21st Century Science Initiative Grant: Studying Complex Systems

The need for an integrated scientific understanding of urbanization

Urbanization is the most widespread and systematic social development happening worldwide today. According to the United Nations, sometime in the last two years humanity crossed an historic threshold with more than half of the world’s population now living in cities. China and India, the largest among other fast developing nations, are less than 50% urbanized, but changing at a breakneck pace. Worldwide, some three billion people are expected to migrate to cities in the next few decades. In the United States and other developed nations the fraction of urban population is still increasing, with poorly understood implications for patterns of land use, human behavior, resource consumption and economic development.

Can a highly urbanized world be sustainable? Will it be stable, or more susceptible to natural disasters and social instability? Can we understand the forces that lead so many people across the globe to move to cities? Can we predict the consequences of such monumental and unprecedented social change?

These and similar questions have remained difficult scientific puzzles. As a consequence, public policy targeting urban development has had a checkered history, reporting few successes even where substantial investments have been made, such as in attempts to revive cities of the American rust belt.

Urbanization has always been connected with social and economic development. Historically, periods of cultural flourishing and economic vitality – from classical times, the renaissance, the industrial revolution or today – have been times of rapid urbanization. Conversely, economic decline and social collapse, for example at the end of the Roman Empire or the Mayan civilization, have coincided with the fall of cities and the loss of large-scale social and political organization.

Despite much historical evidence, the suggestive links between social and economic development and urbanization have remained mostly qualitative. Studies of cities, whether in economics, geography or the social sciences, have until recently failed to engage seriously with a growing amount of large scale empirical evidence and with quantitative methodologies able to reveal the general organizational forms and dynamical mechanisms that underpin urban development.

This situation has now changed. The advent of large quality datasets for the largest urban systems around the world (USA, China, EU, Japan, India, Brazil, etc) and efforts by national and international organizations to collect quantitative indicators of urban development makes possible, for the first time, quantitative and predictive analyses of the nature of cities as social organizations and engines of development. Methodologies from complex systems are also starting to reveal general trends in these data, suggesting the existence of fundamental general principles underlying urban social organization and dynamics.

A rich but dispersed urban phenomenology

The study of cites has traditionally been subsumed and compartmentalized among a variety of domains. A number of academic disciplines, including sociology and economics, were born out of the study of cities and the impact of urbanization on social life. These disciplines developed independently of each other, and as a result created different empirical traditions and conceptual frameworks to explain human development and social dynamics.

Nevertheless, a picture of large-scale social organization in cities has begun to emerge. More than a century ago, during the industrial revolution in the West, rapid urbanization inspired a number of influential qualitative theories of economic and social behavior. For example, social dynamics within cities motivated the work of influential thinkers such as Adam Smith, Thomas Malthus, Herbert Spencer, Ferdinand Tönnies, Emile Durkheim, Alfred Marshall and others. In the United States, at the turn of the 20th century, sociologists working out of the “Chicago School” of human ecology also produced a spate of literature drawing parallels between population growth, neighborhood turnover, and social mobility in American cities.

Several key insights into economic and social organization were conceived by these studies and remain the foundations of urban theory today. First, it was observed that several social adaptations take place whenever people move from the country to the city. New city dwellers come to depend on more abstract forms of governance and conflict resolution and forego subsistence activities in favor of greater economic and social interdependence. An important economic consequence of these changes is a systematically increasing division of labor as cities grow. Conceptualizations of urban society through biological analogies also originated in this classical period, and continue to frame present thinking.

More recent progress in urban economics refined the nature of the advantages of cities for economic production in terms of the identification of increasing returns in productivity to population scale. This phenomenon probably explains why cities exist at all, and provides a mechanism for human development with far reaching consequences. Specifically, increasing returns to scale means that economic productivity and rates of innovation increase on a per capita basis with a city’s population size. The measurement and explanation of this phenomenon from first principles is one of our main motivations. Importantly, this phenomenon goes beyond economic behavior. In sociology the recognition of similar effects, dealing with the disproportionate presence in cities of “unconventional” behaviors, led to the proposal of several possible mechanisms for their origin, specifically Fischer’s “subculture theory,” “social breakdown” scenarios, and “compositional” arguments that emphasize the differential selection of particular types of people into cities. Interestingly these arguments run parallel to mechanisms proposed in urban economics, but deal with very different quantities, viz. crime rates versus economic productivity, for example.

Despite these conceptual advances several practical and methodological stumbling blocks have remained in the way of progress in the direction of integrating knowledge from different disciplines into a comprehensive quantitative theory of cities.

Methodological and empirical challenges

Some of the important phenomena unveiled in economics and sociology also pose important challenges to classical modeling in these disciplines. Fischer identified specific difficulties in sociology, namely: 1) The choice of appropriate units of analysis: including definitions of city and social groupings; 2) Availability of commensurate data in different cultural settings (e.g. nations) and times; and 3) Adequate control for the variation of multiple variables.

In addition, increasing returns to scale open the door to the possibility of dynamical instabilities, in the form of unbounded urban growth and potential collapse. This shifts the study of dynamics of cities away from traditional economic models, which typically rely on equilibrium to define optimal organization and scale. Instead, we have started to show that a deeper empirical understanding of the relationship between innovation, scale and growth is necessary, as well as the infusion of more general mathematical modeling from the study of other complex systems.

All these difficulties are now being overcome. Their careful consideration and resolution has indeed helped establish firm foundations for a new integrated approach to the study of cities, which we detail below.

Analysis of large data sets reveals general principles of social organization and Dynamics

We have initiated preliminary work that not only resolves most of the conceptual and methodological issues that have stumped previous approaches, but also established general properties of urban organization and dynamics, which we have shown to hold across different urban systems and at different times. First, we have now demonstrated that cities must be taken as unified economic and social units. It does not suffice to adopt more or less arbitrary administrative definitions of cities because choosing the wrong units of analysis destroys important statistical urban regularities. Economists and demographers have struggled for some time with this issue and have come up with appropriate definitions of cities as integrated labor markets, in a way analogous to concepts of mixing population in population dynamics and epidemiology. These are called Metropolitan Statistical Areas (MSAs) in the USA, and equivalent concepts exist in the European Union, China and so on.

Second, we have now collected and started to analyze large data sets data covering analogous social, economic and infrastructural quantities in urban systems worldwide and over time, including the USA, the European Union as a whole and in terms of specific nations, and China, Japan and Brazil.

Third, these data cover many distinct types of urban metrics including measures of economic productivity and wealth creation (GDP, wages, income), innovation (patents, patterns of employment), crime, public health, infrastructure (road surface, electrical grid, etc), energy and resource consumption, CO2 emissions, etc. We have also developed analytical methodologies to disentangle multivariate dependencies. The study of multivariate statistical patterns characterizing cities requires accounting for the covariation of urban properties with population size, an issue we address through scaling analysis, a well-known methodology in complex systems.

Forth, we have adopted and developed a set of quantitative methods from the study of complex systems, non-linear dynamics and statistical mechanics that is able to identify quantitative regularities in these large datasets, formalize quantitative and predictive hypothesis and build new theory for cities based on underlying general principles. Having overcome these initial challenges and with new data and methodologies in hand we are now ready to establish a new qualitative and predictive understanding of cities. Progress in this direction has never been more needed. It will help guide economic development towards sustainability and reveal new insights on our social human nature.