How it all began

I first met Axel Leijonufvud at the Certosa di Pontignano in 1993 during a Workshop entitled "Simulating Societies"   in which, together with Luigi Marengo, I presented the paper "Division of Labour and Social Coordination Modes".[1]

In the paper we argued that any new division of labour requires new forms of coordination, and considered many different modes in which activities and plans are coordinated, ranging from one extreme case to another: a pure hierarchical, and a decentralized process.  

By means of a simple simulation model based on Genetic Algorithms and Classifiers Systems, we then compared the dynamic properties of each coordination mode and, in particular, their speed and cost of adaptation to changing environmental situations. The basic architecture was borrowed from Herbert Simon’s theory of problem solving. 

I confess I was not aware that some years earlier ( 1984), in his well known work “Capitalism and the Factory System,” Axel had deeply analysed this question, so I was astonished when after my presentation he courteously came up to me and started a pleasant discussion. What followed was a long friendship and an ever increasing collaboration with the Trento University Department of Economics, which became systematic when he accepted the offer of a “chiara fama” chair at the university in 1995

The factory system

I would like to briefly recall one aspect of his contribution to the question of the “factory system”, where he holds that the neoclassical production theory is unfit to explain the process of industrialization and, according to Adam Smith, suggests to start instead from the process of division of labour.

It cannot be denied that in the initial phases of industrialization,[2] the steam-engine made available an amount of power that was crucial to large-scale production.[3] Some industrial revolution scholars generalize when presenting this historical episode by claiming  that it is the introduction of new machinery  that explains the growth of the factory system. Axel questions this view:

“It is not all that obvious, therefore, what role should be assigned to indivisible machinery, in explaining the emergence of the factory as the dominant form of manufacturing enterprise. Some questions remain. Why, for example, did not the steam engine simply lead many independent masters to locate in the same workplace (and, perhaps, pay rent for the right to attach their new-fangled machines to the overhead, steam-powered shaft)?”

He suggests, instead, a different explanation by designing a process in which the mechanization is a constitutive part of the division of labour. I will rephrase his reasoning in different terms. The emergence of the factory system is the outcome of two processes that unfold and coexist. On the one hand, the process of  vertical division of labour into more and more specialized and simple tasks, which requires  the collaboration of more and more un-skilled workers. It follows that the market of labour becomes easily accessible and it is extremely fluid because the tasks in a production line do not require a long-time learning. Moreover, the production becomes standardized and it is possible to extend them easily to a large size. This is the first part of the story and we should not be surprised by the  negative reactions of the contemporaries to the social consequences of the vertical division of labor and the unfolding of the  movement of luddism.

But there is the second side of the coin: while an increasing vertical division of labour requires rigid coordination among more and more unskilled workers, on the same time the simplification and routinization of the different tasks is the outcome of the design of the new mode of production. Axel argues that the entire process is unfolding thanks to the design/innovation of new technologies and new coordination patterns :

“The mental task of analyzing the production process so as to carry through the division of labor leads to the discovery of these opportunities for mechanization. Once you get the hang of the division of labor, you will discover how to mechanize industry.”

Put in a different way, the division of labour can be interpreted as a process of collective problem solving resulting in the construction of new technologies and new coordinating procedures. This view implicitly assumes that the process is likely to be the outcome of a  “distributed intelligence,” i.e. of the spontaneous actions of individuals’ will, that brings about a new production pattern which can be interpreted as if it were made according to a single plan, although a central planner does not exist.[4]

Therefore, while the factory system is expanding, in parallel with the creation of simplified and routinized jobs, new organizational and technical competences will be created and new jobs will appear on the labor market. This dualistic process replicates itself iteratively. Moreover, along with the development of artificial intelligence, a large number of procedures can be substituted by  artificial program, thus also rendering the new jobs progressively obsolete. As far as a large part of the new competences can be routinized and mechanized this new routinization requires, again, the creation of new skilled jobs.

Therefore, the process is an evolutionary one, in which the proportion between old unskilled and new skilled jobs fluctuates as far as new innovations are realized. This, in my view, well explains the increasing complexity of the process, a process where the increasing differentiation in production applies to the entire productive system. The economy becomes more complex as it expands.

Axel remarks:

“It is this increasingly complex coordination (when it can be maintained!) of larger and larger numbers of specialists that shows  up as increasing productivity. As one proceeds with the analysis of this Classical Division of Labor theory, it increasingly escapes the analytical categories of static neoclassical production theory. The Classical theory becomes a theory of an evolutionary process, rather than a theory of the rational choice between known alternatives. Recall that Smith and Marx both insisted that the New Division of Labour preceded the mechanization of industry. “

I will not dwell on this issue, but the reasoning we have made so far suggest that Axel’s idea of using the “division of labour” as key to unlocking the problem of the factory system is a very fruitful strategy. His view allows us to respond to some questions that were unsolved within the traditional approach, for example why capital hires labour and not vice versa:

“Growth of the market will offer opportunities for driving the functional differentiation of both labor and capital equipment further. In the early stages of industrialization at least, the implications for capital and labor are not symmetrical, however. The typical machine becomes a highly specialized piece of equipment, “dedicated” to particular tasks in the manufacture of a particular product. This specialized machine may have no alternative uses, but it is, on the other hand, not quickly or easily replaced. The typical factory worker works at a specialized task but a basically unskilled one. He could easily and quickly qualify for many other similar jobs elsewhere, but is on the other hand himself replaceable with equal ease. In brief, the functionally differentiated machine has a “thin” market, while the worker performing a highly differentiated function has a “thick” market. This asymmetry in the position of the two factors of production affects their bargaining power.“

Once he decided to open the black box of neoclassical production function, Axel adopted a “institutionalist” view, in the sense that the processes involved in the working of the factory system cannot be reduced to the introduction of machines: technological progress and organizational progress go together and as such need to refer to a wide social, organizational, even political context.[5]

One relevant consequence is that this problem cannot be explored only through the use of  the traditional toolbox of microeconomics and thus the use of simulations and of the mathematics of complexity to represent adaptive and evolutionary processes behind the division of labour becomes necessary.

In Trento, in 1991, we launched a Laboratory of Experimental Economics, in which the first relevant activities were in collaboration with a group from the University of Michigan that included John Holland, Robert Axelrod and Michael Cohen. The group was dedicated to the analysis of social systems as adaptive systems. This was a special interest for Axel, given that adaptative models, machine learning and simulation were useful tools for the analysis of market and factory as social institutions.

The “corridor” and the invisible hand

Axel went beyond the general question of the toolbox for analysing adaptive systems. In Towards a Not-Too-Rational Macroeconomics  he claimed that Economics deals with complex dynamic systems and the related analytical theories and methodologies.  The question was to  find the condition of stability of the economy, portrayed as a machine that has to "compute" the equilibrium not in a centralized way as in the Walrasian tatonnement, but in a distributed way. To this end, he had created at UCLA a laboratory of "Computable Economics", and after his affiliation with the Faculty in Trento, he promoted studies in computable economics at the Trento Lab.

His micro-assumptions behind the macroeconomic processes reverse the mainstream approach, which requires that the more complex the economic problem to be solved, the greater must be the information, competence and computational skills of the economic agents. According to Kirman’s influential ideas  on complexity, Axel notes that many complex and sophisticated collective behavioral patterns are the product of agents who follow very simple rules:

“We might start, then, by asking how believably simple people cope with incredibly complex situations. If we knew a bit about that, we could then go on to study the conditions under which market interaction will and will not configure the complex system into that incredibly smart allocational pattern.”

This leads us to the question of the failures of the “invisible hand,” because people’s choices on the market are reciprocally compatible only when the invisible hand is working well. Laissez faire doctrines invoke Adam Smith to claim that the economy is structured in such a manner that invariably prices converge to equilibrium. But this picture does not correspond faithfully to Adam Smith’s vision of “Invisible Hand”:

"Indeed (the individual) does not generally intend to pursue the public interest, nor is he aware of the extent to which he is pursuing it. When he prefers the support of the productive activity of his own country instead of that of a foreign one, he is aiming only at his own security, and when he directs that activity in such a way that its product is the greatest possible, he is aiming only at his own gain and is led by an invisible hand, in this as in many other cases, to pursue an end which is not within his intentions. Nor is the fact that this end is not always within his intentions always a detriment to society. By pursuing his interest, he often pursues the interest of society far more effectively than when he actually intends to pursue it" (Smith 1974, 174 ).[6]

The conditions under which by pursuing his interest the individual does not pursue the interest of the society are today studied under the label of “market failures Smith is convinced that egoism in political action is socially dangerous, while - unlike them - he suggest that in economic action selfishness can bring social benefits. Let me make a digression on this point. This view got an incredible success since the beginning, because throughout the Middle Ages the evaluation of human economic and political actions [7] were characterized by severe moral principles. The role of interest, which throughout the Middle Ages was seen with a negative connotation, so much that the Church did not allow individuals or organizations to give loans at interest, in The Wealth of Nations is proposed as a positive factor thanks to the process of competition.

Therefore, Smith maintains a “precautionary principle” about human passions that characterized him as the most important political philosopher since Machiavelli, and is neatly expounded by David Hume:

“Political writers have established it as a maxim, that, in contriving any system of government, and fixing the several checks and controls of the constitution, every man ought to be supposed a knave (rogue), and to have no other end, in all his actions, than private interest. By this interest we must govern him, and, by means of it, make him, notwithstanding his insatiable avarice and ambition, co-operate to public good. Without this, say they, we shall in vain boast of the advantages of any constitution, and shall find, in the end, that we have no security for our liberties or possessions, except the good-will of our rulers. “ [8]

Smith shares this view, but argues that individual passions are transformed into a social advantages through a mechanism that is neither constituted nor regulated by a social contract, and therefore is an “invisible hand.” But which are the limits of the invisible hand?

Where the social relations require a visible hand

Smith suggests that the market sustains itself without the need for an authority to establish and enforce the contractual rules  provided that the invisible hand works properly. Only under this assumption market and public authority can be considered to be clearly separate areas.

But more generally this separation is misleading: from the very beginning of the growth of the markets the need for a public authority to guarantee the parties from defaults or violations emerges.

In a hypothetical economy that self-replicates without modifications (the Marxian simple reproduction or the Schumpeterian circular flow) all activities are routinised and there are no conditions for the emergence of opportunistic behavior; the parties trust each other, and the successful completion of contracts is assured. But, in normal conditions, when changes happen in time, even if the parties’ behavior were completely correct, it could easily happen that the conditions of a contract at the moment of realisation were different from those in which the contract had been planned. The occurrence of unforeseen conditions could give rise to different interpretations between the parties and, consequently, to disputes and litigation. Therefore, it is necessary for trade between the parties to be contractually guaranteed, which requires a public authority with enforcement powers.[9]

When there are discrepancies among the parties, the conditions for a perfect working of the invisible hand are not fulfilled, and is no more true that making his own egoistic interest the individual produces beneficial effects on the society. On the contrary the effect of the egoistic actions may harm some individuals or the entire collectivity.

In these conditions the conflicts of the opposite interests between the parties - for example sellers and buyers - are no longer solved by the invisible hand; the conflict emerges explicitly and individuals should make very a complex calculation to find if there are fair solutions; to avoid  retaliations, every individual should limit or exclude actions that could create harm to others and vice versa. To this end, individuals should perform a strategic forward-looking calculation.[10] Of course this calculation is in general very complicated and it is mostly impossible to predict the opponents’ strategies.  Therefore, the search for a solution must be assigned to some third part, in general to a Public Authority.  This is a typical situation of market failure in which the “Coase theorem” does not apply.[11]

There are two radically different conditions under which this happens: in one condition, there is a loss by one part and a gain by the other. In this case it is possible to find a solution on the ground of shared social norms or legal rules under the pressure of a Public Authority.[12] 

But when there is collective loss, a “fair” process of adjustment is not easy to be detected. Axed remarks:

“For relatively small, isolated defaults, a well-developed market economy has well-defined legal rules that specify the priority ordering of creditors and so on and which therefore determine who has to take the necessary loss. The system can tolerate a certain amount of such defaults and still work more or less normally. Here we have assumed that the volume and magnitude of broken promises went beyond this tolerable level.

The enforcement process that is then triggered does not have much to do with optimal calculation and the associated equilibria. It becomes rather a matter of the system mindlessly grinding away, ruining some and saving others in an often highly arbitrary manner. If market processes are just left to run their course, the eventual outcomes will certainly not conform to those notions of justice and fairness that have previously made people willing and accepting participants in the system.

Consequently, when default occurs on a large scale, the rules themselves end up in the political arena. In this process, the effective rights and obligations of agents become still more uncertain and ultimate outcomes very opaque.[13]

This is a very remarkable issue: the collective loss cannot be solved by an invisible hand, but necessarily requires a political authority which, on the ground of shared norms and rules, realises a proposal of mediation that “fairly” distributes the collective loss.

“…..Because, of course, social interaction does not always produce the perfectly rational result. Sometimes, as James Tobin once said, "the invisible hand is nowhere to be seen." Ordinary people also interact to produce booms and busts in real estate, credit crunches and bank panics, great depressions and hyperinflations -- and much other misery besides.

What we should aim for is to model "systems that function pretty well most of the time but sometimes work very badly to coordinate activities" [21; 23].(page 5)

This point is later expanded in the “corridor” hypothesis. It is a crucial question within  Axel’s views, which remains in the background of many of his works: how it happens that an economy can become unstable or remain stable, and therefore under which conditions the metaphor of the invisible hand is no more viable. 

“The economy is a complex dynamical system. In tranquil times, economic agents may make coherent plans up to some fairly distant horizon. In times of financial distress or of high inflation, decision-making is for the most part very short-term in both the private and the public sector. Short-sighted adaptive behavior leads easily into complex system dynamics. In the present context we are interested in the balance between deviation-counteracting and deviation-amplifying (unstable) processes. The former are the familiar market processes that keep departures from equilibrium prices and outputs within more or less stringent bounds. Unstable processes are cumulative but, in the cases of interest here, do converge so that the deviation-amplifying movements are nonetheless bounded. “

Conclusion

Axel was a forerunner of powerful ideas that unfortunately emerged in a historical period in which the mainstream was deaf and sometimes hostile to his views. Instead, the Walrasian development of monetarism, the new-classical macroeconomics which combined the postulate of continuously clearing markets with the rational expectations hypothesis became dominant.

In Macroeconomic Crises and the Social Order (2003) Axel ironically notes:

“Encounters with younger colleagues in Europe or the United States often make me reflect on how the worldview of economists have changed over the last forty or fifty years. Back then, most believed that the private sector was unstable and riddled with market failures but that an enlightened and benevolent government could fix all that. In more recent times, this worldview has taken a 180-degree turn – a rather dizzying turn for people of my generation. The majority of economists now appear to believe that “free markets” will not only be stable but, if not interfered with, deliver outcomes so nearly optimal as makes no difference. Only the short-sighted, time-inconsistent meddling of politicians who tax too much and spend even more cause problems.“

Behind these ironic observations there was the “irresistible rise” of Muth’s rational expectations theory and the resulting rise of the  new-classical macroeconomics, in which to some extent the neoclassical interpretation of the Smithian parable of invisible hand was resurrected again and with major vigour and more sophisticated formal treatment.

But in parallel, an alternative view and methodology was slowly but robustly emerging, the one initiated by Herb Simon’s theory of bounded rationality, which gave rise to a new generation of models in which  uncertainty, complexity, computability and machine learning  are formally treated and are progressively entering the background of economic analysis. It is a research stream that in my opinion is suitable for providing the right toolbox for taking into account Axel’s pioneering ideas.