Economic Evolution, Learning, and Complexity

The twelve papers in this collection grew out of the workshop on “Economic Evolution, Learning, and Complexity” held at the University of Augsburg, Augsburg, Germany on May 23-25, 1997. The Augsburg workshop was the second of two events in the Euroconference Series on Evolutionary Economics, the first of which was held in Athens, Greece in September 1993. A special issue of the Journal of Evolutionary Economics (1993(4)) edited by Yannis Katsoulacos on “Evolutionary and Neoclassical Perspectives on Market Structure and Economic Growth” contains selected papers from the Athens conference. The Athens conference explored neoclassical and evolutionary perspectives on technological competition and increasing returns. It helped to identify the distinguishing features of evolutionary scholarship. The Augsburg workshop was more oriented toward exploring methodological issues in evolutionary and related scholarship. A number of the papers employed new methods, such as genetic programming and experimental analysis, some developed new econometric techniques or raised new empirical issues in evolutionary economics, and some relied on simulation techniques. Twelve papers covering a range of areas were selected for this collection.

Interaction among autonomous decision-makers is usually modelled in economics in game-theoretic terms or within the framework of General Equilibrium. Game-theoretic and General Equilibrium models deal almost exclusively with the existence of equilibria and do not analyse the processes which might lead to them. Even when existence proofs can be given, two questions are still open. The first concerns the possibility of multiple equilibria, which game theory has shown to be the case even in very simple models and which makes the outcome of interaction unpredictable. The second relates to the computability and complexity of the decision procedures which agents should adopt and questions the possibility of reaching an equilibrium by means of an algorithmically implementable strategy. Some theorems have recently proved that in many economically relevant problems equilibria are not computable. A different approach to the problem of strategic interaction is a “constructivist” one. Such a perspective, instead of being based upon an axiomatic view of human behaviour grounded on the principle of optimisation, focuses on algorithmically implementable “satisfycing” decision procedures. Once the axiomatic approach has been abandoned, decision procedures cannot be deduced from rationality assumptions, but must be the evolving outcome of a process of learning and adaptation to the particular environment in which the decision must be made. This paper considers one of the most recently proposed adaptive learning models: Genetic Programming and applies it to one the mostly studied and still controversial economic interaction environment, that of oligopolistic markets. Genetic Programming evolves decision procedures, represented by elements in the space of functions, balancing the exploitation of knowledge previously obtained with the search of more productive procedures. The results obtained are consistent with the evidence from the observation of the behaviour of real economic agents.

We consider an oligopolistic market game, in which the players are competing firms in the same market of a homogeneous consumption good. The consumer side is represented by a fixed demand function. The firms decide how much to produce of a perishable consumption good, and they decide upon a number of information signals to be sent into the population in order to attract customers. Due to the minimal information provided, the players do not have a well-specified model of their environment. Our main objective is to characterize the adaptive behavior of the players in such a situation.

This paper is concerned with the relationship between firm heterogeneity, different firm strategies, and technological spillovers and learning in an oligopolistic market. In a model of heterogeneous oligopoly firms are technologically different with respect to both the kind of production technique applied (as given by the capital-intensity) and the efficiency by which a specific technique is performed. In pursuing technological progress and building up appropriate technological know-how these firms may apply either a learning or absorptive strategy or contrariwise a conservative strategy that does not attempt to learn from competitors. Within this context the paper investigates the general direction of technological progress in the sense of labor or capital-saving. Applying simulation technique it is shown that the direction of progress changes purely stochastically in the case of conservative firms. Contrariwise, when firms follow the absorptive strategy a quite ordered pattern of development pursued by all firms is observed.

Market share instability, during certain stages of the industry life-cycle, has become a stylized fact in the industrial organization literature. In the finance literature, volatility in the form of excess volatility, i.e. the much larger volatility of stock prices than dividends (although stock prices should in theory trace the present value of future dividends), has given rise to controversies regarding stock price determination (Campbell and Shiller, 1988; Shiller, 1989). Recent evolutionary models, both theoretical and empirical, have tied the presence of market share instability to industry specific variables, such as specific periods in the industry life-cycle and specific “technological regimes”. The object of the paper is to explore whether there is a relationship between market share instability and stock price volatility and to what degree this relationship is connected to the concept of the industry life-cycle, and hence to industry specific factors. To do so, we explore the relationship in one particular industry, the US automobile industry. Since neither life-cycle nor finance theories attack this problem directly, we use insights from both approaches to build hypotheses which guide the data analysis. The empirical results confirm many of these hypotheses, suggesting that the degree of excess volatility is indeed partly affected by industry specific factors.

This paper sheds light on the questions, Why does knowledge spill over? and How does knowledge spill over? The answer to these questions we suggest lies in the incentives confronting scientists to appropriate the expected value of their knowledge considered in the context of their pathdependent career trajectories. In particular, we focus on the ability of scientists to appropriate the value of knowledge embedded in their human capital along with the incentive structure influencing if and how scientists choose to commercialize their knowledge. We conclude that the spillover of knowledge from the source creating it, such as a university, research institute, or industrial corporation, to a new-firm startup facilitates the appropriation of knowledge for the individual scientist(s) but not necessarily for the organization creating that new knowledge in the first place.

This paper reports on the first stages of an empirical investigation into the evolutionary dynamics of the competitive process in the Italian Steel Industry. Central to this investigation is that the mode and tempo of change within well-defined populations of firms are driven by the evolutionary competitive processes of development and selection. The purpose of the paper is to assess the relative importance of these different evolutionary processes to the pattern of change in the Italian Steel Industry. To provide for this, first, a discrete time selection model is introduced. Second, this model is confronted with the empirical data. It is shown that in a mature sector — as the Italian Steel industry — the development term is the major factor driving the decrease in the average unit costs in the population, whilst the selection factor seems to play a major role.

This paper offers an econometric methodology for the detection of self-organisational change (defined in terms of the presence of time irreversibility, structural change and fundamental uncertainty) in economic processes that follow logistic diffusion growth paths in historical time. The approach we adopted is built upon recent developments in ‘moving window’ spectral methods which are applied to the scaled residuals generated by estimated logistic diffusion models. We illustrate the use of such methods by examining the case of a financial instrument, namely, the Australian Building Society Deposit, which experienced logistic growth in its market share until bank deregulation was enacted in the 1980s. We show that there is clear evidence that selforganisational change is present over the historical period considered.

Modelling evolutionary economic processes presents a challenge to traditional econometric methods. One of the authors has proposed an augmented logistic diffusion model to represent growth through a period of self-organisational structural change. In this paper the use of one type of fuzzy regression is explored by applying the technique to a combination of an augmented logistic diffusion model and data that have elsewhere been extensively investigated using spectral methods. In the fuzzy regression approach, the parameters of the logistic diffusion model are estimated as fuzzy numbers. Local parameter corrections are produced for each parameter as part of the estimation procedure. The local parameter corrections are plotted graphically and interpreted, in the light of self-organisation theory, as showing the change in variability of model parameters through time. The results of the fuzzy regression support the conclusion that the data are produced by a self-organising process.

For any evolutionary theory of economic development, the understanding of the determinants of variety and its effects on economic systems is of central importance. On the one hand, increasing returns tend to standardize technologies thus reducing product variety. On the other hand, the resulting efficiency gains generate resources that allow for the development of new products that contribute to long-term growth. The net variety effect on economic systems remains an empirical question. We propose two variety measures, the entropy measure and Weitzman’s maximum likelihood procedure. It is argued that the two measures are complementary since entropy indicates the variety in a frequency distribution, while Weitzman’s measure is based on a distance measure. We apply both measures to data on product characteristics of aircraft (1913–1984) and helicopters (1940–1983). Aircraft variety has been increasing rapidly after the emergence of a dominant design in the thirties. The rising variety of aircraft is related to the increase of the span of performance, which is analogous to the size of the habitat of a biological species. The larger the habitat, the larger the number of niches which is expected to occur. A dominant design is established in each niche. Helicopter variety has been decreasing after a dominant design emerged as innovations hardly in-creased the span of performance. The presence of aircraft technology in the higher end of the helicopter market limited the commercial potential of technological improvements in helicopters.

We reassess Mankiw, Romer and Weil’s [MRW] version of the Solow model using, as did MRW, cross-sectional data to estimate the steady-state equation governing income per capita levels. The model fails in two critical areas. First, plausible factor shares obtained by MRW are not robust to the substitution of two measures of human capital that are more precise than the secondary school enrollment rates used by MRW. Second, the null hypothesis of an exogenous and identical level of technology in all countries is rejected. We also explain why the Solow model performed well despite the above shortcomings.

We find that in a predictable market regime (stable, relative foreign prices) long run growth benefits from slow market adjustments. Fast reallocation creates price instability, erroneous expectations and (easily) cost overshooting. Rapid new entry, however, is always growth promoting in the long run.

The positive scenario for slow market adjustment is completely reversed under an unstable (unpredictable) external market regime, when flexibility in the production system is needed to reallocate resources smoothly and without cost escalation.

Different methods of econometrically representing and estimating the magnitudes involved in non-linear evolutionary models are presented.

This paper examines the interaction between two random processes: (1) a process of technological competition among several mature polluting technologies and a new technology which is environmentally friendly, but the cost of which depends on the adoption rate; (2) a random process of learning about the environmental impact of technologies by the public policy maker who may decide, according to the results, on taxation or prohibition of certain technologies. The ability of the public authorities to ensure the survival of the most environmentally friendly technology is analyzed. The role of various parameters is discussed, especially the interaction between the choice of risk thresholds which determine the intervention of public authorities (i.e. the implementation of the precaution principle) and the policy maker’s learning rate for environmental risks (i.e. the speed with which beliefs in risks regarding the most recent research results are adjusted). Some paradoxical effects arising from the implementation of the precaution principle on the survival of the most environmentally friendly technology are pointed out. The model illustrates problems encountered in practice like the choice of the European Union policy in the face of NO_x emissions from cars.