What money couldn’t buy: social protection for migrants in India’s lockdown

ABSTRACT We analyze findings from a large-scale survey of over 11,000 respondents across 64 districts in India, conducted between December 2020 and January 2021 to examine the impact of the lockdown on internal migrants in India. We find that compared to the households without migrants, households with migrants were relatively advantaged in income levels before the pandemic but faced more severe food and financial vulnerability even nine months after the first lockdown. In addition, governmental social security support was more difficult to access for households with migrants. The paper joins several scholars in arguing for greater policy attention and social protection for migrants.


Introduction/Background
Even by more conservative central government estimates, 11.4 million migrant workers returned to their homes during the pandemic in India (The Hindu Business Line 2021). 1  Migrants were left jobless overnight and, in the absence of any means of transport available, often walked several hundred kilometers over many days to reach back to their villages.Data on a complex phenomenon like migration will always fall short of being able to capture reality comprehensively and accurately.However, the government's acknowledgment that it did not have any data on deaths or job losses of migrant workers (Nath 2020)a phenomenon if not visible earlier, became starkly visibilized during the pandemicreflects the acuteness of the shortage. 2 In the absence of any official narrative, the story of an Indian Migrant worker needed to be pieced together by data collected by private entities.For example, through the data from the Centre for Monitoring Indian Economy (CMIE), the Center for Sustainable Employment at Azim Premji University (Kesar et al. 2021) and Jan Sahas, among several others.Adding to this picture, we use data collected by Rapid Rural Community Response to Covid 19 3 (RCRC) to compare the experiences of "migrant households" (households including at least one member who was seen by the household as a migrant) and non-migrant households.In addition to other surveys it has done (and continues to do), the collective collected data from nearly 12,000 households around nine months after the first COVID-19-induced national lockdown imposed in March 2020, providing a rich data source focusing exclusively on some of the most distressed geographies in rural India.
Internal migration is thought to be an important cog in the development story, not only in India but also in other countries like Tanzania (De Weerdt and Hirvonen 2016), China (World Bank 2014) and others (Bell et al. 2015).Even back in 2009, United Nations Development Program reported that three-fourths of the estimated billion migrants worldwide are internal migrants.The Covid story of India narrates the underbelly of this development (Department of Economic and Social Affairs, United Nations 2011).Rao et al. (2020) describe "Migration" as mimicking a social protection strategy, but one that has been provided privately and independently of the state.While migration is often dependent on social networks and relations, policies still see migration as an outcome of personal initiative, drive and risk-taking ability.For example, the draft National Policy on Migrant Workers proposed by the NITI Aayog, India's apex public policy think tank, says that the efforts need to "enhances the agency and capability of the community and thereby remove aspects that come in the way of an individual's own natural ability to thrive" (Mehrotra 2021).While the economic value of migration, the fact that it's a self-initiated and reliant activity, is argued to provide a sense of pride and independence to economic migrants; the rural-to-urban migration in India is largely short-term and remains distress-driven (Mishra 2020;Srivastava 2020).
In forcing a shared experience of being "locked down" without access to market-based instruments of income and services, the pandemic coalesced this otherwise diverse and dispersed group of citizens in their needs and neglect.There was a disproportionate burden of the pandemic-induced shutdowns on the poor, especially on migrants, daily wage earners and the self-employed (Buheji et al. 2020;Gupta et al. 2021).The migrant population has always largely been disadvantaged, lacking influence, social networks and policy protections.However, the pandemic has further reinforced their position as "clients" in their cities and not "citizens with rights" and thus the need to recognize them as a "distinct constituency" has emerged even more strongly (Rao et al. 2020).With migrants taking the streets attempting to reach homewhile others remained off itthe pandemic visibilized the group as clearly distinct from the rest with shared vulnerabilities.
Despite reporting being better off on standard economic indicators like income before the pandemic, we find that the economic advantage did not necessarily enhance the ability to deal with the shock of the pandemic and concomitant lockdown.Migrant households were more likely to report losing jobs in this period, but they were also less likely to be able to absorb the shocks of losing them.Migrant households report experiencing worse development or welfare outcomes on several dimensions including health, hunger and mental well-being, and find themselves in greater financial stress and indebtedness, raising questions about their future well-being.In addition to these individual-level adversities, and possibly contributing to them, migrant households were less likely to avail of government-provided social protection programs despite returning to their "source" villages and found themselves less likely to receive help from social institutions and government functionaries.
Our results not only provide support for demands for greater urgency for the governments to recognize the vulnerabilities facing migrant workers but also raise questions about the sufficiency of policy instruments like cash transfers to provide social protection during times of distress.Not only did money fail in buying them better outcomes, but it also failed the migrant workers in protecting them from worse ones.While there are several narratives that point in this direction (Breman 2020;Deshingkar 2022;Rajan and Bhagat 2022;Srivastava 2020), we supplement these narratives with data that allows us to compare the relative experience of households with migrants to those without them.Our results suggest a paradox and reinforce the need to not only look at migrants as a distinct group but also the necessity of universal social protection responses that are inclusive of migrants, not exclusive to them.
Even as we acknowledge the limitations of our work in capturing the complexity of these experiences, we contribute to the literature by drawing on data from some of the most disadvantaged communities in India.
The rest of the paper is organized as follows: we first introduce the data, provide sample background characteristics and how the sample compares to the average population.To situate our sample in the context of other studies that have also studied the rural migrant experience during the pandemic.We then discuss the findings in four broad areas: incomes and livelihood, health and nutrition, financial liquidity and the demand for institutional support.

Data
We use data collected by Rapid Rural Community Response to Covid 19 4 (RCRC), a collective of civil society organizations, in December 2020-January 2021, a period about nine months after the first COVID-19-induced national lockdown imposed in March 2020. 5 RCRC collaborated with 43 organizations to collect data on food security, livelihood, access to government schemes and services, and several other topics across 11 states.The collaboration resulted in a diverse sample of 11766 respondents across 128 Blocks from 64 districts from Assam, Bihar, Chhattisgarh, Gujarat, Jharkhand, Madhya Pradesh, Maharashtra, Odisha, Rajasthan, Telangana and Uttar Pradesh.The break-up of the sample across states can be seen in Table 1.

CJDS / LA REVUE
Without any fixed total sample size, organizations were requested to survey at least 200 households across two blocks -20 random households in at least 5 villages in each block (i.e. 100 households in each block).Even though the median organization surveyed 200 households across 2 blocks, the process followed may have varied and sample size for organizations ranged from 60 households surveyed in a single block to over 2000 households surveyed across 15 blocks in 4 districts, depending on their resources and capacity.
All organizations and their enumerators were trained on the questionnaire and data collection process by the central research team to conduct in-person surveys with mobile phones and tablets.Since the efforts from the organizations were voluntary, it meant that they deployed enumerators that had prior experience and in-person access to disadvantaged households which was otherwise difficult during that period.Thus, this resulted in the sample primarily comprising "beneficiary" households.About 64% of households were those that the NGOs were "currently working with," about 13% of households had a previous association with the NGO and 23% of households had no association with the NGO.
While these results of the survey may not be representative at the block/district/state level, this effort represents one of the largest surveys conducted during that time and is likely to be indicative of the experiences of vulnerable households in rural India.This is especially due to the capacity and embeddedness of surveying organizations among poor and vulnerable communities in their respective geographies.
Thus, unsurprisingly, the sample consists of relatively poorer households than the average population.More than 55% of households reported having pre-pandemic incomes less than Rs.5000 per month (2019 incomes).Almost 85% of the households had monthly incomes of less than Rs.10,000 per month. 6The share of Scheduled Caste (SC) and Scheduled Tribe (ST) 7 households, which were likely to be more vulnerable, comprised more than 50%.The share of ST households exceeded 30%more than three times their average share in the population. 8See Table 2 for more details about the sample.
A notable share of households (about 25%) had individuals who were migrating outside their village to work. 9This is the primary category of interest for us since the focus of our paper is to compare the experiences of households with migrant workers to those without on several livelihood and well-being outcomes.In our analysis, we first compare these two groups of households on demographic and socioeconomic indicators and then explore the differences in experiences and outcomes.We present descriptive differences in outcomes and then account for demographic, geographic and socioeconomic characteristics using a multivariate linear regression model.Each outcomea binary variable such as whether any member in the household lost their job or incomeis the dependent variable.These are regressed on whether the household has any migrant workers (the main variable and coefficient of interest), along with other characteristics such as their social category, income category, which state they reside in, and whether the district they reside in is part of the government's Aspirational District Program. 10 The standard errors have been clustered at the block level. 11

Findings
Before we present the differences in outcomes across households with migrants compared to those without, we first report on how they differ on background characteristics in Table 3.  Source: Authors' calculations using primary data collected by RCRC.

CJDS / LA REVUE
We find that households with migrants that were surveyed were relatively more advantaged on some "baseline" indicators compared to households without a migrant.While decisions to migrate are influenced by a multitude of factors and remain a complex phenomenon, this finding is in line with previous evidence suggesting that relatively wealthier households (within caste groups) may have a higher propensity to migrate (Munshi and Rosenzweig 2016).In our sample, migrant households had higher pre-pandemic incomes, were more likely to report having health insurance coverage, 12 fewer belonged to the ST category, and were slightly more likely to have a smartphone in the household.They were also more likely to be receiving support or being beneficiaries of the NGO surveying them at the time of the survey.
Additionally, among households with migrants, almost 75% reported that the migrating members had returned back to the village during or right after the first lockdown (between March to June 2020).Among those households who had a migrating member come back to the village, 57% reported having gone back to the destination city at the time of the survey (December 2020/ January 2021), which was around six months after the national lockdown was lifted.The poorer households may have found it relatively harder to go back to the cities -50% among households earning less than Rs.2500 per month had returned, compared to around 60% among households who earned more than Rs.5000 per month.
Of the remaining households who had migrants that hadn't returned, about 60% expressed their desire to go back to the destination city for work.

Incomes and livelihood
The suspension of all economic activity during the COVID-19-induced lockdown was bound to have a significant impact on incomes and employment activities.Recent estimates indicate that the middle class in India shrank by more than 32 million households in 2020-2021 (Kochhar 2021).
We find this downward shift for relatively poorer households in our sample.Over 70% of households claimed to have faced income losses due to the pandemic.In Table 4, we see how this was more severe for households with migrants (at 82%).Similarly, migrant households were also more likely to face job lossesmore than 35% of households said someone in their house had lost a job compared to 13% among households who did not have a migrant.
This also may have led to the relatively higher demand for social protection schemes like the Mahatma Gandhi-National Rural Employment Guarantee Act (MG-NREGA) - Notes: *The sample is restricted to those that wanted and tried to get work under the policy.Source: Authors' calculations using primary data collected by RCRC.
the national employment scheme in rural areas that guarantees a minimum of 100 days to all householdsamong households with migrants, as observed in our data.Over 40% of migrant households reported trying to get work under this policy (within the last two months of the survey i.e.October to December 2020), compared to less than 30% among other households.Unfortunately, the demand for work was filled for around half of migrant households seeking jobs, around nine percentage points less compared to households without migrants; potentially suggesting why the majority of migrant workers had gone back to their "destination" city or were trying to go back at the time of the survey.We present the regression results in Table 5.We find that the differences between the two groups largely remain, even after factoring in the differences in their pre-pandemic incomes, their social category (SC, ST, etc.), their state of residence, and whether they belonged to an aspirational district.Households with migrants were more likely to have lost their jobs, faced a reduction in incomes and tried to get work under MGNREGA.

Health and nutrition
The survey suggests that households continued to cut down on food, even nine months after the first lockdown, amid the seemingly "revival" period.While around 40% said they had cut down on food during the first lockdown, about 25% of households continued to do so at the time of the survey (December 2020 -January 2021).Further, households cut down on nutritious itemsabout 80% had cut down on either milk, vegetables, pulses and/or oil.
We find that this situation too was worse for households with migrants as seen in Table 6.About 46% of households cut down on food during the first lockdown compared to around 38% among households without a migrant worker.Similarly, a higher share (28%) continued to cut down on food compared to the other group (23.5%).
In terms of physical and mental healthabout 15% of all households reported there was someone in the household who had been unwell within a month of when the survey was taken.The survey also asked about whether treatment of other ("non-COVID 19") illnesses was impacted due to lack of access to medical facilities due to the lockdowns and mobility restrictions, among other challenges.Around 25% reported that such treatments were affected, and we observe a similar trend with respect to households with migrants here as wellalmost twice the share of households reported having someone fall unwell in the household (around 23% among migrant households compared to 12% among others), and similarly a higher share (30%) reported that treatment for other illnesses was affected due to the ongoing crises.Households were also asked about whether they experienced an increase in stress (sleeplessness, anger, fights at home, etc.) due to the ongoing situation.About 70% of the respondents also reported that people in their household felt tense/ stressed due to the ongoing situation.They were mainly stressed about the well-being of their family, employment and money.This, too, was found to be higher among households with migrants.
The regression results in Table 7 show that all these differences continue to hold even after factoring in the socioeconomic and demographic differences between the two sampled groups.Households with migrants are more likely to reportcutting down on food (at the time of the survey and during the first lockdown), someone falling sick in the house and non-COVID-19 treatments being affected.Note: *The sample is restricted to those that reported feeling tense.Source: Authors' calculations using primary data collected by RCRC.

Financial liquidityseeking loans and pawning household assets
To further understand the degree of financial insecurity and precarity that households were pushed into, we asked households about the loans that they had taken or if they were seeking one at the time of the survey, and if they had pawned any asset (like ornaments, vehicles, livestock, etc.).This can be seen in Table 8.More than 15% of households claimed to have taken loans "within the last 3 months" (referring to the period between September and December 2020), and over 30% of households were seeking loans.Among those who took loans, over 50% reported using at least a part of it towards food and medicines.These numbers were not too different for those seeking loans as wellaround 50% said they'll use at least a part of it towards food and more than 40% said for medicines.
Following the previous trend, here too, households with migrants were substantially more likely to have taken loans (25% compared to 13%) and sought loans (42.5% compared to 26%).They were also more than twice as likely to have pawned one of their assets (14% compared to 6%).
The regression results in Table 9 show that all these differences continue to hold even after factoring in the socioeconomic and demographic differences between the two sampled groups.Households with migrants are more likely to take loans, seek loans and pawn their assets.
These numbers only further indicate the debilitating food, financial and health-related insecurities households continued to face even nine months after the first lockdown.Households may not have gotten a chance to build back, and many needed to take on debt and/or pawn their assets to meet even basic expenses.
Many migrants had to borrow money due to income insecurity and lack of provision of free transport to their homes.The lack of adequate support that could have helped migrant workers absorb the shock, further pushed them into financial precarity (Rao et al. 2020).A study conducted in Nawada, Bihar, found that migrants "over-relied" on credit and a significant increase in credit taking was seen since the onset of the COVID-19 pandemic (Sinha, Narain, and Bhanjdeo 2022).Note: *The sample is restricted to those that reported they were seeking a loan.Source: Authors' calculations using primary data collected by RCRC.

Demand for institutional support
The survey also inquired about whether households demanded and received any institutional supportfrom critical government functionaries like Accredited Social Health Activists (ASHAs), 13 Anganwadi workers 14 and Gram Panchayat (or village council); community associations like Self Help Groups (SHGs); and non-governmental or Community Based Organizations (CBOs) operating in their villages.Each of the government functionaries and community associations was responsible for playing critical roles at the time in alleviating issues around health, nutrition, financial insecurity, access to government services and schemes like MGNREGA, and lack of information/misinformation (about COVID-19 or otherwise), among others (Kesar et al. 2021).
A large share of households also reported desiring support from these institutions but not receiving it.Around 50% of households reported not getting desired support from SHGs, ASHAs, Gram Panchayats and Anganwadis.Households with migrants were more likely to not receive the desired support across each of these institutions.The  Source: Authors' calculations using primary data collected by RCRC.
difference was largest among those seeking support from the Gram Panchayat (55% compared to 45% for households without a migrant).
The regression results (in Table 11) show that while all the differences do not hold once other factors are controlled for, they continue to remain significant for households with migrants receiving less support from ASHA workers and their Gram Panchayattwo critical frontline functionaries responsible for ensuring the continuance of several services and access to social protection strategies initiated by the government.While Anganwadis, ASHAs and panchayats are formal institutions, adequate access to these during a time when they had limited capacity and resources may have depended on informal ties and networks which households with migrants are more likely to lack.

Discussion
The paper discusses the seemingly paradoxical outcomes for migrants.We find that while households with migrants were economically better off to start with, the outcomes for these households post the lockdown were worse off on economic and nutrition outcomes.The specific experiences of migration would vary over regions (Tumbe 2018;2019).For instance, Rajendra, Sarin, and Singhal (2021) found that migration experiences and vulnerability varied along factors like skill levels, caste and religious identities in districtlevel data in Bihar.However, an underlying thread that connects these stories is the paradox of being economically better off but worse off in absorbing economic shocks.
Migrants form a crucial part of the neoliberal paradigm of development, contributing to 10% of India's gross domestic product (Mehrotra 2021).In our analysis, although the incomes are higher than non-migrant households, they are still low.Insufficient incomes with low social infrastructure imply poor outcomes on health, especially for women and children (Ravindranath, Trani, and Iannotti 2019).During the pandemic, migrants immediately became an "unrecognized category" in their destination cities without any protection or mechanism to access basic needs amid a debilitating crisis (Rao et al. 2020).
Earlier studies have pointed to potential pathways that help us make sense of our observations.Munshi and Rosenzweig (2016), point to a trade-off between economic CJDS / LA REVUE gain due to migration and the social insurance from caste-based social networks in the villages.They argue that male migration reduces social insurance because of diminished access to rural caste-based networks.Consequently, male migration occurs primarily when the income gain to the household is large enough to compensate for the loss in this insurance.They predict that "relatively wealthy households within their caste benefit less from the network and so will be more likely to have migrant members ceteris paribus" (49).Migrant households would therefore be typically characterized by higher incomes and poorer networks.De Weerdt and Hirvonen (2016) have observed a similar trade-off in Tanzania's internal migration and characterize migration "as a risky but profitable endeavor" (81).Lagakos, Mobarak, and Waugh (2023) based on work in Bangladesh also find that migration came with a considerable "non-monetary disutility." The government response to the COVID-19 crisis involved several protection programs such as cash transfers through Jan Dhan Yojana, free ration through the Public Distribution System (PDS), and expansion of MGNREGA.
Studies have also pointed to the distress in incomes post the lockdown and the role of MGNREGA, which has consistently been seen as important employment insurance at times of distress and crises (Imbert and Papp 2020;Morten 2019).While the implementation of MGNREGA has remained uneven across India since its inception (Afridi, Mahajan, and Sangwan 2022), its potential and success in being a protective measure during times of distress and crises have been noted (Afridi, Mukhopadhyay, and Sahoo 2016;Imbert and Papp 2020;Morten 2019).In a survey conducted in villages with high out-migration in West Bengal, Gupta et al. (2021) found that households lost almost 90% of their "local incomes" and over 60% of remittance incomes right after the lockdown.Afridi, Mahajan, and Sangwan (2022) find that districts that had better public employment-generating capacity through MGNREGA were more likely to absorb job losses in rural areas.However, disbursement was not immediate, and the implementation remained patchy and varied during the crisis.
There is a need not just to strengthen formal social protection mechanisms within the village but also to improve it at the migrant destinations (Sinha, Narain, and Bhanjdeo 2022).Post the series of lockdowns, some state governments have begun running pilot projects to increase access to social security schemes (Chakraborty 2022; Skill Development and Employment Vertical, NITI Aayog 2021), although the efforts have been far and few.International Labour Organization, Aajeevika Bureau and Centre for Migration Inclusive Development ( 2020) and a Working Group on Migration (Bhan et al. 2017) have also charted a way forward for the consolidation of migration policy in India.This significant loss in nutrition and health outcomes are likely to be a consequence of job and income losses, and inadequate support from social protection schemes like.The lack of access to food during the pandemic for migrants has also attracted significant policy attention (Choudhury et al. 2020).The portability of the ration card has been one of the predominant policy focuses of the central government.However, given the seasonal nature of the migration, the strengthening of food provisions in the form of canteens is offered as a potentially more viable alternative (Khera 2019).Several states already have versions of state-run canteens, Amma canteens in Tamil Nadu and Indira canteens in Karnataka.
A future policy response to the concerns of migrants has been designing portable social security benefits.Given the complexity of implementing schemes targeted at migrants, including seemingly straightforward ones like the One Nation, One Ration 15 ; we must develop a more nuanced understanding of the experience of migrant workers in India.The implementation, as in the case of the portable ration card, is often imagined using the Aadhaar infrastructure (serving India's biometric identification program).However, many have raised concerns about data privacy and exclusionary errors (Khera 2019).Ideas around an urban employment scheme are also being discussed (Drèze 2020).The need to include these security measures in local governance systems, except for Kerala, has been conspicuously absent from these future imaginations (Rao et al. 2020).The paper joins the large scholarship on migration to draw attention to the need for meaningful policy attention.

Notes
because of the sampling design (independent of influences resulting in and stemming from migration itself)?Alternatively, are there unobserved factors affecting participation in the survey related both to a household's migration status and the outcomes we measure?While we cannot rule out these possibilities, we have no reason to believe that enumerators systematically surveyed different types of households (in terms of vulnerabilities) based on whether or not they were migrant households.Nevertheless, this possibility points to the limitations in being able to generalize our results to a larger population of migrants and non-migrants.12. Health insurance coverage in India, including passive enrollment under government schemes such as Ayushman Bharat Yojana, can vary in terms of effectiveness and benefits received.Therefore, this might as an indicator of enrollment and awareness rather than a measure of comprehensive coverage and outcome 13.ASHAs are community health workers serving under the National Rural Health Mission of India 14. Anganwadi workers are part of India's Integrated Child Development Services (ICDS) and responsible for providing nutritional, health, and education services to children and women 15.A scheme by the Government of India allowing holders of Ration Cards under the National Food Security Act to access their food subsidies and entitlements in any part of the country.See Vamakshi (2022) for a discussion on its (in)effectiveness.

Table 1 .
Distribution of sample across States.

Table 2 .
Basic characteristics of the sample.

Table 3 .
Differences in background characteristics between households with and without migrants.

Table 10 .
Support from NGOs, Panchayats and others.

Table 11 .
Support from NGOs, Panchayats and others (Regression results).