Migration and Remittance Profile
Across the four hotspots, most migrants were married young men of 21–30 years with secondary or higher education levels (Table 1). This is consistent with migration literature suggesting that older men are less inclined to migrate and educated younger men with limited access to resources and input into household decision processes are more likely to do so [11, 43, 75, 77]. The gender balance in outflows from the study hotspots suggested that household migration remained male dominated. Previous studies have established that gender is an important form of social differentiation that influences migration in developing countries . Migration requires sociocultural acceptance, and economic and physical capacities that are not equally available to women [10, 50]. However, there is diversity in female migration across the study hotpots and sites within the hotspots, with some sites reporting higher proportion than others. For instance, the mountain areas in the Gandaki and Teesta river basins, the Kendrapara and Bhadrak districts of Mahanadi deltas in India, and the Barisal and Khulna divisions of the GBM deltas in Bangladesh reported higher female migration rates. The reason for higher female migration in high mountain areas was the sociocultural acceptance of migration by women, coupled with a comparatively higher education attainment. Many respondents reported better job opportunities and opportunity to pursue further education as the major reason for migration of women. Social norms in high mountain areas are more liberal as compared with stringent patriarchal system followed in the plains. Furthermore, in high mountains, environmental conditions necessitate multi-local livelihoods with established familial and friendship networks in destinations. This social norm makes the acceptance of women’s mobility easier. In deltas, destruction of local livelihoods and existential threat from a number of extreme climate events was reported as the major driver for women to migrate. The study areas experienced significant losses and increased household vulnerability from cyclones Sidr 2007, Aila 2009 and Phailin 2014.
Where Do People Migrate and the Role of Remittances
Consistent with national statistics and other studies [36, 41, 77, 104], migration was predominantly internal across the four study hotspots. The prevalent types of movements were daily commuting from peri-urban to urban areas, seasonal/circular migration (< 6 months/year) and long-term migration (> 6 months). International migration was high in certain study sites such as the Gandaki river basin (Nepal), GBM delta in Bangladesh, and Faisalabad district in the semi-arid plains of Pakistan. Among international migrations, the most popular destinations were the oil-rich Gulf countries and Malaysia. This finding is consistent with that of Siddiqui et al. ’s assessment in the Hindu Kush Himalayan region. Most migrants worked in informal sectors, usually hired as daily wage labourers in industries such as construction or small-scale retail and hospitality.
Migration affected both origin and destination areas through remittances—financial and social. Remittances, it is argued, can provide flexibility in livelihood options, supply capital for investment, and spread risk . Across our four study hotspots, 80% of migrant households reported receiving remittances (slightly lower in deltas, where 66% of respondents reported receiving remittances). Certain regional differences were observed in deltas. For example, only 48% of households reported receiving remittances in the Mahanadi delta in India against 85% in GBM in Bangladesh. This lower remittance transfer in deltas was associated with a short duration of migration. For example, 82% of migrants in Mahanadi delta moved for less than 6 months, and carried remittance in cash and kind in person. Overall, the average annual remittances from internal migration (USD 543) were much lower than those recorded for international migration (USD 1703) in the study sites.
Remittances have the potential to enable rural households to overcome credit and risk constraints by the spatial diversification of labour and income . Moreover, if invested in modern agricultural technologies, tools and livestock, subsistence farmers can increase their productivity and complete the transition from familial to commercial production, which is instrumental in the diversification of rural economies [9, 62]. It is the capacity to mobilise new resources that can enhance household resilience against the pervasive effects of climate change. However, our empirical evidence suggested that across the four study hotspots, remittances were scantly employed to enhance asset base or invest in income-generating activities locally. For example, in the deltas and semi-arid plains, 89% of surveyed households used remittance income to help pay for food, health, education, debt repayment, and household appliances. Consistent with other studies [34, 56], our empirical evidence suggested that remittances acted as a financial buffer against economic losses, thus representing reactive (albeit powerful) coping mechanisms during hardships. In the river basins, 17% of households reported using remittances to meet household food and non-food needs during hardship brought about by extreme environmental events.
In addition to financial remittances, migration brought social remittances such as new ideas, knowledge, skills and technologies from destination to origin areas for development in the areas of origin [9, 57]. Empirical evidence across all four study hotspots supported this theory. In deltas, 75% of surveyed households reported benefits from new ideas and knowledge to build adaptive capacities at origin. Similarly, 29% of migrant households in semi-arid plains reported learning new knowledge and skills compared to 20% of non-migrant households. The difference was statistically significant at 5% level of significance. Similarly, labour migrants from semi-arid regions and deltas acquired new skills such as masonry, carpentry, and catering at their destination. As a result, migrants were able to diversify their livelihoods upon returning to place of origin. In the Upper Ganga areas migrants had introduced mobile apps to the communities. These apps provided weather-related information, allowing daily agricultural activities to be planned.
Why Do People Migrate?
Consistent with previous findings on motivations for migration [36, 41, 77, 104], the main driver of migration as reported by migrants across the four study hotspots was economic, often associated with better employment opportunities elsewhere. When asked ‘What was the primary reason for migrating?’ 55% of respondents in semi-arid plateau, 82% in semi-arid plains, 48% in deltas, and 44% in river basins reported economic reason as the primary reason. Other important reasons reported were to pursue higher education, meet family obligations, diversify from unprofitable agriculture, earn better wages, overcome landlessness, and moving for marriage. These responses revealed that unequal development, leading to lack of economic opportunities and access to basic services, was a major driver for people to move from rural areas to urban areas.
Only 6% of the respondents in deltas and 12% in semi-arid plains cited environmental causes such as drought, flood, cyclones, increased temperature, and erratic rainfall as their main motivation for migration. This result highlighted that, in most cases, households did not identify migration of a household member as being related to the environment. It illustrated, rather, that decisions to migrate were multi-causal, in which economic reasons were the primary motivation. Consistent with previous findings [1, 101], environmental drivers had weak attribution in the migration decisions in the study sites. The discrete research designs employed in this study had certain limitations which inhibited the capture of environmental attributes to migration. As previous studies suggest, it is difficult to measure the relative significance of environmental factors vis-á-vis other drivers in labour migration . This was particularly true in cases where empirical evidence was collected via cross-sectional surveys in sending areas. Direct causal signals are stronger in displacement or permanent relocation following sudden-onset environmental hazards. However, recently emerging literature with methodology to capture household’s perceptions regarding climate change and household migration decision paves new ways to investigate this linkage [54, 55].
Barriers to Migration
Across the four study hotspots, migration was not always available to households exposed to environmental stressors. The proportion of households that reported migration of one or more members was only 29% in the river basins, 41% in semi-arid plains, 39% in semi-arid plateaus, and 24% in deltas. There were several barriers to migration. Sociocultural norms, household composition, and intra-household work sharing norms made migration a highly gendered process, typically curtailing women migrating . In the study hotspots, lack of financial capital and social network in destination were the most prevalent barriers. In case of Pakistan (Indus river basin and semi-arid plains), having sufficient male adults in the household to send for migration was reported as another important barrier. Lack of access to safe accommodation in destination areas and constant preoccupation about maintaining family commitments in places of origin were reported as significant constraints preventing women migration during the focus group discussion in Bangladesh deltas. Marital status was also highlighted by respondents, with married women enjoying more autonomy to migrate compared with single women.
The barriers discussed above have limited migration as a response to climate change in the study areas. Many surveyed respondents mentioned the desire to diversify their income sources by adding migration into their livelihood portfolio. In the semi-arid plains and deltas, for example, the proportion of immobile households was 59% and 35%, respectively.
Migration for Household Adaptation
There is a growing body of literature on climate change adaptation and its effect on reducing vulnerability to climate change impacts at household levels [27, 70]. Research endeavours in measuring adaptation have focused heavily on assessing the vulnerability or adaptive capacity through various indices [38, 45]. But linking migration to adaptation outcomes is complex with changes over temporal and spatial dimensions being critical [35, 64, 104]. The research analysis conducted by the four CARIAA consortia employed a suite of analytical tools to examine this relationship in the study hotspots. HI-AWARE examines the adaptation behaviour of the households, DECCMA analyses the differences in adaptation measures of households, PRISE assesses the difference in livelihood resilience of the households, and ASSAR examines changes in the material and subjective well-being of migrant households. Each analytical tool offers a different lens in the migration and household adaptation interlinkages. While the different analytical tools employed make it challenging to compare the results directly across the hotspots, we report the key findings on the implications of migration for household adaptation and attempt to synthesize them. In each section, we briefly describe data and analysis techniques employed in each study hotspot followed by key findings. We highlight that in this synthesis report, we present findings derived from both quantitative and qualitative data and analytical techniques.
In the river basins, we examined the adaptation behaviour of households to climate change impacts separated by migration status (migrant and non-migrant households) of the household in four crucial livelihoods sectors—agriculture, livestock, water, and forest. Households were classified as adaptors or non-adaptors—adaptors were households that had reported undertaking at least one measure to reduce the negative impacts of climate change in the particular sector in the year prior to the data collection. A household was categorised as migrant if it had at least one member involved in labour migration for at least 3 months in the year prior to the data collection. The statistical relationship between adaption behaviour and migration status was tested using Pearson chi-square test of independence in the four livelihood sectors. Test results showed that in agriculture sector, migration played a statistically significant role in adaption behaviour of households (Table 2). In other sectors, the differences were statistically not significant. For majority of the households in the study sites, agriculture was a major economic sector and households depended on farm production to meet their food security. This probably explains the reason for migrant households’ investment on adaptation measures in agriculture sector. The most commonly used adaptation measures in agriculture were introduction of new crop varieties, use of pesticide/insecticides, adjustment of timing, improved irrigation, shifting to non-farm activities, etc.
Although more than 90% of households perceived changes in climate as compared with situations a decade before, less than one-third of the households reported undertaking adaptation measures to reduce the negative impacts of such changes. Overall, higher proportion of households reported undertaking at least one adaptation measure in agriculture and water sector and least in forest sector. Consistent with findings of Hussain et al.  in the Koshi river basin, most of the adaptation measures undertaken by households were autonomous rather than planned. This illustrates the importance of identifying and verifying the autonomous local adaptation practices; however, they might be unable to manage new risks and extreme changes in future .
The analytical approach in deltas was two-pronged, examining (1) who were migrating as an adaptation response to climate change (assessed through exposure to floods, droughts, sea level rise, erosion, and salinity), and whether or not they deemed that to be a successful adaptation; and (2) differences in adaptation measures adopted by migrant and non-migrant households using Chi-square tests to see if there were statistically significant differences.
Migration was an adaptation response in all three deltas: by 28% of households in Bangladesh; 53.3% in the Indian Bengal delta; and 23% in the Mahanadi. Interestingly, though, migration was not among the top three strategies considered successful adaptations in any of the deltas. Temporary mobility associated with production-related moves was also captured in the survey by way of people reporting work outside the village. Such an adaptation response was mentioned widely by between 51 and 71% of respondents across the three deltas.
Table 3 shows results for the most prevalent adaptation responses. The statistically significant differences between migrant and non-migrant households relate to the use of adaptations that were finance-related. In terms of increasing income to the household, more migrant households took loans in the Indian Bengal, Mahanadi, and the Bangladesh portion of the GBM deltas. This may have been because remittances could act as a guarantee for such loans. Migrant households also adapted by modifying their homes: for example, by constructing more permanent structures that better withstood flooding and cyclones. This was an important adaptation in deltas, and qualitative interviews in the Indian Bengal delta indicated that this was one of the primary priorities of households.
In Bangladesh, however, it is important to note that there was only a marginal (statistically insignificant) difference between migrant and non-migrant households in terms of taking out loans. This was likely to have been because of widespread household access to microfinance. Bangladesh has been a pioneer in microfinance operations since 1980 . Work outside the village was the only adaptation response that recorded statistically significant differences across the three deltas. Although the survey did not capture the type of mobility associated with that answer, there was a wide range of temporary moves including daily, weekly, and circular trips. In practice, that underpins the livelihoods of people worldwide, and this is particularly the case for smallholder and subsistence farmers living in marginal environmental conditions. Such movements occur for a wide range of reasons including labour and economic motivations.
In contrast, there was no statistically significant difference between migrant and non-migrant households in the use of livelihood-based adaptations, for example diversifying crops; planting climate-tolerant crops; increasing use of irrigation; or using new farming and fishing equipment in the Indian Bengal, Mahanadi, and Bangladesh deltas.
The approach in the semi-arid plains sought to understand the relationship between migration and resilience by using a livelihood resilience index. The study defines livelihood resilience as ‘the capacity of all people across generations to sustain and improve their livelihood opportunities and wellbeing despite environmental, economic, social and political disturbances’ . The resilience index was constructed following the method developed by Cutter et al. . Livelihood resilience index is composed of three components: adaptive, absorptive, and anticipatory capacity. Following Bahadur et al. , adaptive capacity is defined as the ability of social system to adapt to multiple, long-term, and future climate change risks, and also to learn and adjust after a disaster, Similarly, anticipatory capacity as the ability of social systems to anticipate and reduce the impact of climate variability extremes through preparedness and planning capacities, and absorptive capacity as the ability of social systems to absorb and cope with the impacts of climate variability and extremes. Each component is further divided into sub-components (see Bahadur et al.  for detail). Indicators defining the three types of capacities were selected through a careful literature review and customised to match the local context of the study areas.
We found that overall migrant households were more resilient than non-migrant households and they scored better than non-migrant households in all the three components (Fig. 3). They had greater adaptability to shocks as a result of higher and more diversified income sources, better housing, and a higher employment rate. Migrant households also had higher levels of subjective well-being, as they were more comfortable in coping with stressors, more at ease in making life decisions and were exposed to a wider range of opportunities to learn new skills and improve their livelihoods. They were also more adept at planning for the future, more informed about climate change impacts, and had higher capacities to anticipate and deal with shocks such as climate extremes and food insecurity .
In the semi-arid plateau sites, in addition to the household surveys, life history interviews (n = 37 across rural, peri-urban, and urban sites) were used to examine temporal vulnerability and changes in the material and subjective well-being of migrant households across the rural–urban continuum [84, 85]. These life histories were supplemented by historical timelines to reconstruct local ecological, socioeconomic, and politico-institutional changes, and consequent livelihood dynamics. The in-depth life histories analysed ‘family trajectories of accumulation or impoverishment over time and of particular matrices of vulnerability’ (: 489) and thus constructed an understanding around temporal vulnerability and how households followed ‘(household) trajectories towards vulnerability or resilience’ (: 3). These narratives also allowed a nuanced inquiry into how personal and family aspirations, asset constraints, and needs interacted with social norms, agency, and larger-scale institutional and economic changes to shape livelihood choices .
Migrant life histories demonstrated that permanent migration, especially accompanied by accumulation of education, skills, and social capital can, over a period of time, be a positive adaptive response, expressed through increased incomes, better living facilities, improved access to services, and lower exposure to extreme events as compared with where they moved from . However, we also found instances where permanent migration was also detrimental, especially when there was inadequate employment or poor living conditions associated with tenure insecurity in destination areas, entering into unsafe livelihoods, or living in flood-prone areas in the cities they moved to . Critically, the life histories demonstrated that the outcomes of migration decisions change over time with some houses improving their assets and capabilities and hence adaptive capacities while others (typically those with poor social networks, low assets, and belonging to marginalized social groups) remaining trapped in cycles of coping. Furthermore, migration outcomes were differentiated within the household, with men and women within migrating household reporting differential impacts on well-being. In keeping with previous research (e.g. ), women reported higher work burdens and reduced time for leisure once they moved. However, there were also examples contradicting perceptions of gendered migration: in peri-urban areas, women reported migration improving subjective well-being and adaptive capacities through improved incomes, more autonomy, and changing intra-household gender relations .