The problem with urbanisation in developing countries

Excellent new paper by Edward Glaeser and Wentao Xiong that summarizes the case for urbanisation for developing countries. 

The authors examine the reasons for the wide income and productivity differentials observed between cities and rural areas, as well as within industries in the same city. They examine whether the income gaps are due to productivity differentials or due to other factors like labour regulations, high housing costs and disamenities in cities, or unobserved worker human capital. Their findings are summarised here. This post though critically looks at the deeper forces driving urbanisation. 

The story about beneficial agglomeration effects goes like this. Firms and workers prefer to concentrate geographically. For firms, the economies of scale and network effects increases the likelihoods of access to suppliers, service providers, partnerships, skilled workers, distribution and marketing, and consumers. For workers, the same effects increase the likelihood of more opportunities, ideas sharing, partnerships, social amenities, similar friendship networks and so on. These locations then become attractive entry points for external technologies and ideas. Finally, there is the positive feedback loop associated with such locational equilibriums - the Mathew effect of agglomeration. 

But are these agglomeration effects real? Or they are because of the confounding effects of unobservable migrant and firm characteristics - the more skilled and enterprising workers and the more dynamic and productive firms (and industries) self-select themselves and locate in cities? Or is the concentration driven by the spatial attributes of the geography - ports, geographic proximities, special zones etc? This is a big research challenge. And it would be a very high value endeavour. 

If it is the latter two, then the agglomeration benefits may only be secondary and the real benefits accrue from having higher quality self-selected workers and firms moving into cities or specific locational advantages of the area. In that case, we may actually be ending up with something more closer to a zero-sum game, or one where the more skilled or competent vote with their feet to certain geographies (urban areas here). This potentially leaves the remaining areas, especially villages, less likely to be able to benefit from exactly the same set of dynamics (enterprising workers and productive firms) that drive the growth of the cities.

The most extreme and disturbing form of this dynamic is the one happening within cities, in the form of gentrification. See this, this, and this. If the current trends play themselves out, it is not unlikely that in the general equilibrium, the city cores will become affordable only for those above the upper middle-class. We would then have massive doughnut sprawls, with deficient infrastructure and poor quality of living even within the urban agglomeration. And, as is happening, if political power gets captured by the rich and (therefore) those who reside in the core, these trends will be accentuated and then urbanisation itself start hitting its limits.

It is for this reason that I am increasingly inclined to the view that housing and transportation are going to be the two most important determinants of the future of urbanisation, and, therefore, national economic growth itself. A time may have come to view gentrification (as we perceive it, a blighted area becoming upmarket) as a negative externality. We may no longer be able to afford leaving this to the dynamics of markets. Public policy may have to play an increasingly important role in moderating the gentrification effects, even with the risk of potential excesses.