Anmol Agarwal and Suchika Chopra 10 June 2019
India’s Economic Survey 2016-17 estimates the average annual migrants in India at a staggering 9 million since 2011, much higher than the 3.3 million figure given by multiple Censuses. While results from international studies like International Organisation for Migration suggest that the rate of internal migration in India is fairly low, the economic survey findings based on railway passenger traffic data suggest that absolute number of migrants are much larger than commonly perceived. Most of the recent discussion on migration worldwide, is focussed on the economic impact on regions facing high inflow of migrants. However, understanding the motivations and circumstances under which people migrate is equally important and this study highlights some crucial aspects regarding the same.
We use the India Human Development Survey data1 (2011-12 round) to analyse the push factors behind migration. In the IHDS data, about 90% of the migrants from 2007-11 come from rural areas and more than two-thirds move to urban areas. Women only make up for 10% of the total migrants and hence we focus exclusively on men. We further restrict the age group to 20-50 years, which potentially eliminates migration for studies, thereby allowing us to concentrate specifically on work-based out migration. Work based migration usually result from getting trapped in a vicious cycle, where lack of initial or ancestral resources results in lack of accumulated skills through inferior education, which further exacerbates earning potential. In order to understand the ‘why’ of migration, we focus on the traditionally referred to as ‘push factors’.
We start by comparing the education levels of migrants and non-migrants in the data. Migrant refers to a person who has migrated at least once from their district in the past 5 years (2007-2011). Education level is characterised by the number of years of schooling (for instance, 12 stands for finishing standard XII). Using data for 375 districts2, we match the average education level of migrants in a particular district to the education level of non-migrants in the same district. Same district matching allows us to compare people living under similar conditions and better understand the marginal gains and losses from migration. Using national averages in a heterogeneous country like India, can give dubious findings. For instance, a great deal of insights are lost when a migrant from Rajasthan is compared with a non-migrant from Kerala. Figure 1 shows the kernel-density plot of the difference in education levels between migrants and non-migrants. About 75% of the density curve lies to the left of 0 showing a high probability of non-migrants being better educated than the migrants in given district.
Figure 1. Higher education level of non-migrants (v/s migrants) in the same district
Often an important concern for migrants undertaking the costs of migration is the prospects of better future income flows. Figure 2. tries to discern the income of non-migrants from out migrants. To compare income, we again use district matching between migrants and non-migrants while controlling for differences in education. For instance, income of a migrant holding a bachelor’s degree is compared with income of non-migrants holding a bachelor’s degree in the same district. Figure 2. elucidates that the median income of non-migrants is higher and the income gap, while being negligible at low levels of education, widens considerably at higher levels. It reflects the fact that highly educated migrants struggle to gain employment commensurate with their skills, but as we’ll see later, they are also the ones more likely to linger on in the migrated regions.
Figure 2. Higher income of non-migrants (v/s migrants) at different education levels
Push factors often force people to move out of their domestic territory, sometimes against their will. Perpetrators of this phenomena can be shocks such as an income shock due to scanty rainfall or harvest, or social-economic beliefs in a region – usually developed gradually over time.
An interesting factor is the extent of confidence people put in the institutions in a region. In the face of lack of confidence in public institutions—judiciary, banks, etc. – people may perceive the prospects of opportunities to be bleak and hence migrate out to seek better opportunities. In figure 3, data delineates the aforementioned. On the y-axis, the proportion of migrants is the ratio of total out migrants in the past 5 years in a district to the total number of people (males aged 20-50) surveyed (minimum threshold for inclusions is 100) in the district. The IHDS-II surveys report peoples’ level of confidence in 12 institutions (politicians, military, police, state government, news media, panchayats, government schools, private schools, government hospitals, private hospitals, courts and banks) on a scale of 1 to 3, where 1 reflects highest confidence and 3 the lowest. We take the average of confidence in these 12 institutions as our measure of confidence. The plot shows that districts with lower level of confidence in its institutions have higher proportion of people migrating.
Figure 3. Positive correlation between confidence (lack of) in institutions and proportions of migrants (bin scatter for 183 districts)
Figure 4. Positive correlation between income inequality and proportion of migrants (bin scatter for 172 districts)Source:
While the Economic Survey 2016-17 shows a positive relation between out-migration and real income per capita, we focus on economic inequality as a factor motivating migration. To measure economic inequality, we use the 20-20 rule – ratio of average income of the top 20% of earners to the bottom 20% in a region – as used by the United Nations Development Program (UNDP). A positive correlation between migrant proportion and economic inequality in a district, obtained in figure 4, indicates that people desire to live in more equitable societies and move out of the districts with considerable income inequality.
Since majority of the migrants belong to rural areas (90%), it is likely that they depend on agriculture for their livelihood, directly or indirectly (as workers on others’ farms). Consequently, rainfall plays a major role in their decision-making process. A bad monsoon can engender severe stress and coerce people to migrate. For examining this aspect, we use the district level rainfall data in 2010 from Indian Meteorological Department. To avoid the cases of extremely low and high rainfall, top and bottom 5% districts are excluded. Figure 5 reiterates the conjecture that the proportion of migrants (considering migration 1 year prior to survey here) in districts with lesser rainfall is higher than those where rainfall was sufficiently high. The result aligns with some recent work demonstrating drought frequency in the origin region as an important push factor behind migration. (Dallmann and Millock).
Figure 5. Negative correlation between annual rainfall in 2010 and proportion of migrants (bin scatter for 170 districts)
Source: IHDS-II, IMD
It is worth noting that the duration of migration shows considerable heterogeneity across states. Figure 6 alludes to the heterogeneity in average number of months people migrate for in each cycle of migration across Indian states. Migrants from most of the western part of the country are relatively short term (less than 6 months) as compared to the eastern regions. Short-term migration may be due to both demand and supply-side factors. On the demand side, people migrate with a premediated brief-term view. These are mostly seasonal migrants opting to migrate during specific months of the year when work in home district is not particularly rewarding—a common phenomenon for households dependent on agriculture. Farmers who must leave their land fallow due to lack of irrigation facilities or other structural barriers aren’t left with much choice but to migrate temporarily. On the supply side, failure in finding remunerative employment, net of the cost involved in moving (especially for people with lower education level) forces people to return in a few months’ time. Figure 7 lends credibility to the intuition—the average duration of migration is positively correlated with the education level of migrants. People with better education are more likely to forge social networks crucial to get jobs and job-related information. Moreover, their motivation to migrate may be tethered to the signal that their education is—poised to improve their long-term economic prospects.
Figure 6. Heat map showing the heterogeneity in average duration of migration per cycle amongst Indian states
Figure 7. Positive correlation between education level and average duration of migration per cycle (bin scatter for 2569 individuals)
The analysis brings to the fore some characteristics of migrants and the push factors motivating their decision to migrate. By focussing on males in the age group 20-50 and employing same district matching, we see that people prone to migration have, on average, lesser years of schooling and earn lesser income. Factors like confidence in domestic institutions, income inequality and rainfall shocks, certainly, clout people’s decision to migrate with added nuances of heterogeneity in the duration of migration across states. Individual specific factors also propel people to migrate. The degree of risk aversion among people can play an important role in influencing peoples’ decision irrespective of the condition of regions they live in. For example, migration for extremely poor is akin to a highly risky gamble – given their meagre endowments and fixed cost of moving – which they may prefer to not take[AA1] [sc2] . Therefore, a formal statistical analysis to check such biases is warranted.
Overall, in order to better understand the consequences of large ‘in migration’on the economic conditions of a region, as has been the major policy concern recently in global discussions, it is imperative that the impulses and incentives of migrants are better understood.
- The India Human Development Survey (IHDS) is a nationally representative, multi-topic survey of 41,554 households in 1503 villages and 971 urban neighbourhoods across India, organized by University of Maryland and National Council of Applied Economic Research (NCAER). The first and second rounds of interviews were completed in 2004-5 and 2011-12 respectively.
- Data for Jammu & Kashmir has been excluded as migration there can result from frequent riots, not suitable for this study.
[AA1]Don’t understand this
[sc2]It just means that relative fixed cost for poor is more as compared to someone say lower middle class. A poor may have to sell their house altogether just to get money to be able to move.