Spatial disparities in India: Have Mumbai and Chennai become too congested?

Klaus Desmet, Ejaz Ghani, Stephen D O'Connell, Esteban Rossi-Hansberg 13 June 2012



In the last two decades the Indian economy has been growing at unprecedented rates, but that development has led to widening spatial disparities. While some cities such as Hyderabad have become major high-tech hubs with world-class companies and real estate developments reminiscent of Silicon Valley, many other places remain mired in poverty and stagnation.

Rising geographic disparities are obviously not specific to India. They are common to the development experience of many countries. As documented by the 2009 World Development Report, growth is often accompanied by the increasing spatial concentration of economic activity. In China, for example, the coastal provinces have attracted the lion’s share of the country’s economic growth, and in Brazil much of its economic activity is concentrated in the metropolitan areas of Sao Paulo and Rio de Janeiro. 

Policymakers tend to frown upon the uneven distribution of economic activity, and are keen on promoting policies aimed at reducing such spatial differences. Given the huge congestion in cities such as Mumbai or Kolkata, this seems a reasonable policy in the context of India. However, those cities also benefit from important agglomeration economies so that there is a need to analyse the trade-offs between the costs and benefits of economic density before articulating policy recommendations.

Compared to other countries at similar levels of development, India’s growth stems disproportionately from its burgeoning service sector (Ghani 2010). When focusing on the geographic distribution of economic activity, this difference is relevant. Evidence from the US and Europe suggests that these sectors behave very differently (Desmet and Rossi-Hansberg 2009). In both the US and Europe manufacturing is becoming more equally spread across space, whereas the service sector is becoming increasingly concentrated in medium-sized locations. In those intermediate places agglomeration economies dominate congestion forces, making them attractive locations for service industries to grow and expand.

If the US and Europe provide any guidance, India’s engines of growth should be its medium-density cities. In a recent paper (Desmet et al. 2012), we analyse whether the trade-off between agglomeration economies and congestion is indeed similar in India. This provides valuable insights into what types of regional policy interventions may be useful and effective.

'India is Different'

Using district-level employment data in India for the years 2001-2006, we find that the manufacturing sector exhibits a spatial growth pattern in line with what we see in the US and Europe. Manufacturing is moving away from high-density clusters to areas that are less congested. As a result, over time industrial activity is spreading more equally across space. The same is not true for the service sector. Figure 1 below shows the result of a nonlinear kernel regression of the annual growth in services employment as a function of initial services employment density (including the 95% confidence intervals). As can be seen, although low- and medium-density service locations experience spatial dispersion, the high-density service locations are continuing to attract more employment. That is, high-density hubs are gaining relative to locations with intermediate levels of density.

Figure 1. Services density 2001-06

Source: Authors' calculations based on National Sample Survey

In contrast to the US and Europe, high-density service clusters in India therefore do not seem to be running into decreasing returns. For example, service employment in the well known IT hubs of Hyderabad and Chennai are growing at annual rates of, respectively, 11% and 4%. To make the comparison more precise, agglomeration economies in the US service sector peak at a density of between 50 and 150 employees per square kilometre. Three of the main high-tech counties in the US fall within that range: Santa Clara, Calif. (Silicon Valley), Middlesex, Mass. (Route 128) and Durham, NC (Research Triangle). In India, instead, agglomeration economies continue to be strong in places where service employment density reaches into the thousands. For those levels of density, US locations exhibit decreasing returns, reflecting substantial congestion.

Providing a different angle on the same issue, suppose the relationship between density and growth in the service industry were the same in India as in the US In that case many of the slow-growing medium-density service districts would grow much faster. Growth would become more concentrated in the coastal regions, especially in Southern states such as Tamil Nadu and Kerala, as well as in Northern states such as West Bengal, Bihar and Uttar Pradesh. Of the well known IT clusters in India, the medium-density ones such as Ahmedabad and Pune, and especially Bangalore, would continue to grow at sturdy rates, whereas the higher-density places, such as Chennai and Mumbai, would experience a substantial decline in their growth.

The overall picture that emerges suggests that India’s engines of growth are its high-density service hubs. Maybe the difference with the US and Europe has something to do with India being an emerging economy. This does not seem to be the case though. China, for example, although similar to India in being a fast-growing developing economy, exhibits spatial growth patterns in the service industry more similar to those in the US. Another possibility is that the underlying trade-off between agglomeration economies and congestion forces is fundamentally different in India. But this is hard to defend as we have no strong reason to believe that Indians have less aversion for congestion than Americans. A more likely explanation is that India suffers from certain frictions that put a break on the growth of medium-density locations. It is not immediately obvious what those frictions are. One possibility is India’s notorious lack of infrastructure, which limits the production possibilities and the market access of all but a handful of highly dense clusters. Further exploring what other frictions might contribute to these growth frictions is an important research and policy question.

One way forward is to look at Bangalore, perhaps the best counterexample of the problems faced by medium-sized locations in India. Its density is similar to that of Silicon Valley. Interestingly, its origins go back to the Electronics City of the 1970s, set up as an industrial park 18 kilometres south of the city, with good infrastructures and away from the downtown congestion. This success story suggests that it is possible to eliminate the growth restrictions confronted by many of India’s intermediate-density locations.


Desmet, K and E Rossi-Hansberg (2009), “Spatial Growth and Industry Age”, Journal of Economic Theory, 144:2477-2502.

Desmet, K, E Ghani, S O’Connell, and E Rossi-Hansberg (2012), “The Spatial Development of India”, World Bank Policy Research Paper 6060.

Ghani, E (2010), The Service Revolution in India, Oxford University Press

World Development Report 2009: Reshaping Economic Geography, World Bank, Washington DC.



Topics:  Development Industrial organisation Productivity and Innovation

Tags:  India, manufacturing, services, Mumbai, Chennai


Perhaps one explanation for the authors' findings could be the tenuous link between employment and output in India. India's formal sector, which provides most of the output in both manufacturing and services, exhibits something akin to jobless growth, or at best maintaining employment shares. India's informal sector, meanwhile, exhibits tremendous employment growth but stagnating productivity.
For instance, the decomposition in de Vries, et al (2012; ssrn:1998072) of labor productivity and employment between informal and formal sector manufacturing using national accounts data shows that observed shifts of employment into manufacturing has been mostly into informal firms. Because of poor productivity, this has not been growth enhancing. The few statistics they provide for services suggest a similar pattern.
Is it possible the informal sector, which dominates Indian employment, is exhibiting declining returns, but heedless unskilled workers coninue to be drawn to big cities for other reasons? Meanwhile the formal services sector - more sensitive to agglomeration vs. congestion effects - exhibits a more typical spatial dynamic? Hence employment dynamics that don't make this distinction would not provide much information about sectoral returns or agglomeration vs. congestion effects. Is it possible to distinguish bettween informal and formal service sector employment in the NSSO data?

Klaus Desmet

Professor of Economics, Universidad Carlos III and CEPR Research Affiliate

Ejaz Ghani

Lead Economist in Economic Policy and Debt, PREM Network, World Bank

Stephen D O'Connell

PhD Candidate in the Department of Economics at The Graduate Center, City University of New York

Professor of Economics and International Affairs, Princeton University