Left behind? Internally migrating children and the ontological crisis of formal education systems in South Asia

ABSTRACT In-country migration is widespread in South Asia, and the region hosts the world’s largest number of out-of-school children. Yet the relationship between internal migration and inclusion in formal education has received only limited academic and policy attention. The Agenda 2030 pledge to leave no-one behind prompts us to argue that when it comes to migrating children, formal education systems in South Asia are in ontological crisis. Cases of mobile pastoralists and seasonal labour migrants illustrate that being ‘left behind’ is produced by intersecting norms of modernity, which formal education systems reproduce, via logics of rights, human capital and ‘terms of inclusion’. Formal education systems are antithetical to mobility-dependent livelihoods, discount situated learning, and perpetuate unequal social relations. ‘Alternative’ education has done little to contest these norms and their productions. Terming this an ‘ontological crisis’ signals both the disjuncture between realities and Agenda 2030’s moral exigencies, and the opportunity to act.


Introduction
The Education For All (EFA) movement and successive UN-led Monitoring Reports (e.g. Reaching the Marginalised (GMR 2010); Education for All 2000: Achievements and Challenges (GMR 2015; and Migration, Displacement and Education (GEMR 2019)) have highlighted that migrating groups are highly excluded from formal education provision around the world (Dyer 2014). The Sustainable Development Goal (SDG) framework (Agenda 2030) recognises 'migrants' as one of the most vulnerable communities who must be 'empowered' and provided 'inclusive and equitable quality education', and explicitly commits to 'leave no one behind' (UN 2015).
Ironically, however, current discourses concerning 'migrants' tend to construct their subject in ways that include many, yet exclude more numerous others. Concerns over refugee and displacement-related migration have led 'migration' scholars to focus so sharply on international, cross-border migration (Arnot, Schneider, and Welply 2013) that, in what Hickey and Yeoh (2016) term a 'semantic shift', migration has become conflated with international migration. This turn is replicated at global policy level in the 2016 New York Declaration for Refugees and Migrants (United Nations 2016), and in the Agenda 2030 framework, which implicitly assumes that 'migrants' are 'refugees' and 'displaced' persons. Globally, however, the numbers of people undertaking various forms of internal migration are far higher than those who cross national borders to work (IOM 2019; WMR 2020), a fact that is 'easy to forget' (GEMR 2019, 13).
In South Asian countries, internal migration is extensive, and larger in scale than international/regional migration (WMR 2020). In South Asia, temporary/seasonal migration is not only a more prominent form of internal migration than permanent migration (Babu et al. 2017); it is also on the rise, propelled by economic growth and urbanisation (WMR 2020). Yet the nature and complexities of internal seasonal migration are inadequately addressed in legislation and policies, where it has received, at best, fragmented references (UNICEF 2012). While South Asia also has the world's largest number of Out-Of-School Children (OOSC) (GEMR 2019), the relationship between internal, seasonally linked migratory patterns and in-and exclusion in formal education has received very limited academic and policy attention (Bernard, Bell, and Cooper 2018;Bengtsson and Dyer 2017;Cameron 2012;Dyer 2016;GEMR 2019;Smita 2008).
The Agenda 2030 construct of 'left behind' requires a political engagement with fundamentally unequal structures in society and schooling, which is what prompts us here to discuss what we see as an 'ontological crisis' in South Asian formal education systems. That is, modern school systems' ontological privileging of sedentary populations/children has 'designed in' an internal systemic incapacity to ensure inclusive and equitable quality education for internally migrating children. Until this ontological crisis is identified and addressed, only 'stopgap' education interventions for these children can be imagined; and it is these that are currently pressed into inadequate service of the ambition to 'leave no-one behind'. This paper substantiates its claim of ontological crisis by questioning the construct of 'left behind' with reference to two broad groupings of people who use internal, seasonal/temporary migration as a livelihood strategy: mobile pastoralists and seasonal labour migrants. Children in these groupings remain among the most excluded from formal education provision in South Asian countries (Bengtsson and Dyer 2017;Chandrasekhar and Bhattacharya 2018;Cameron 2012;CIER & UNICEF 2009;Dyer 2014Dyer , 2016GEMR 2019;UNICEF 2014). In this respect, they constitute a significant 'left behind' population segment in the region, with which Agenda 2030 must be concerned. With reference to these illustrative cases, we argue that the ontology of contemporary education systems in South Asia makes current system potential to 'put the last first' and meet the moral exigencies of Agenda 2030 very doubtful; yet to see this is an 'ontological crisis' is to enable recognition of a productive moment that can precipitate change.
We note at the outset that the literature available on this topic for the region is very slim, and dominated by studies of India. Given the urgency of the situation, we attempt, nevertheless, to reflect also on Afghanistan, Bangladesh and Pakistan, but directly comparable data on different parameters are not available, and we have not extended our focus to include Nepal, Sri Lanka and Myanmar.

Mobility as a static referent
Agenda 2030 calls for 'high quality, timely and reliable' data, including migratory status, to envision systemic change (United Nations 2015). 'Migratory status' is complex, and this terminology implies a fixity that is in tension with ground realities when the concept of 'migration' is extended beyond the field of refugees and international border crossing (Arnot, Schneider, and Welply 2013). Our emphasis on the need first to interrogate 'migration' and mobility is, perhaps ironically, in recognition of a persistently sedentary bias in perceptions of what development is and should be (Danaher, Kenny, and Leder 2009). This bias in turn fuels a tendency for policy documents across sectors to ignore mobility, or to define it as a problem (De Haan and Yaqub 2010). We note, too, that discourses of modern childhood normatively posit childhood as a site of 'residential fixity' and 'domestication', which has led to claims of a 'moral panic' around the lives of children who are mobile (Ní Laoire et al. 2010).
At present, no comprehensive and systematic literature is available to shed light on the complex mobility patterns of internally migrating populations in South Asian countries (Babu et al. 2017;Bengtsson and Dyer 2017;UNICEF 2014). This, coupled with absence of systemic and comparable data, makes it difficult to understand the actual scale and nature of various forms of mobility (Srivastava and Pandey 2017) and how education systems could better respond. Terms in the migration literature used to indicate variations in migration patterns feature aspects such as place (international/regional/internal), duration (permanent/semi-permanent/temporary/seasonal), pattern (rural-rural/ruralurban/urban-rural/urban-urban) and nature (economic migrants/political migrants/ environmental migrants/refugees/asylum seekers/internally displaced persons/stateless persons) (e.g. IOM 2019). These categories in reality defy neat boundaries: there are multiple intersections and complexities around them; and those, also, vary across countries. Further, individuals and/or families often undertake multiple forms of migration over the life course. Mobile pastoralists generally fall outside the purview of 'migration' scholars altogether (Dyer 2012), although the UN's 2018-19 Global Education Monitoring Report (GEMR) Migration, Displacement and Education does note, in its chapter (2) on internal migration, that both pastoralist and seasonally migrating children are denied their right to education, and devotes a sub-section to each (GEMR 2019).
In respect of education inclusion, the scarcity of accurate data is compounded by an emphasis on population stocks (absolute numbers at a given moment) (Carr-Hill 2012). While education services need to know stocks, flows (movements over a given interval of time), too, are important (Deshingkar and Akter 2009). Mechanisms currently in use may capture permanent and semi-permanent movement, but not that which is shortterm and seasonal (Srivastava 2012;Srivastava and Pandey 2017) or highly variable. For example, India's National Sample Survey Office (NSSO) has attempted (from its 55th round onwards) to capture short-term movements separately, and in its 64th round defined short-term migrants as people who 'stayed away from the village/town for a period of 1 month or more but less than 6 months during the last 365 days for employment or in search of employment' (GoI 2010). This definition does not capture movements outside these parameters (Chandrasekhar and Bhattacharya 2018;Srivastava 2012), such as families who earn their living from the charcoal industry in India who move intermittently for about three weeks each time during the annual school cycle (Bengtsson and Dyer 2017).
While it is important to understand mobility itself better, using it as an isolated and static referent is problematic. Mobility is integrated into a wide variety of livelihoods and associated socio-cultural values; it is adopted for very different reasons; and its patterns are dynamically shaped by the -often erratic -economic landscapes of current times.

Two mobility-dependent livelihoods: mobile pastoralism and seasonal labour
Let us, then, illustrate how the issues we have discussed above are manifest in relation to mobile pastoralism and seasonal labour migration, the two mobility-dependent livelihoods on which this paper focuses.
Mobile pastoralists are livestock producers in a system that uses mobility to access the natural resources on which it largely depends, and is practised across South Asia's drylands and mountainous areas (Dyer 2014, Rao andCasmir 2008;Sharma 2011;Sharma, Koeller-Rollefson, and Morton 2003). Reflecting this ecological variation, animals herded include yaks, camels, cattle, sheep, goats, and even ducks. Well-known examples of local pastoralist communities include Kuchi, Bakkerwal, Raika, Gaddi, Gujjar, and Rabaris. In mountainous regions, much migration is 'vertical' -up for summer pastures and down for winters; across the plains, it is 'horizontal' (Sharma, Koeller-Rollefson, and Morton 2003). Seasonal patterns of timing and direction are relatively constant, with variations according to monsoon quality; day-to-day movement responds to local variability of fodder and water. Pastoralism is a sophisticated and important dryland livelihood, but there are many pressures on it. The South Asia Pastoralist Alliance, a much needed regional platform founded in 2015, identified as key problems: decreasing grazing land and fodder supplies and ineffective policies to address constraints; poor access to markets; missing information on pastoralist systems in South Asia; and education and healthcare services for people and animals that are not adapted to mobile livelihoods (WAMIP 2015).
It is virtually impossible to provide up-to-date and accurate population counts of mobile pastoralists in the region: estimates are about 35 million for India, according to its Centre for Pastoralism, 1.2 for Pakistan and around 2 million for Afghanistan (Dyer 2014). Carr-Hill (2012) sets out aspects of omission 'by design' in counting mechanisms that are designed to capture stocks of permanent, sedentary populations. There are also issues of definition at multiple levels: global scholarship tends to use 'mobile pastoralists', a distinctive term that draws in both mobility and livelihood type, while regional sources tend to refer to 'nomads', or 'nomadic pastoralists'. In Afghanistan, the generic term Kuchi (those who go on migration, in Persian) is used in policy documentation to refer to all Afghan nomadic communities. This gloss confers a single political identity on an internally diverse group, some of whom are Kuchi, pursuing increasingly differentiated livelihoods.
Seasonal migrant workers are, according to the 1990 International Convention on the Protection of the Rights of All Migrant Workers and Members of Their Families, Article 2(b), those 'whose work by its character is dependent on seasonal conditions and is performed only during part of the year' (United Nations 1990, 2). Their patterns of labour movement are often complex and depend on various factors such as the length of migration cycles; annual frequency of migration; type of destination; nature of work; and whether migration is undertaken by single individuals or families (Srivastava 2012). Prominent sectors in which they are employed (some of which the International Labour Organisation categorises as 'hazardous' forms of labour) include garment/textile industries; construction; domestic work; plantation work; mining; fishing; brick-making; salt-making; tile-making; charcoal-making; rice mills; sugar cane harvesting; and stone quarrying. Various micro studies and surveys show that short-term and seasonal migration has been rising, in response to a range of choices and structural factors (Babu et al. 2017) such as changes in livelihood opportunities in rural areas, regional disparities between rural and urban areas and changing patterns of urbanisation and informalisation (Srivastava and Pandey 2017). These often prompt movement to informal urban settlements and work in informal labour sectors (Deshingkar, Sward, and Estruch-Puertas 2012), or from agriculturally poorer regions to rural irrigated areas (Srivastava and Pandey 2017). Whether such movements amount merely to 'survival and coping' or 'accumulation' (Deshingkar and Start 2003) requires more systematic analysis, but it is evident that seasonal, short-term and circular migrants are prone to exploitation and vulnerable living and working conditions (Cameron 2012).
Numbers of these internal migrants in South Asia are also not accurately known, and there is huge discrepancy between claims of official statistics and of micro-level estimates. For example, in India, while the NSSO's 64th round estimated the number of short-term migrants at around 15 million, Deshingkar and Akter (2009) estimated about 100 million circular migrants. Though these figures may not be directly comparable because of definitional and methodological complexities, the scale of such migrating communities is, clearly, heavily underestimated in policy circles. The often interchangeable and different use of the terms short-term/temporary/circular/seasonal in different country contexts, further, makes it difficult to understand the exact nature of flows of these populations.

Invisibility in Out-Of-School Children discussions
UNICEF's current estimates of OOSC in South Asia put totals at about 11.3 million at primary school level and 20.6 million at lower secondary level (UNICEF 2015), while Beteille et al. (2020) report a higher total of around 35 million. The discrepant totals aside, such aggregated data inadequately capture the status of mobile children (UNICEF 2014), such that they are likely to be over-represented among OOSC statistics yet paradoxically largely invisible within reporting. UNICEF's 2014 South Asia regional study, for example, identifies children from families who migrate seasonally for work alongside children in other groups where exclusion is common (such as girls and street children), but does not mention mobile pastoralists (UNICEF 2014). The global Out-of-School Children Initiative (OOSCI) that UNICEF leads has begun to devote attention to 'Invisible OOSC', raising issues of child (in)visibility to reporting and recognising who is out of school and why (UNICEF/UIS 2016). Pastoralists are a case in point, as UNICEF's own reporting shows, but there is an evident need to focus deliberately on mobile out-ofschool children (cf. Dyer and Echessa's (2019) use of MOOSC to do so).
The binary implied in counting children as either in or out of school, which is inherent in the term OOSC, is itself misleading. Many migrating children are 'nominally enrolled' (Srivastava and Dasgupta 2016) as their names are retained in the home school register even when they move out during the school year (Chandrasekhar and Bhattacharya 2018): they are thus 'in' but also 'out'. A notable example of an attempt to pay special attention to irregular attendance, even if not specifically identifying mobility as a possible cause, is found in the Indian state of Karnataka, which changed the definition of 'drop out' in 2013 from 'continuous absence of a child for 60 days' to 'continuous absence of child from the school for seven days' (Rajan 2019). In general, however, children's absence from school is poorly recorded, under-reporting is widespread and patterns of irregular attendance are rarely taken into account while calculating the number of OOSC (Bhatty, Saraf, and Gupta 2017).

Discourses of inclusion and universality: how rights and human capital perspectives create notions of 'left behind'
Vast gaps between the formal education provision that is available and the ground realities of migrating children's lives give rise to numerous barriers to initial access, participation and progression (UNICEF 2014). These barriers reflect 'terms of inclusion' (Dyer 2013) that are imposed by norms of geo-spatially fixed schooling that are exclusionary for migrating children, and often conflict with their families' livelihood imperatives. Fundamentally, terms of education inclusion are constituted by conflicting normative views of childhood, mobility and modernity, which create ontological dissonance. We illustrate this dissonance and how it produces being 'left behind' with reference to two key ideas in policy discourses of universal education inclusion in South Asia: education as a right; and education as a means of developing human capital.

Rights-based individualism and 'other' childhoods
The 1989 UN Convention on the Rights of the Child (UNCRC) is ratified by all countries in South Asia, but rigorous research about the impact of a broad human rights framework on policies of education across the region is lacking (Bajaj and Kidwai 2016). States have long recognised education as a fundamental right, but adoption of a rights-based approach to formal education supported by legislative frameworks is mostly recent. Based on criteria of constitutional and legal provisioning, Bajaj and Kidwai (2016) classify only India and Sri Lanka as providing a full constitutional guarantee of free and compulsory education, while other countries provide a partial guarantee.
The normative view of childhood that underpins rights discourses, however, reflects a hegemonic Western construct of childhood as a developmental stage, sharply distinct from adulthood, which is at odds with social constructs of childhood within the collective orientation of South Asian society (cf. Zaidi, Couture-Carron, and Maticka-Tyndale 2016). Understanding childhood as a social construct highlights the uneasy co-existence of Western modernity and the politics and possibilities of policy frames -whether international, such as the UNCRC, or national -that invalidates the existence of diversity by assuming childhood is a static and universally defined phenomenon (Burman 1996). Categories that have objective existence in the hegemonic discourse of Western childhood, such as 'age' and 'child rights', are based on Western notions of individualism, a nuclear family, child protection, and formal education. Taken in conjunction with the assumption that children have no economic role, this notion of childhood leads to the construction of other -i.e. Southern -childhoods as deficient and lacking (Nieuwenhuys 1998).
With respect to migrating children, in addition, rights discourses are organised in policy through sedentary imaginaries of formal education as fixed-place schooling. For migrating children, mobility is an additional axis that intersects with caste/gender/class/ place/religion, which are structures of inequality and marginalisation that modern schooling tends to reproduce (Jeffrey, Jeffery, and Jeffery 2004). Dyer (2019), for example, shows how the notion of the 'neighbourhood' school that India's 2009 RtE Act invokes assumes a fixity of 'neighbourhood' that does not exist for migrating communities.
The idea of an individual 'right' lacks resonance among mobile pastoralists and seasonal migrants. Pastoralist families make decisions about whether an individual child goes to school with a view to the labour requirements (to which children contribute) of mobile animal husbandry-as well as the logistics of accessing provision. The family may decide that its collective interest is best served by selecting some children for school and others for pastoralism, a strategy that may require the structural constraints of school provision to be overcome by household 'splitting' (a child/some children stay in the home village where they can access school, leaving others with animals and out of school) (Dyer 2014). In families undertaking seasonal labour migration, particularly in circumstances of poverty, 'accompanying' children may work, to contribute to the family income. In neither case is childhood a protected space where children have no claims on their time and are free, as individuals, to attend school. Gender norms also play out in decision making about who goes to school: among pastoralists (older) boys are more likely to stay in herding; but generally, education inclusion tilts towards boys, reflecting the gendered nature of domestic work (Srivastava 2012).

Discourses of human capital development, poverty alleviation and reducing inequalities
While a rights-based approach is relatively recent in the region, national policy narratives (GoA 2016; GoB 2010; GoI 2019a; GoP 2018) have long affirmed formal education as an individual and social good, largely with reference to its role in developing human capital for economic growth. Across South Asia, there has also been emphasis on formal education as a means of reducing inequalities associated with social hierarchies, particularly gender and caste. States also use schooling systems instrumentally to pursue goals of national integration and unity. Two examples illustrate: in Pakistan's 2017 -2025 Education Policy, 'Education is the only source of human capital formation and producing responsible citizens in the country' and 'priority goals' include 'poverty alleviation and integrated human development, universalizing access and quality education, women empowerment and elimination of all forms of discrimination, community mobilisation and strengthening partnership of Public and Private Sector' (GoP 2017, 4). In Bangladesh, the 2010 National Education Policy states that 'Education will help [citizens] to grow up as non-communal, patriotic and efficient persons free from superstitions. And simultaneously, it is only education that can equip the nation to acquire the qualities and skills that will strengthen Bangladesh to work with equal capacity and pace of the global community' (GoB 2010, 1).
In practice, because these kinds of policy goals are largely pursued through formal schooling, 'inclusion' is a site of tension between what they state and the ground reality that schooling is itself a structured system of social inequality (Drèze and Sen 1995).
Education inclusion is a process of drawing people into this contested arena; and, amid claims that education acts an equaliser, schooling is part of a state project of educating people into modernity, to improve their capital and capacity as citizens. Since mobilitydependent livelihoods are, as we have shown, located in contradictory ways within the modern nation state, the nature and value of the 'human capital' associated with people pursuing them is also contested. Schooling largely functions to distance people from livelihoods that modernity delegitimises, and holds out the promise of improved social capital as a reward.
There are many reasons why pastoralists' inclusion in late colonial and post-colonial citizenship frameworks is problematic (Bajrange, Gandee, and Gould 2020). A view of pastoralist mobility as 'wandering' in the 'jungle', which is nefarious and/or irrational, is a persistent perception among state officials (Agrawal and Saberwal 2004) which shapes the dominant stance that sedentarisation is desirable (Dyer 2014). Policy discourses favour sedentary livestock husbandry and agriculture, and position mobile pastoralism as irrelevant to the modern capitalist economy. Mobile pastoralists themselves understand the 'capital' that their animals represent in ways that defy the notion of capitalist accumulation on which modern societies depend, and which shape how growth, wealth and poverty are measured within these societies. Rather, for pastoralists, animal holdings traditionally constitute social standing; and pastoralism is a spiritual way of life in which animal wellbeing is both prioritised and a source of human wellbeing (Sharma, Koeller-Rollefson, and Morton 2003).
Pastoralists also develop human capital through education, but not via schooling. The pastoralist workforce is sustainably reproduced through situated learning, where children learn contextually, from experienced others, how to be good pastoralists (Dyer 2014). This way of developing 'human capital' suited to pastoralism may be perceivedincorrectly -by outsiders as children 'wasting time' when they could be at school (e.g. Murtaza, Iqbal, and Khaleeq 2016). Such a perception reflects an undervaluing of indigenous knowledge and the learning of it in an era where 'education' has broadly come to refer to the acquisition of decontextualised knowledge, packaged for age-grade delivery in schools and measured in learning achievements. The erasure of the standing of indigenous knowledge and rise of formal education as a marker of social status have made being 'educated' (schooled) intrinsic to a 'respectable identity' in contemporary South Asian society (Jeffrey, Jeffery, and Jeffery 2004;Dyer 2014).
While mobile pastoralism is marginalised, migrant labour has become central to South Asia's economic growth and development (Deshingkar 2006). Seasonal and short-term migrants are a source of 'cheap' labour in the wider context of increased informalisation and casualisation of the workforce; in India, one study has suggested that they contribute 10 percent to national GDP (Deshingkar and Akter 2009). Yet at the same time, internal migration often entails a citizenship status that is restricted, whether in terms of legality, rights, identity or belonging (Abbas 2016;Srivastava and Pandey 2017). Temporary/ seasonal labour migration is a survival strategy in rural areas for the poorest and socially marginalised people, who have the least skills and schooling (De Haan and Yaqub 2010; Keshri and Bhagat 2012;Srivastava 2012). Precarity and informality of labour shape circumstances of mobility: in circumstances of poverty, extended family support for child care at home may be lacking, so children migrate with parents (as 'accompanying' children) and contribute to the family's survival, if not directly by paid work, then through enabling work such as sibling care, watching over the tents while adults are out working, and doing domestic tasks such as cooking, cleaning and fetching water (Rajan 2018). Work, whether paid or unpaid, and intergenerational dependence, are integral to the lives of children living in such economically and socially marginalised contexts (Heissler 2017).
Perceptions of the value of formal schooling, and its relationship to both economic productivity and social status, vary significantly according to context (Heissler 2017;Rao 2009Rao , 2010. Where the informal labour market structure provides children in seasonally migrating families with good opportunities for employment and sale of produce, working can be more attractive than schooling: processes of modernisation are opening channels -such as consumption -that may more effectively and quickly deliver gains to social status and incomes than schooling, if local labour markets are conducive (Rao 2009). Poverty and uncertain contexts of mobility often adversely affect families' opportunities to invest in schooling (GEMR 2019). Migrant remittances can positively influence access and closing the gender gap of children who remain in the home village, as Mansuri (2006) found in rural Pakistan, although poor quality residential provision to support schooling is widely reported (Shah 2015). Coffey's (2013) quantitative study in 70 villages in rural north-western India with 1,980 children in short-term migrant families found a significant statistical difference between the educational attainments of migrating and non-migrating children (see also Cameron 2012). 'Accompanying' migrant children who do manage to access formal schooling, on and off, experience huge learning gaps -caused by schools' rigid structure and organisation, and lack of quality -that contribute to the likelihood of their dropping out and engaging in child labour (Roy, Singh, and Roy 2015;Schapiro 2009).
There is, then, a deep fissure between global and national policy discourses and actual manifestations of educational inclusion. The normative imaginations of inclusive policy are mediated through structural inequalities that largely serve to exclude migrant populations from formal education.

The ontological crisis of modern education
As preceding sections have shown, South Asia's current development regimes situate mobility-dependent livelihoods in complex and contested positions. Their schooling systems embody a fundamental 'ontological crisis' (Rajan 2020), in that they offer the promise of inclusion for all on terms that are persistently exclusionary for migrating children (Dyer 2013), and appear unable to evolve a 'mainstream' model that can accommodate these learners.
A strong policy focus on universalising access, reflecting the priorities of the second Millennium Development Goal, underlined that guaranteeing 'attendance' and 'completion' of elementary education for migrating children is a massive task (Chandrasekhar and Bhattacharya 2018). The geo-spatial zones that are home to mobile pastoralists and many seasonal labour migrants are often known to others as 'interior' or 'remote rural' locations. Such locations typically have a low population density, and common infrastructural deficits in education systems including teacher shortages, poor material conditions, absence of water and sanitation facilities, and monitoring challenges. For migrant learners, additionally, standard features of formal schools, such as daily timing, annual calendar, requirement of daily presence, and age for grade structure of progression, are 'terms of inclusion' (Dyer 2013) that may exclude them altogether, or fuel patterns of initial enrolment, irregular attendance and subsequent drop out. The mutually constituted and 'immobile' ideals of modern schooling (such as age, grade, learning level, curriculum, language of learning, externally appointed teachers, bureaucratic governance) position migrating children in a constant state of facing, and attempting to mitigate, a 'learning crisis' (Rajan 2020). Within schools, informal practices of exclusion and discrimination persist: arrangements to help learners catch up with missed work are usually lacking; teachers may be reluctant to re-enrol children who have temporarily left or to take responsibility for taking extra measures for migrating children (Coffey 2013); and if a child is re-enrolled, missed learning may lead to low performance and demotivation, or indeed the child being blamed for falling behind (Smita 2008).
Exclusion of migrant children from formal schooling is not limited to rural areas. For temporary migrant families, the presumed epitomes of the city as a site of progress and equitable access, and of migration to the city as a process of optimal labour allocation and human capital accumulation, rarely hold true. These families are often pushed into the urban margins, which in turn hinders children's physical access to schooling. Even for those who gain physical access, social and pedagogical inclusion is a huge challenge, given their marginal social contexts and, often, language barriers in the destination cities (Bengtsson and Dyer 2017;Cameron 2012).
The argument of ontological crisis, however, is not only about these exclusions and limitations. It is also about the need for, and limitations of, 'alternative' provision, which gained ground in during the EFA movement. Internally migrating children in South Asia have been offered inclusion via adjuncts to geo-spatially fixed provision of day schools, such as residential schools, hostels, bridge schools, and some attempt to develop 'networked' schooling (Dyer 2014;Dyer and Echessa 2019); and non-formal or 'alternative' basic education (ABE) which includes fixed and mobile provision. These variations are, nevertheless, largely bound in the sedentary imaginaries of formal education as schooling, which complicates the inclusionary claims of EFA that they put forth, while open learning for migrating children remains significantly under-explored (Morpeth et al. 2009)

Adjuncts and in-system adaptions
Residential facilities offer continuity in fixed-place provision but are often also places of discrimination, limited support, and physical and sexual exploitation. Establishing a seasonal hostel, or residential centre, in the place of permanent residence helps children to remain at school (Smita 2008). In some seasonal labouring families, this removes the possibility of exposing children to risk in 'hazardous' work environments, and sidesteps the problem of commonly poor access to schooling in destination sites (Bengtsson and Dyer 2017). The positive intention of assisting children identified by the state as needing special protection can perpetuate rather than address social segregation, as Rao (2006) found in relation to pastoralists in Jammu & Kashmir, although conversely, Dyer (2014) reports that segregation and supporting community values are essential to the success of a pastoralist residential school in Gujarat. India's Kasturba Gandhi Balika Vidyalaya scheme, in which a reported 5,970 residential schooling facilities are said to reach 725,000 girls from poor socioeconomic backgrounds (GoI 2019b), is critiqued for being operated through patriarchal assumptions of protection, surveillance and alienation (Saxena 2012). For many migrating children, the convention of enrolling in one school for the whole year is persistently implicated in the production of being 'left behind', as it contradicts the realities of their lives. Networked schooling (Dyer 2014;Dyer and Echessa 2019), is an innovation that enables a child to enrol in a succession of schools across the network of provision, and as such replicates both the advantages and disadvantages of that provision. A brief attempt to do this in government schools in Gujarat in India was discontinued because state actors felt the level of support required exceeded state capacity to plan and manage resources effectively (Dyer 2014). Given the pervasive 'learning poverty' across South Asia (Beteille et al. 2020) which reflects weak basic infrastructure and poor learning conditions in government schools, the addition of temporary learners further exacerbates fragility (Bengtsson and Dyer 2017); funding norms are not designed to accommodate variability of this nature; and temporary presence in a school does not necessarily lead to social inclusion or learning.
In places of high outmigration, bridge schools -which India's 2009 RtE Act, for example, explicitly recommends -provide short-term, catch-up provision aiming to support reintegration into the 'home' school. Bridge schools tend to be stop-gap arrangements in which practices are disparate, quality is uneven, and continuity is not monitored: some offer little more than nutrition and recreation (Rajan 2018) while others report that they have enabled thousands of children to reintegrate successfully into their 'home' school after absence (in Bengtsson and Dyer 2017). The need for support for 'accompanying' children to attend school in the place of destination is less recognised (Babu et al. 2017), although the 'Urban Slum School' and 'Bridge School' projects run jointly by the NGOs BRAC and Educate A Child in cities in Bangladesh are examples of such interventions (EAC, n.d.b, n.d.c).

Alternative basic education
The EFA call in 1990 for flexibility and recognition of disparate learning needs signalled some recognition of structural constraints, but in the name of 'reaching' the 'hard to reach', rather than addressing the fundamental ontology of schooling, EFA legitimated alternative provision that fails to challenge the 'exclusionary workings of mainstream provision' (Dyer 2014, 184). South Asia now has a proliferation of 'alternative' provision, such as Community Based Education (CBE) in Afghanistan and Alternative Learning Programmes (ALP) in Pakistan, although India's Alternative and Innovative Education (AIE) has been discontinued: the 2009 RtE Act's notion of 'quality' and 'neighbourhood schooling' precluded the flexible parameters for responsive provision that characterised AIE under the Sarva Shiksha Abhiyan framework, and legally required it to be phased out.
In terms of access dimensions of being 'left behind', ABE provision can respond to mobile learners by being mobile itself, in a tent, boat, or bus (e.g. Suliman, Shah, and Ullah 2017), and by accompanying learners or bringing resources to where they are. Bangladesh's boat schools were innovated to serve the peripatetic Bede community; and in the Haor basin flood plain, a collaborative 'Boat School' project targets 13,000 'hard to reach' OOSC (EAC, n.d.a). The capabilities of mobile provision may not, however, extend to match complex learner mobility. Experience from India and Afghanistan shows that a constellation of mobile pastoralist learners may fragment in the search for natural resources (Dyer 2014;Sharifi 2013), leaving teacher(s) unable to follow all learners. A hybrid 'mobile' school model provided by the government for Bakkarwal pastoralists in India's Jammu and Kashmir, which comprised a home and a seasonal school in the high pastures, was found to offer 'pathetic' conditions in the seasonal school component: a lack of infrastructure, despite government claims that tents and other materials had been provided; 'haphazard' learning in disorganised multi-grade, multi-subject teaching; and a demotivated, underpaid teacher in provision that the state failed to monitor (Suri 2014).
ABE claims to offer flexibility over curricular content and pace, and to respect community social values. Where girls are 'left behind', these kinds of attributes may be conducive to enrolling and retaining them, as Afghanistan's CBE for newly settled pastoralist communities, run by the Swedish Committee for Afghanistan in close collaboration with the government, has shown (Dyer 2014). All states place extensive reliance on partnership with civil society organisations for funding and delivery, which raises questions about sustainable funding. ABE delivered through such partnership is now widely conceived as a 'second chance' education (Banerji 2015) that uses accelerated learning and is intended to enable catch up and transfer into formal provision. Pakistan, for example, has about 16,000 non-formal basic education institutions, largely comprising community-based schools: accelerated learning is the dominant strategy within such provisioning (UNCF 2017). This widespread policy move affirms the hegemony of the mainstream -and sedentary -school curriculum, and values underpinning it, as the currency of universal inclusion.
Regional scholarship is critical of the often sporadic and piecemeal approach of ABE, and raises questions over quality, equity, sustainability and social integration. Rampal (2000), for example, critiques the politics embedded in the flexible curriculum of 'those' children and the 'highly guarded curriculum' of 'our children'; and the inability of non-formal provision to influence and change formal provision, which Dyer (2014) also highlights. Resorting to differential provisioning to mitigate leaving some children behind is in itself evidence of the ontological crisis of formal education systems, although not recognised as such in differing national positions on whether 'alternative' provision is a legitimate approach to inclusion. Structurally, ABE does not comprise an alternative education system, since it focuses only on 'basic', i.e. lower levels, and then relies on 'mainstream' schooling, with all its attendant exclusionary terms intact, to enable progression.
There is a strong tradition of Open Learning in the region (e.g. Bughio, Qazi, and Shah 2014;Morpeth et al. 2009). Despite this, an open learning system for internally migrating children, which might utilise the many affordances of e-learning to respond to realities of geographic flexibility, language of instruction, flexible admission and programme completion, timing and individualised pacing, has yet to be fully investigated or proactively supported.

Conclusion
This paper began with a concern that people who migrate internally and pursue mobilitydependent livelihoods are being pushed outside the current purview of migration studies and marginalised yet further in policy gazes by the elision of migration with refugees and internally displaced persons. The role of education in social integration and economic mobility, and learning equivalence/transferable qualifications, are both widely discussed in relation to international and refugee migration (Nicolai, Wales, and Aiazzi 2017). Such discussion is equally pertinent for people who migrate within a country, but still -at best -is at only a nascent stage (cf. Cameron 2012; GEMR 2019).
With respect to South Asia's internally migrating children, there are significant gaps in the evidence needed to address the challenges associated with 'leaving no-one behind'. Indeed, difficulties of evidence, albeit in our desk-based searches of publications in English, are in themselves a dimension of being 'left behind'. Official data have gaps and elisions that serve to perpetuate invisibility to policy, or misrecognition. The 'grey' literature, which provides some useful information, is diffuse, often anecdotal and also often written in a 'reporting' format that focuses on numbers, targets and outcomes. Project websites offer insights into intentions, but rarely provide detail on progress. Reporting on projects that target 'left behind' children displays a conspicuous absence of reflection on lessons learned about the contradictions of education provision that these projects surely encounter. Academic literature around the nature and extent of relationship between internal migration and education in South Asia is slim. Insight about pastoralists comes only from very few, small scale, ethnographic studies, while short-term seasonal migration has attracted some quantitative work at scale (e.g. Coffey 2013; Cameron 2012) which establishes a significant gap in outcomes attainment between migrating and non-migrating children. Overall, systematic, scholarly studies that provide detailed empirical findings about the lived realities of internally migrating children and their experiences of education inclusion are in very short supply.
While stronger evidence and better data have roles to play, other significant challenges to the Agenda 2030 pledge, which are highlighted as ontological in this paper, are at least as pressing. Our first case shows that mobile pastoralism is a productive livelihood in which situated learning is embedded, yet frameworks of the modern economy operate to delegitimise this production system and associated cultural values, and to narrow the frame of what counts as 'education'. Formal education represents an exit route, rather than a resource for pastoralism that can augment situated learning in useful ways, such as engaging with print and e-technology. Our second case highlights that for people who use seasonal/short-term/circular migration to access alternative income-generating opportunities, these opportunities tend to situate them in ways that perpetuate intersecting inequalities -of caste, class, gender, place, and poor access to the formal education system. These two cases, with all their contrasts and similarities, show that the ontological foundations of formal education systems within South Asia render those systems inadequate in design to incorporate the lived realities and educational needs of very large numbers of children in the region.
The principles of 'leave no-one behind' and 'reaching the last first' are the core moral exigencies of the 2030 Agenda for Sustainable Development. This paper has interrogated some of the socio-political and moral contingencies that shape the production of a 'left-behind child' in South Asia when that child is in a family which uses mobility as a livelihood strategy. We have shown a range of ontological dissonances between realities for these families and policy discourses -around norms of childhood, human capital, rights, citizenship and legitimate 'modern' livelihoods. This leads us to conclude that contemporary formal education systems are embedded in, and perpetuate, parameters of 'modernity' that construct 'the last' by being antithetical to, and/or exploitative of, particular livelihoods, discounting indigenous knowledge, and perpetuating hierarchical social relations. In so doing, they tend to reproduce rather than interrupt existing socio-cultural and economic inequalities. In calling this an 'ontological crisis' of formal education systems, we underline the seriousness and scale of the disjuncture between these realities, the aspirations of SDG 4 and the moral exigencies of Agenda 2030; and we signal that this crisis is a productive opportunity to act for change.

Disclosure statement
No potential conflict of interest was reported by the author(s).