Are the world’s megacities too big?


Alexis Tostado: Torre Mayor, Ciudad de México, Mexico. 2016 (Unsplash)

by Klaus Desmet and  Esteban Rossi-Hansberg

Republished from

Are the world’s megacities becoming a sprawling, overfed, and uncontrollable mass that needs to be restrained for the good of society and the environment? This column suggests that policies aimed at reducing the dispersion in city sizes will hardly improve the wellbeing of the people who live there. If anything, in some developing countries, such as China, large cities may actually be too small.

The trend in urbanisation is continuing unabated across the globe. According to the UN, by 2025 close to 5 billion people will live in urbanised areas. Many cities, especially in the developing world, are set to explode in size. The Nigerian city of Lagos, for example, is expected to increase its population by 50% to nearly 16 million in the next decade and a half (UN 2010). In recent years there has been much debate about whether restricting the growth of megacities will improve the quality of life (see for example the recent Economist debate here).

But the debate is not so much about whether people should move to cities or stay in the countryside. It is about whether (some) of the world’s megacities have become too large. People flock to cities in search of higher paying jobs and better amenities. Many of the world’s large metropolises, such as Los Angeles and Mumbai, are highly productive and are located next to large bodies of water. As cities grow in size, however, they start suffering from increased congestion, higher crime rates, and air pollution. How fast the benefits of efficiency and amenities erode with population size because of increasing congestion costs depends on the quality of governance, responsible for the provision of road infrastructure, sewage systems, clean water, and security.

Following this argument, if New York is bigger than Williamsport, it must be better in terms of efficiency, amenities, or governance. More generally, the huge differences in city sizes must reflect huge differences in these three basic characteristics. Putting it differently, if cities were to become more similar in terms of efficiency, amenities, and governance, they would probably become more equal in size too. The world’s megacities would likely become smaller, and maybe quality of life would improve.

– But how much better, if at all, would life become?

– And how much would the spatial distribution of population change?

In recent research (Desmet and Rossi-Hansberg 2010), we propose answers to these questions. After estimating efficiency, amenities, and governance for US metropolitan areas, we ask what would happen to the city size distribution if all locations had the same levels of efficiency (or amenities). As expected, we find large reallocations of population. For example, the metropolitan area of Los Angeles would lose 29% of its population if it had average efficiency. The corresponding figures for the metropolitan areas of New York and Chicago would be 77% and 46%. Maybe surprisingly, the cities that would gain population are not the smallest ones, but the intermediate-sized ones. Many of the smallest cities thank their existence to a particular advantage. Take it away, and the city essentially disappears. One such example is Santa Fe, known for its scenic beauty. If it had average amenities, it would lose 82% of its population.

Going beyond the experience of individual cities, many regions of the West Coast and Florida would lose out if all metropolitan areas had average amenities. This makes sense. As argued by Rappaport and Sachs (2003), the concentration of population in coastal areas has to do increasingly with the quality of life. If, instead, we eliminated efficiency differences, central regions would lose population, and so would much of the Northeast. If all cities would have the same quality of governance, most of the Midwest and the Northeast (the region that includes the Rust Belt) would gain population, suggesting that some of the problems of those regions are related to governance issues, which could include labour market frictions and unionisation. For example, with average levels of governance, the population of Rochester would increase by 37%.

But would people really be better off if New York and Los Angeles became smaller? Surprisingly, in spite of the large reallocations of people, it turns out that the welfare effects of a more equal distribution of city characteristics are negligible. Endowing all cities with the same level of efficiency (or the same levels of amenities or governance) would never change real income by more than a couple of percentage points. If, say, a city experiences a negative productivity shock, the effect on wellbeing is mitigated because people will work less (and enjoy more leisure) and congestion costs drop (because some people move out).

Of course these types of thought experiments are hypothetical – even if we tried, cities will never all have access to the same amenities. A more realistic exercise would be to take the inherent characteristics of cities as given, and to determine how much better off people would be if resources were optimally allocated across space. Indeed, given that part of a city’s productivity and amenities is inherent but that another part depends on a city’s size through externalities, we cannot be sure that the observed city size distribution is the socially optimal one. In line with the limited effects of the previous thought experiments, moving to the optimal city size distribution would produce a meagre welfare gain of just 0.6%. And although the effect on the shape of the size distribution is limited, the optimal size distribution would call for the larger cities to become larger and the smaller ones to become smaller. Capping the size of cities does not seem to be warranted.

Beyond the US: The case of China 

The small welfare effects suggest that policies to relocate people across space with the aim of improving the overall quality of life are pointless, at least in the US. But what about in other countries? A similar exercise for China reveals important differences. For example, if all Chinese cities had the same level of efficiency, welfare would increase by 47%, and if all had the same level of amenities, welfare would rise by 13%. These figures are an order of magnitude larger than in the US, where the corresponding numbers are 2.5% and 2.3%.

One would expect a more equal distribution of efficiency or amenities to lead to a more equal distribution of city sizes. But quite the opposite happens. If all Chinese cities had the same level of efficiency or amenities, the city size distribution would become more dispersed, with the larger cities being larger and the smaller cities being smaller. In the case of amenities, this is easy to understand; in contrast to the US, the larger cities in China have on average worse amenities. Giving them average amenities therefore allows them to further grow in size. In the case of efficiency, the largest cities would lose if they had the average level of efficiency, but some of the intermediate-sized cities, endowed with good amenities, would grow so much that they would become larger than the largest cities today.

In other words, any policy that would reduce differences in efficiency or amenities across Chinese cities, would not only dramatically improve welfare, it would lead to more dispersion in the city size distribution. This confirms earlier findings by Au and Henderson (2006) who argue that Chinese cities are too small.

In sum, far from megacities disappearing, we may need more of them, and they may need to be even larger.

Piece originally published at Vox under copyright.


Au, C-C and JV Henderson (2006), “Are Chinese Cities Too Small?”, Review of Economic Studies, 73:549-576.

Desmet, K and E Rossi-Hansberg (2010), “Urban Accounting and Welfare”, CEPR Discussion Paper 8168.

Rappaport, J and JD Sachs (2003), “The United States as a Coastal Nation”, Journal of Economic Growth, 8:5-46.

UN-Habitat (2010), The State of African Cities

About the Authors:

Klaus Desmet is Professor of Economics at Universidad Carlos III, Madrid and is a CEPR Research Affiliate. Esteban Rossi-Hansberg is Professor of Economics and International Affairs at Princeton University.