The mass exodus of distressed migrant workers from Indian cities after the nationwide Covid-19 lockdown has got everyone talking about this visible yet unnoticed segment of our population. Over the past month, we have seen several mitigation measures from the Union and state governments to help migrant workers to stay put in their respective destinations. However, due to the high economic cost of the lockdown, the government has started slowly easing restrictions in a phase-wise manner, allowing stranded migrant workers to return to their home-states.
Since a lot of such movement will be a point-to-point occurrence, this initiative requires proper collaboration among states to precisely map the migration networks, in order to facilitate not only massive logistics requirements, but also to plan the quarantine and testing procedures for migrant workers who have returned to their source areas. This is important as the majority of the source states for the inter-state migrant workers are relatively less developed in terms of healthcare infrastructure, and mitigation is necessary to resist further spread of the virus. Along with the National Disaster Management Authority (NDMA), as many as 27 states have started web-portals to register the details of return migrant workers. Coupled with such real-time data, can district-level data from the Census also be useful to map major inter-state migration corridors from the infected districts to source districts? While many of the returnees may be short-term migrants workers not captured by the Census, do they follow the same corridors?
What are the major migration corridors?
We use data from the Census of India and the crowd-sourced site, www.covid19india.org, which provides district-to-district mobility data to derive a source-destination matrix for all the districts in the country and to identify districts highly infected by Covid-19. Though district-to-district mobility data has not been released since 2001, it can still be useful to map the major corridors of inter-state migration, given the deep and consistent networks that sustain them. We have considered the urban areas of the fifteen districts with the maximum number of Covid-19 patients, and these districts accounted for 65 per cent of the total cases in India (As per data from www.covid19india.org taken on 9 May 2020). This list includes some of the major cities in India, like Delhi, Mumbai or Kolkata and is borne out by the fact that 31 per cent of the total and 47 per cent of the urban inter-state migrant workers, during 1992-2001, were accounted for by these districts (including all fifteen largest cities in 2001, except Bangalore, Kanpur, Lucknow and Nagpur. The four cities not in top fifteen but included here are Agra, Thane, Nashik and Jodhpur).
A deep dive into the data reveals that these districts are connected to 244 source districts with high-migrant volumes (>2000 people. Three districts – Agra, Jodhpur and Nashik – didn’t have any source districts with more than 2,000 inter-state migrant workers). Together, these source and destination districts are linked by 471 unique pairs, and these corridors constitute 75 per cent of the urban-ward inter-state migration to the destination districts (3.6 million). These pairs include 53 cases of interlinkages across the highly-infected districts themselves. These movements (9.5 per cent of the total movement across the corridors), nonetheless, require the highest priority in terms of health screening and treatment (these pairs account for 7% of the total inter-state migration of the destination districts).
The volume of migration across the remaining 418 corridors varies widely, along with the distance between the source and the destination districts. Depending upon these two factors, transportation needs to be arranged accordingly, along with planning for controlling diffusion of the disease. For example, a short distance like Delhi – Aligarh could be completed by road travel, and also is likely to see a lot of autonomous travel. On the other hand, a corridor like Mumbai-Jaunpur is likely to see less autonomous travel and more travel by rail, though these assumptions are being increasingly challenged on the ground.
To identify the high volume and long-distance corridors, these 418 pairs are plotted in terms of the distance within the source and destination districts on the x-axis and the share of each pairs to total urban inter-state in-migration in the destination districts on the y-axis, with total volume of migrant workers representing the size of the circle (Fig 1).
For example, the Ganjam-Surat corridor, (returnees from Surat have already resulted in a spurt in Covid-19 cases in Ganjam), is 1196 km long (the straight line distance between Ganjam and Surat), and has recorded 45,760 migrants workers (over 1992-2001), which constituted 11 per cent of the total urban in-migration to Surat district. The 20 long-distance, high share corridors, in the top right quadrant, account for 15 per cent of the total movement across all the 418 corridors. Another 13 per cent is contributed by 42 short-distance but high-share corridors, while 132 long-distance, low-share corridors in the bottom right hand quadrant account for another 20 per cent. More than half of the migration is dispersed, in 224 low-share, short-distance corridors. This last characteristic points to the local hinterland of these high-infection districts, and the need for regional governance structures, which are currently absent in the management of this pandemic. It also shows why a number of migrant workers are trying to independently reach home.
What can be told about source districts and their connectivity?
The analysis also highlights source districts that are exposed to multiple highly infected destination districts. Out of the total 244 source districts, 112 are connected to multiple destinations. Fig 2 classifies these 244 districts into four groups, of which 112 are in groups 2, 3 and 4 and more. Together, these districts consist of 69.3 per cent of the inter-state migration across all the corridors and pose more challenges in terms of risk mitigation from incoming migration. Although some of the highly connected districts include a set of large cities like Mumbai, Kolkata or Bengaluru, most of them are fairly rural in nature, therefore will have limited health infrastructure and non-farm job opportunities. This is more evident in districts with 3 or more connections, i.e. more volume and diverse set of return migration. The urban structure of districts with multiple connections are also based a lot on small and medium towns, with 41.8 per cent of the total urban population living in non-Class I (<1, 00, 000) cities. Once normalcy returns, these smaller urban areas within some of the source districts may provide a source of employment for these returned migrant workers, before they get back to the destination districts.
Need for better data
The Covid-19 pandemic has highlighted the importance of updated and granular information to respond to the migrant worker crisis. While migration data for Census 2011 was partly released in 2019, important details, such as the sectors where migrant workers work, are yet to be released. The district-to-district migration data used here is only available from 2001. In 2017, the inter-ministerial Working Group of Migration report recommended an accelerated time-schedule for the release of data, along with the release of migration data at administrative levels below the districts, in order to identify micro-geographies of movement. Ironically, the Ministry of Home Affairs, which is managing the lockdown, is also responsible for the Census. To begin with, migration corridors emerging from Shramik Special Trains data and state and national NDMA portals can be compared with corridors identified from the Census to understand the links between the direction and magnitude of short-term and longer-term migration. Such analysis may help not only to manage future movements, but also design a more responsive and coordinated welfare architecture.
The authors are senior research associates at the Centre for Policy Research.
Views are personal.