even if we understand in this way, statement (1) is in contradiction with the fact that there are some economists who think complexity approach hopeful but does not oppose to the neoclassical research program, as it is explained in section 7.
This is only nitpicking but there is more serious problem in (1). Let me cite a paragraph from Frank Hahn’s General Equilibrium Theory (Hahn 1984, Chapter 3).
Although Hahn did not use the term “complexity,” it is evident that he and other neoclassical economists were concerned with the complexity of the economic system, its chaotic and anarchic nature. General equilibrium model of Arrow-Debreu type is for them “the most important contribution that economic thought has made to the general understanding of social processes.” (ibd.) Beker’s statement like (1) reveals simply un-understanding of neoclassical economics and does not hint many of young students the necessity to change their orientation. In order to criticize in a true way, it is necessary that we present better explanation than neoclassical economics on how market works and misfunctions.
The case of statement (2) is simple. One important example of neoclassical general equilibrium model is named DSGE. The capital D here signifies dynamic. Then some students misunderstand that Beker has no knowledge of DSGE and criticizing neoclassical economics based on his old knowledge before 1980’s.
Statement (3) is not false as itself. But, this is not the true reason why neoclassical economics preferred to exclude increasing returns. If returns were not supposed to be decreasing, one cannot define firm-level (and by consequence economy-wide) supply functions. In addition, it would be necessary to point that multiple equilibria do occur from the demand side, even if returns are decreasing. Convexity or decreasing returns are necessary conditions but not sufficient conditions.
Statements (2), (3) and (4) are related. Even if equilibrium is unique, it does not imply that the economy converges to the equilibrium. It is widely known among neoclassical economists that there is no good theory that warrants the economy to converge to the equilibrium. Ordinary stability condition of equilibrium is not a true stability which must prove that the economy out of equilibrium shifts to equilibrium state. There is no such theory. If readers of Beker’s paper know these facts, they may judge that Baker is not really versed in the state-of-the-art of microeconomics.
The second kind of uneasiness is the tone in which Beker talks about non-linearity and chaos theory. I willingly admit that non-linearity and chaos are one hopeful method of analysis, but we have to admit at the same time that they are by no means a magic stick which solves everything. For an observer like me who followed chaos theory since its emergence in 1970’s, the achievements that chaos theory and dynamic systems analysis in general brought about are rather disappointing. Qualitatively it gives us some new insights, but there was no substantial improvement in quantitative predictions. Even if we put aside the Holy Trinity, newly obtained results are not illuminating.
Beker spend a whole section (section 5) on the detection of chaos in economic time series. This very fact demonstrates that chaos is not very useful in the trade cycle analysis. The existence of a positive Lyapunov exponent does not imply that the time series is chaotic. It may simply be divergent. Moreover, estimating Lyapunov exponent for a real time series is quite difficult as opposed to the imaginative dynamical systems written as a system of differential equations.
Section 6 argues interactive systems. It is true that a financial market like exchange rate is extremely interactive. It is easy to criticize that neoclassical economics is neglecting these interactions between agents. However, pointing Keynes’s beauty contest and herd behavior is not sufficient. It is necessary to develop an analytical tool to examine what is happening and what happens. Yet, we have no good model to analyze these herd behaviors.
As far as I know, the best model that describes herd behavior is Orléan (1990). It is a model of financial contagion. Each agent chooses by a toss one of two possibilities, high value or low value. The market price is given by the mean of their choices. This market continues for infinite times. Each agent uses the market price as information about how other people are acting and changes his/her probability of choice. This generates a dynamic process. As this is a stochastic process, each process traces a different pass but Orléan had succeeded to get the mean distribution of this process. What is astonishing to me, this stable (i,e. timeless) distribution can be expressed by a polynomial which depends on only one parameter s. Big s means that the agent principally uses fundamentals. Small s means that the agent mainly follows the market. Thus the parameter s determines the stable distribution. This permits us to make many interesting interpretations.
In my opinion, Orléan’s model was exceptionally successful model of this kind but it seems Orléan himself was not satisfied by the model and abandoned it without writing a full paper in English. Of course, there are limitations to his model, but it may have been developed further.
This anecdote teaches us that merely pointing the existence of phenomena like beauty contest and herd behavior does not make economic analysis progress. It must be the starting point but only a starting point. I feel Beker talks too optimistically the existence of some difficult-to-analyze mechanisms in the economy. To find them may be important, but if we cannot find a tool to analyze them, we have progressed very little. We should take account of this fact.
Beker points as “another promising line of economic modeling” Agent-based Computational Economics (ACE). I myself prefer to use ABS, which is an abbreviation of Agent-Based Simulation. I worked on this theme for a long time. Indeed I have started U-MART project with my colleagues in information engineering from 1989. U-MART is a simulator to which human agents can participates as well as machine agents. I omit all details, because we have already published many papers and two books (Shiozawa et al. 2008 and Kita et al. 2016.)
I have a big hope for ABS. I am thinking that ABS may become a similar analytical tool for future economics just like mathematical analysis was for 20th century economics. It is the indispensable tool from various viewpoints. Major reasons will be two necessities (1) to incorporate heterogeneous agents and (2) to implement process analysis. Mathematics is hopelessly powerless in these implementations. This is one of the reasons why mainstream economists cannot intrude in this field even if they also know the necessity to do so (Shiozawa 2016).
However, it is also important to note that ABS is still in a very premature stage. At present, we can implement ABS in a very realistic way but it is very difficult to extract any firm knowledge from the simulation. Many of our trials are still in the stage of “garbage in, garbage out.”
Computation is now one of three principal modes of scientific examination: experiments, theory and computation. ABS in economics has not arrived to this stage. It is necessary that we develop this new tool (let me say the third mode of scientific research) further. Until it becomes a true tool of analysis, it may take other two or three decades. (For more detailed argument, please see Shiozawa 2016.)
As a conclusion, I want to say this: stop talking of mainstream economists as fools. They are making their greatest efforts. They are very intelligent, powerful, and strong rivals. We are still wondering on a wild field. The only difference between them and us is that they are heading a wrong direction whereas we are probably heading a more correct direction.
Reference
Hahn, Frank (1984) Equilibriun and Macroeconomics. Basik Blackwell, Oxford.
Kita, H., K. Taniguchi and Y. Nakajima (Eds.) Realistic Simulation of Finacial Markets, Springer, Tokyo.
Orléan, André (1990) Le rôle des influences interpersonnelles dans la détermination des cours boursiers, Revue économique 41(5): 839-868.
Shiozawa, Y. (2016) A Guided Tour of the Backside of Agent-Base Ssimulation. In Kita et al. (2016) Chap. 1, pp.3-50.
Shiozawa, Y., Y. Nakajima, H. Matsui, Y. Koyama, T. Taniguchi, and F. Hashimoto (2008) Artificial Market Experiments with the U-MART System. Springer, Tokyo.
First of all, I want to thank Professor Shiozawa for his thorough reading of my article on Complexity and Economics and for his detailed comments on it.
His arguments make me think that perhaps my line of reasoning was not very clearly presented, so I will take this opportunity to try to make them clearer.
1.- The point of departure of my argument is that I can´t imagine anybody objecting the idea that the economy is a very complex system. I assume that economists –either orthodox or heterodox- coincide on this point.
However, there are two basic ways of studying that complex system. The traditional one has been the reductionist approach. A new one is the complexity approach.
Up to now both orthodox as well as heterodox economists have mostly used the reductionist approach. The difference lies in what features of the real world are kept in the theoretical model and what features are disposed of.
In the same way, the complexity approach has been used in the neoclassical theory context and also in a heterodox one. That´s why I emphasize that it is still too early to conclude whether the complexity approach is another twist of orthodoxy or constitutes a heterodox paradigm.
2.- DSGE models have been a contribution by New Keynesian economists, not neoclassical ones. Although I don´t think they are truly Keynesians I wouldn´t call them neoclassical (see my article in the real-world economics review n° 80).
3.- I never said that convexity or decreasing returns are sufficient conditions.
4.- Prof. Shiozawa remarks that “Even if equilibrium is unique, it does not imply that the economy converges to the equilibrium.” Once again I don´t see where the discrepancy lies. We all know that one thing is the existence of equilibrium and another one its stability.
5.- Section 5 gives some reasons why the detection of chaos in economic time series faces more difficulties than in the natural sciences. They are some of the reasons why I think that economics should give chaos theory another chance -as I explain in my 2014 article mentioned in the references of my article- in spite of the disappointing results obtained so far.
6.- I appreciate the references to Orléan´s paper as well as to Shiozawa´s own contributions on ABS –and I suppose many readers will also be thankful to Prof. Shiozawa for the same reason.
7.- I agree with Prof. Shiozawa´s concern about those many young economists who are still thinking within the neoclassical framework but feeling uneasiness on various parts of their economics. The key issue is to try to open their minds to alternative approaches and convince them that there is nothing like ‘the’ economic theory but only a collection of economic theories -our collective diversified intellectual portfolio- that are in competition with each other.
Finally, dear Prof. Shiozawa, let me thank you once again for the time devoted to the analysis of this paper and for your contributions which will let me improve its final version.
Whatever scientific examination and methodology pathway we follow, our accounts of society have to reduce any complex level of human economic activity into workable units of book-keeping. Exponential knowledge automation will speed up the complex frequencies of all our economic trans-actions, and this will be the greatest challenge for the economic discipline as profession and science in the coming years.
Very useful overview for this complexity novice!
1. I’m interested in agents who do more than optimize, and I’m interested in the role that ideas and theories play in decision making. However, your point about the difficulty in building models to study areas like economics where most of the data is of the low frequency variety is a good one. I did, however, come across Robert Shiller’s 2017 AEA Presidential Address on “Narrative Economics,” which uses Google searches to trace the “rise and fall” of different ideas (memes?) over time. These can be simple narratives, such as “don’t wait until you’re priced out of the housing market,” or more complicated theories like the so-called “Laffer Curve,” which played a big role in the justifying the tax cuts of the Reagan and Thatcher administrations.
2. You point out that Richard Day (1994) says, “An economic system is dynamically complex if its deterministic endogenous processes do not lead it asymptotically to a fixed point, a limit cycle, or an explosion.” Can this be roughly translated as saying, “An economic system is dynamically complex if its future ‘path’ cannot be predicted”? Does this mean it has no equilibria? (Apologies if these are silly questions).
3. Don’t know if this is of interest to anyone besides me, but my paper (for this conference) seems to match up with 3 or 4 of the 5 summary criteria of complexity put forward by Robert and Yoguel’s (2013): i) “heterogeneity, ii) disequilibrium and divergence, iii) interactions and partial information iv) network architecture, and v) emergent properties.” I have i) autonomous agents with ii) multiple beliefs, who iii) interact in many institutions and environments on the basis of diverse beliefs, which produces iv) some kind of disequilibrium, if only in the form of disappointed expectations.
4. You write, “The multiplicity of equilibria means that there are many possible worlds.” A philosophical question for you: Are all these worlds made of the same basic stuff or can they be made of different “materials,” e.g., different concepts? This isn’t a well-formed question.
5. You write, “From this perspective, the economy can be seen as a process of self-organization: the system ‘chooses’ between the different options that are presented to it.” I don’t think you mean that the “system,” itself, is an agent, but rather are employing something like Daniel Dennett’s “intentional stance.” The danger in this is well-described by the Wittgensteinian philosopher, P.M.S. Hacker, in a number of books, but especially in The Philosophical Foundations of Neuroscience, where Hacker argues that you make decisions, your brain doesn’t. You can Google to find Dennett and John Searle’s side of the argument if you’re interested.
Thanks for sharing your article. There’s a lot more in it that bears comment, but you’ve certainly got me thinking.
Dear Greg,
Thank you very much for your interesting comments on my paper. I will try to answer your questions.
1.- The relationship between complexity and prediction is not a simple one. For instance, see Mihailovic et al. (2014) at https://www.hindawi.com/journals/amete/2014/878249/
2.- I think you´re right in your argument on the relationship between your interesting paper and Robert and Yoguel´s criteria.
3.- I´m not a philosopher, so I presume you have a better answer to questions 4 and 5. What I want to underline is that multiple equilibria implies that the result attained is path-dependent. History matters.
Thank you for the references in point 5. I will read them with interest and benefit.
Very Good insights into complexity economics. Complexity economics is promising
and more theoretical and empirical work is needed to reveal uncertainty
in complex economics as well as improve this new approach.
Duality of human knowledge is the fundamental principle of every theoretical explanation in economics . Economic reality, since entirely constructed by human being in accordance with his or her value preferences (which can considerably deviate from rational ones), must become an object for itself; such a transformation of a pure subject into an object for itself is impossible without primary duality residing within the person (the person is at the same time both the subject and the object of nature). This duality cannot be reduced, since, being the general condition of conceiving economic phenomena, it is also, according to our theoretical approach, the principle of every theoretical explanation. Now every theory need be aimed solely at reducing all opposites of the economy to the primary opposition of the cognising subject, who is no longer himself, but appears as a manifestation of economic phenomena. Economic systems, also like living systems in nature, maintain themselves through a process of eternal turnover, separating on the one side from what they connect with on the other, and connecting here with what they separate from there.
Dear Professor Beker,
I enjoyed reading your research on the complexities in Economics. I certainly support the notion that the comportment of economics all together is much more multifaceted than the performance of its sections. The complex nature of the subject is even more telling because of the prevalence of time series in many economic analyses. In addition, time series, more often, follow a non-linearity pattern in regression analyses. The introductions of non-linearity, even with its problems, offer several options in dealing with stable and volatile features in time series data. Your research was rich and purposeful in describing the options of non-linearity models in economics, together with adding a novel knowledge of the complexities of economics.
Dear Victor A. Beker, I had gone through your abstract which is not only very interesting but also contains areas where we need to not only think but to investigate.
The answer to getting to grips with complexity in macroeconomics is to consider the various functions that different parts of the system have in common. It will be seen that these functions are much more limited in kind than the huge numbers of people who are performing them. In my working paper SSRN 2865571 “Einstein’s Criterion Applied to Logical Macroeconomics Modelling” (which is freely available on the internet), I show that by considering these functions as being idealized aggregates whitin the the Big Picture, there are actually only about 19 variables all together (of money flow verses goods, services, access rites, valuable paper, taxes, etc.) These various flows pass between only 6 entities or agents who have input and outputs of these flows and the money parts mostly balance, as the particular agency adjusts its internal activity, consumes the material mutual flows, etc. Then this “fear” of our subject becoming too complicated to properly understand is seen to be a hypothetical situation which is unrealistic when one makes the simplifying assumptions noted above.
There are further complications due to the need for subsequent decision-making by the entities and its effects, but compared to the scale of how this subject begins, they are small.
Becker has enough reason to argue that complexity economics could add more interesting features to existing economic models. Shiozawa is also right that we should be cautions at the early stage of complexity economics.
From my observation, Brian Arthur made important contribution in explaining the existence of Silicon Valley by INCREASING RETURNS, which was inspired by Prigogine’s idea of constructive role of positive feedback, such as auto-catalysis in chemical reaction; while increasing returns to scale is excluded in microeconomics since supply curve does not exist in optimization.
We observed that neoclassical growth theory has conflicting thoughts. Solow model of exogenous growth based on CONSTANT RETURNS to scale predicting a CONVERGENT trend; while Romer’s model of endogenous growth base on INCREASING RETURNS implies a DIVERGENT trend. Both models failed to explain RISE & FALLl of great powers in history. We proposed METABOLIC GROWTH theory with DYNAMIC RETURNS for a tech WAVELET. Here, nonlinearity is only a math representation. The real challenge is selection of PROPER MATH for STYLIZED EMPIRICAL PATTERNS. See. Ping Chen, Metabolic Growth Theory, Journal of Evolutionary Economics, 24(2),239-262 (2014).
Another example is the misleading image of MATHEMATICAL CHAOS that sounds like greater DISORDER rather than HIGHER kind of ORDER OUT OF CHAOS (Prigogine). We discover that the so-called WHITE CHAOS in difference equations is simple in math modeling but irrelevant for empirical research, since we do not have intrinsic time unit in biology or economics. We prefer COLOR CHAOS model in differential equations that is capable of describing monetary chaos and stock market, since it has irregular amplitude but narrow frequency band. We consider COLOR CHAOS model is a good math model for BIOLOGIC CLOCK or ECONOMIC ORGANISM by Schumpeter or SPONTANEOUS ORDER by Hayek in business cycle theory, More importantly, COLOR CHAOS model is structurally more stable than LINEAR OSCILLATOR, such as Samuelson model of accelerator-multiplier. In this regards, the idea of BUTTERFLY EFFECT, SAND PILE model and EDGE OF CHAOS cannot survive as biological and economic structure. Schrodinger had better idea in his classical book WHAT IS LIFE (1948), he pointed out that organism needs both STABILITY and VARIABILITY to adapt changing environment. His concept of META-STABILITY can be visualized by COLOR CHAOS or a potential WELL with FINITE DEPTH.
Neoclassical economics only consider equilibrium and stability in linear models, while complexity scientists may over-state instability in chaos models. In fact, color chaos model has local instability with multiple equilibriums but global stability characterized by STRANGE ATTRACTOR.
We must have a BALANCED understanding of stability and variability in biological and economic dynamics. See. Ping Chen, Economic Complexity and Equilibrium Illusion, Essays on Market Instability and Viable Market, Routledge (2010).
In sum, we should be optimistic for the future of complexity economics plus cautions in applying complexity modeling in economic analysis.
Ping Chen, Professor of Peking University, Beijing, China
Thank you very much, Professor Ping, for your valuable contribution to the debate.
Dear Prof Beker,
I read your paper and enjoyed the insight story which is mentioned as complexity of the economics. This paper also focus on the traditional approach of economic analysis. Really, it’s fantastic and investigative deep discussion regarding the economic complexity.
Thank you so much Professor in such a nice way of discussion.
My best wishes Prof Beker