The idea of a missing middle in India typically refers to the relative paucity of mid-sized firm. India has many very small firms, and a few very big ones (by Indian standards), but the small ones tend not to grow. This characteristic raises a host of questions about the efficacy of government policy and policy thinking, since it is the growth of firms that creates new jobs, and new jobs are sorely needed in India. But there is also a new angle on the missing middle, which looks at how economic activity is geographically distributed. Here, the missing middle is the lack of growth in medium density clusters of economic activity. The big cities and now the villages (especially those connected by new roads) may be doing well—though starting from different bases—but the challenge for future growth will lie in India’s towns and smaller cities.
Klaus Desmet, Ejaz Ghani, Stephen O’Connell, and Esteban Rossi-Hansberg have recently analysed how employment density in India in the year 2001 was related to employment growth over the subsequent five years. Employment density is closely related to city size, with the big metros having the highest densities. In the case of manufacturing, the pattern they found suggested that subsequent growth was, on average, higher the lower the initial employment density. This pattern is indicative of employment spreading out to smaller cities and towns (with lower initial densities). It is what we would observe if the disadvantages of increased congestion outweigh the advantages of clustering in a location that already has infrastructure, both physical and social.
However, manufacturing in India is a small, anaemic sector of the economy. Fixing its problems relates to the first kind of missing middle, that of the size distribution of firms. The country’s growth has relied heavily on the services sector. Here, the evidence from India is different. In services, initial employment density—that is either low or high—is associated with subsequent employment growth, but this is not so much the case for the medium-density clusters. This result holds for informal as well as formal sector firms, and contrasts with the pattern for