Martin Zoričák, Vladimír Gazda, Denis Horváth, Oto Hudec (p)

Danny Czamanski

Several theoretical concepts (e.g. the Marshallian district theory, New Economic Geography) effort to explain why the firms from the same industry have a tendency to locate close to each other and to constitute a population of firms forming an industrial district. The productivity in the industrial districts is enabled by the collective learning in the emerged industrial clusters thanks to the proximity of the economic agents through innovation and imitation. A firm often starts with not only one industry and several activities. Its location is backed by looking for a free location near other firms in the same industry but working in different - complementary activities to maximise the market position.
A simulation model of historical industrial evolution based on Agent-Based modelling and Cellular automata is inspired by the Bak-Sneppen model of biological evolution. The two-dimensional cellular automata (two-dimensional periodic lattice equivalent to the two- dimensional torus) predesignated to demonstrate real-world spatial cooperation and competition among the firms are proposed. As the evolution of industries is modelled, the ancient time (initial conditions) is set in the firm profile by a random initial assignment of the single industry. Later on, the producers gradually began to recognise the advantages of specialisation and cooperation.
Performance of a firm is considered as dependent on the fitness of its industries and activities with its neighbourhood. In the course of evolutionary simulation, the least fitted firms are repeatedly forced to adapt to the changing environment by partial mutations of their profiles. The mechanism of evolution in the critical state can be thought of as an exploratory search for local better fitness, which is rarely successful, but sometimes has enormous effect on the ecosystem. As achieving the self-organised criticality, even a small change in an industrial profile can cause massive waves of firm restructuring resulting in a new spatial pattern reorganisation.
In the long-term, new industrial profiles emerge and firms become self-organised in spatial clusters evolving towards Zipf's rank-size distribution. The proposed simulation model demonstrates its ability to explain appropriately the long-term evolution of industrial economic structures in both time and space. Moreover, spatial modelling, starting from the basic division of labour to the present, discovers what combination of the number of industries and activities is the most advantageous for a firm to survive and to increasing its fitness.

Danny Czamanski, Dani Broitman (p), Itzhak Benenson

John Östh

Endogenous growth in a spatial economy

Dani Broitman, Itzhak Benenson and Daniel Czamanski


The Lucas paradox points out that contrary to the general equilibrium model, capital does not flow from rich to poor economies. We resolve the paradox by focusing on the self-organizing nature of modern economies that leads to prolonged far from equilibrium conditions. We propose a model with local positive feedbacks that tend to possess a multiplicity of asymptotic states or possible emergent structures. We study growth processes and periodic fluctuations that arise endogenously within a system of cities. At the same time our model addresses simultaneously two stylized facts concerning cities. Under certain conditions, our model generates entire life-cycles for individual cities. At the same time, it produces a power law for an assemblage of cities.

We present a dedicated agent-based model that includes several types of agents including, firms that make decisions what to produce and whether to hire additional workers and individuals that make decisions whether to migrate and become entrepreneurs as an alternative to working for existing firms. Migration generates innovative activities. Newcomers have better chances to boost innovation, and therefore a firm receiving new employees have a higher chance to realize the next big idea. However, innovation also decays with time. Based on existing human capital within an economy, the adjacent possible is the set of all entities that can be produced within an evolutionary system in each successive step. It is the set of ideas that are one step away from ideas that actually exist and generate incremental modifications and re-combinations of the existing ideas. Thus, development is driven by entrepreneurial actions that introduce technological innovations creating new products, services and production processes. The characteristic non-linear, combinatorial interactions can cascade and reach magnitudes causing industrial transformations, termed by Schumpeter creative destruction. In the absence of innovations, the economy will slip into a stationary equilibrium state.

Paul Kilgarriff, Remi Lemoy (p), Geoffrey Caruso

Oto Hudec

Seen from a satellite, whether looking at land use in the daytime or at night lights, most cities have rather circular shapes, organised around a city centre. As a consequence, the distance to the centre is the first spatial differentiation to consider when studying the internal structure of cities. We conduct here a radial analysis of urban land change in order to understand what the recent changes in urbanisation are across Europe and how it relates to city size. We focus on the fundamental differentiation regarding urban land use: is it natural, like most land on the planet, or has it been artificialised for human uses. Using spatially detailed data from the EU Copernicus Urban Atlas, the profiles of artificial land use (ALU) are calculated and compared between two years, 2006 and 2012. Based on the homothety of urban forms found by Lemoy and Caruso (2018), a simple scaling law is used to compare cities after controlling for population and allows for the internal structure of cities, as determined by distance to the city centre, to be compared across years. One of the advantages of using a radial analysis, is the ability to examine the complex two-dimensional intra-urban structure of a city in a one-dimensional space. The land use profiles produced from the radial analysis represent ALU with respect to distance to the centre. We present evidence of tilting profiles of artificial land use which shows that given total population growth, urbanisation is relatively shrinking up to a rescaled distance of ~ 20km (using London as a reference) on average across Europe between 2006 and 2012. This contrasts with further expansion and increase in artificial land use, beyond a rescaled distance of ~ 20km. Grouping cities based on population, highlights that ALU in the largest cities (population > 1.5 million) is on average flattening around the core but upward sloping at distances around the periphery. Cities with a population below 100,000 are upward sloping on average across all distances to the CBD. We explore these changes focusing on similarities between cities by disentangling the role of city size. Our findings have important implications relative to the sustainability of cities as this evidence is pointing to increasing urban sprawl and stagnant growth in urban centres across cities of all sizes. It also bears theoretical implications on the nature of sprawl and its scaling with city size.

John Östh (p), Aura Reggiani Reggiani , Laurie Schintler

Itzhak Benenson

The Central Place Theory (CPT) originates from the works of Walter Christaller (1933), in which he sets up a set of rules in terms of population threshold and commuting range for the provision of any specific service such as schooling or the running of a hospital. The model became very influential and guided the regional delineation of Estonia in the 1930ies and Sweden during the post-war years. In this study, we revisit the underlying principles of population threshold and commuting range as used in the CPT by comparing the delineation of municipalities (designed to follow the CPT) to the functional distribution and use of mobile phones. We follow (for 24hours at a time, then phone IDs are renewed) the diurnal mobility patterns of phones over the course of a week. We have no information about the phone user, but we have geocoded data that can be used to link user to likely socio-economic and demographic belonging on the basis of the night-rest locations of phones. By observing duration and timing of trips, and the socio-economic and demographic characteristics of the area of origin, trips are categorized into commuting, service trips, long-distance travel and other trips. By comparing trip-characteristics and flow-direction to delineations of regions, municipalities, and city parts we can compare to what extent observed phone mobility replicate the geography of the functional regions on different hierarchical levels.