Technology as a Driver of Agglomeration
by Diane Coyle
Urbanisation has for centuries been a marker of economic development, while Alfred Marshall provided the basic economic analysis of the forces of agglomeration in his 1890 Principles of Economics. Yet economic research into cities and agglomeration – into the geography of the economy – has revived significantly since the late 1990s, including work by prominent economists such as Ed Glaeser, Paul Krugman and Tony Venables.
For most of the mid-20th century economics largely lived up to its caricature as a discipline analysing the world in terms of atomistic optimising individuals in linear models, and paid decreasing attention to the specifics of history or geography. The profession rewarded the ability to manipulate mathematical models while steadily dropping from the curriculum the requirement to study the world in all its untidy detail. So what was the reason for the 1990s renewal of interest in agglomeration, the spatial distribution of economic activity? Digitalisation was starting to change the dynamics of the economy in the 1990s. When I started writing about the economic and social effects of digital technologies around the same time, it seemed clear to me that the forces driving urbanisation would intensify (although others predicted the opposite effect, the loosening of geographical ties or ‘death of distance’). Marshall’s original explanations for the concentration of activity in the same places – closeness to market, depth of the labour market and proximity to ideas – still stood but the importance of exchanging ideas was growing as the role of high value added services and intangible (‘weightless’) activities in the economy expanded.
In these ideas-driven activities, Michael Polanyi’s tacit knowledge looms large. In any new domain of activity it anyway takes some time for the information needed to operate a new machine or process, say, to become systematic enough to be codified – written down in instructions that someone else can follow – as James Bessen describes in his outstanding book Learning By Doing. This is the situation now with areas such as AI and big data; although the computational and data handling processes are central, operating them is still a craft skill, passed on between practitioners. Moreover, when it comes to ideas-based work in general, it is difficult-to-impossible to pass on know-how without conversation – although the pandemic is an enforced test of whether improved videoconferencing can finally substitute for face-to-face contact (as Richard Baldwin predicts).
As economists rediscovered the importance of place, the importance of history also re-emerged, again prompted by the arrival of the new technologies. History is the source of evidence about the economic impact of the periodic arrival of general-purpose technologies with wide application such as digital or AI, so researchers started looked back to the Industrial Revolution or even the printing press. An influential example is Paul David explicitly comparing the diffusion and productivity effects of electrification and computerisation. More recent work has specifically highlighted the role of another 19th century technology, the steam train, in driving substantial urban agglomeration. And the role of ideas and technology in economic growth gained broader traction through Paul Romer’s endogenous growth theory.
It has taken some time, however, for the full implications of geographic agglomeration to filter through both economic research and particularly economic policy. The role of both historical and geographical context, of path-dependence in economic trajectories, of the dynamics of self-fulfilling processes, stand in contrast to the (mainly) linear and context-free tradition of economics for much of its 20th century practice. This has recently started to change, driven perhaps by growing understanding of digital dynamics (or by the overlap with the dynamics of natural systems in environmental economics) with a new focus in economic research on increasing returns, network effects and tipping points.
Evidence has also underlined the need to take agglomeration seriously. The growth of global cities has been obvious. Patricia Melo finds that productivity drops off with distance from city centres. Influential research by Enrico Moretti and David Autor among others indicates that in the US the big city lead is accelerating: the often-observed occupational and income polarization has a geography.
However, the policy implications of the polarising, snowball-type dynamics of an increasingly digital economy have taken some time to become clear. The first iteration in policy was probably the desire many city authorities had to become a locus for the ‘creative classes’, focusing on amenities and culture, or alternatively their competing bids to have science campuses or other high-skill magnets. A second reaction was the argument that it was pointless to resist the market dynamics, which would self-equilibrate by pushing up prices of housing and creating congestion in the attractive places.
Both have some truth. Amenities and physical facilities can act as magnets. Markets will bring about some adjustments in behaviour. Neither is an adequate approach to policy in contexts where the big city/small town & rural divide is starting to have significant political consequences – for recent voting trends in many countries reflect to some extent the geography of economic division. To make matters even more urgent, the covid-19 pandemic is clearly amplifying existing societal inequalities of all kinds.
Yet there are no simple policy recipes in contexts of hard-to-predict non-linear and path dependent dynamics. History casts a long shadow and points of inflexion depend on the interactions of many variables in a complex system. Changing the dynamics will require the alignment of a number of different policy interventions, just as a key needs to align all the tumblers in a lock before the door will open. One area of policy I have explored, with Marianne Sensier, is the use of cost-benefit analysis (CBA) for the appraisal of public transport investments. This frequently-used (in the UK) policy tool uses place-specific land values and productivity measures to determine whether an investment is worthwhile, resulting in a strong bias to approving them in the already-most productive places. The rationale is the wish to contribute as much as possible to national productivity but of course it plays to the idea that agglomeration is a natural result of the way markets operate and reinforces spatial inequality within the nation. Nor can CBA methods take into account the counterfactual returns to an investment if other policies were put into effect at the same time: housebuilding, investment in amenities, and training alongside an upgraded commuter line, for example. Thinking about policies one by one rather than as a suite aiming to shift a system outcome cannot overcome the powerful technology-driven dynamics.
At a minimum, policymakers need a far more granular understanding of places other than the handful of high skill global cities. In the super-centralised UK, the availability of data at sub-national level has improved dramatically in recent years but we still know too little about the geography of supply chains or skills, although complexity theory and other innovative approaches are providing new insights.
The combination of the ‘levelling up’ policy agenda and behaviour change following the pandemic will surely make prospects outside the big urban agglomerations the focus of policy in the near future. But unless there is a reversal of the historical complementarity between technologies of communication and face-to-face contact, human proximity in major cities will continue to be the engine of economic growth. That means finding a way to make the forces of agglomeration deliver prosperity without polarisation will continue to be the analytical and policy challenge.
Diane Coyle is Professor of Economics and Public Policy at the University of Cambridge.
Other posts from the blogged conference:
Urban Agglomeration, City Size and Productivity: Are Bigger, More Dense Cities Necessarily More Productive? by Ron Martin
The Institutionalization of Regional Science In the Shadow of Economics by Anthony Rebours
Cities and Space: Towards a History of ‘Urban Economics’, by Beatrice Cherrier & Anthony Rebours
Economists in the City: Reconsidering the History of Urban Policy Expertise: An Introduction, by Mike Kenny & Cléo Chassonnery-Zaïgouche