Abstract: Agglomeration economies are a persistent subject of debate in regional science and city planning. Their definition turns on whether or not larger cities are more efficient than smaller ones. Here, we complement existing discussions on agglomeration economies by providing a sensitivity analysis of estimated externalities to the definitions of urban agglomeration. We regress wages versus population and jobs over thousands of different definitions of cities in France, based on an algorithmic aggregation of spatial units. We also search for evidence of larger inequalities in larger cities. This paper therefore focuses on the spatial and economic complexity of the mechanisms defining agglomeration within and between cities.
This great book edited by Jeffrey Johnson, Paul Ormerod, Bridget Rosewell, Andrzej Nowak, and Yi-Cheng Zhang brings together many contributions from an EU project which lead to several workshops and conferences about a new form of social science – out of equilibrium, far from equilibrium, in disequilibrium as the world always is. The book is open access and you can download it here.
Here is an explanation of what is contained within. Between 2011 and 2014 the European Non-Equilibrium Social Science Project (NESS) investigated the place of equilibrium in the social sciences and policy. Orthodox economics is based on an equilibrium view of how the economy functions and does not offer a complete description of how the world operates. However, mainstream economics is not an empty box. Its fundamental insight, that people respond to incentives, may be the only universal law of behaviour in the social sciences. Only economics has used equilibrium as a primary driver of system behaviour, but economics has become much more empirical at the microlevel over the past two decades. This is due to two factors: advances in statistical theory enabling better estimates of policy consequences at the microlevel, and the rise of behavioural economics which looks at how people, firms and governments really do behave in practice. In this context, this chapter briefly reviews the contributions of this book across the social sciences and ends with a discussion of the research themes that act as a roadmap for further research. These include: realistic models of agent behaviour; multilevel systems; policy informatics; narratives and decision making under uncertainty; and validation of agent-based complex systems models.
This coming month (January 2017) sees the launch of another of Nature’s flagship journals Nature Human Behaviour. In it, Andres Gomez-Lievano, Oscar Patterson-Lomba, and Ricardo Hausmann explore an integrated model which explains the prevalence, scaling and variance of urban phenomena in which they suggest a new theory of city size that embodies ideas from economic complexity and cultural evolution. This provides a rich basis for speculating on the economic structure of cities suggesting hints as to how old cities might regenerate their past prosperity and new ones generate more success. They do not dwell on applications for they are most careful to ground their argument in caveats – largely due to the fact that they make many plausible but non-testable assumptions to generate their model. In fact, this is little different from most work in the urban systems (and indeed in economic science) where bold assumptions are always necessary to cut through the richness and noise that plagues our understanding and descriptions of human behaviour.
I have written and News and Views comment on this and you can find it here. In fact, the article that this is a response too, is also available for gratis – Nature’s first issue is essentially downloadable for free – you can get their article by clicking here. What they essentially do in the paper is demonstrate how scaling arises due to the economic complexity of any city. They ground their work in the key result that as cities get bigger, they get more than proportionately more wealthy, implying a scaling law that is consistent with old ideas about urban agglomeration but which was demonstrated nearly 10 years ago for US cities by Bettencourt; but a law with much more ambiguity when it comes to measuring the wealth of older and more polycentric city systems such as those in the UK. To this end, and I quote from my review: “ … they introduce a simple but enticing model which extends these scaling laws to embrace the prevalence of how different activities scale with city size. First, activities will become less prevalent across all cities as they get bigger. Second, they will become relatively more concentrated in bigger cities, on the average. And third, the deviations from this average will widen. Their theory and the model that comes from this suggests that as an activity requires more and more inputs of both a general and specialized nature, it will become less prevalent as a city gets bigger. But as cities get bigger, only then can they access more inputs; so the theory predicts that the prevalence of more complex and specialized activities will also become more concentrated. To an extent, this increase in the numbers of factors or inputs with city size is consistent with their related ideas on economic complexity. A consequence of the way they structure the model is that the variation in the degree of prevalence for a city will increase as the activities get more complex while the total prevalence of the activity gets less.”
There is a lot more than they say in their article and you should read it as it takes a different view of economic systems, technological progress and cultural affinities than anything that has come before. But it is consistent with a good deal of more general thinking about the nature of cities such as that contained in the writings of Jane Jacobs for example. Exciting times for cities and city planners.