¹ See for instance Serra, Renata, An Economic Analysis of Child Fostering in West Africa (Unpublished Ph.D. dissertation, Faculty of Economics and Politics, University of Cambridge, 1996)Google Scholar; Kielland, Anne and Sanogo, Ibrahim, Burkina Faso: Child Labor Migration from Rural Areas (Washington DC: World Bank, 2002)Google Scholar.

² Kanbargi, Ramesh, Child Labour in India: Carpet Weaving in Benares, in Bequele, Assefa and Boyden, Jo (eds): Combating Child Labour (Geneva: International Labour Office, 1988)Google Scholar.

³ This is a rapidly evolving field of research. Contributions within economics, sociology, geography and anthropology, include Iversen, Vegard, ‘Autonomy in Child Labor Migrants’, World Development, 30 (5): 817–834, 2002Google Scholar; Punch, Samantha, ‘Youth transitions and interdependent adult-child relations in rural Bolivia’, Journal of Rural Studies, 18: 123–133, 2002Google Scholar; Iman Hashim, ‘Child Migration: Pathological or Positive?’ (Mimeo, University of Sussex, 2003); Young, Lorraine, ‘Journeys to the Street: the complex migration geographies of Ugandan street children,’ Geoforum, 35: 471–488, 2004Google Scholar; Whitehead, Ann and Hashim, Iman, Children and Migration (London: Report to the migration team, Department for International Development, 2005)Google Scholar. See also the papers presented at the workshop: ‘Children on the Move in the Developing World: Sharing Research Findings’ at University of Sussex, May 2008, which are available from www.migrationdrc.org. For a recent study using a large scale data-set to identify child migration determinants in North India, see Edmonds, Eric G. and Salinger, Philip, ‘Economic influences on child migration decisions: evidence from Bihar and Uttar Pradesh,’ Indian Growth and Development Review, 1 (1): 32–56, 2008Google Scholar.

⁴ Jeffery, Chris and Sherington, Geoffrey, Empire and Child Migration (London: Woburn Press, 1998)Google Scholar; Coldrey, Barry M., ‘“a place to which idle vagrants may be sent.”—the first phase of child migration during the seventeenth and eighteenth centuries’, Children and Society, 13: 32–47, 1999Google Scholar.

⁵ Retrospective data collection involves methodological hazards: In an empirical assessment of recall quality, James P. Smith and Duncan Thomas: ‘Remembrances of Things Past: Test-Retest Reliability of Retrospective Migration Histories’, Journal of the Royal Statistical Society, Series A, forthcoming, found that respondents typically remember salient moves and usually forget the details of local or short-term migration. The migration episodes reported in this paper were mostly life-changing events.

⁶ Interviewees were selected using a two stage design. In the first stage, we purposively selected eight study villages—four in Karnataka's Coastal belt (now Udupi district) and four in South-Central Karnataka (Mandya district). In the second stage, we identified individuals with a history of child labour migration from randomly selected households within each of these eight study villages. For maps of the study areas and a more detailed methodology discussion, see below.

⁷ Candidate foci would include interdependencies and youth transitions to adulthood, e.g. Punch, ‘Youth transitions’, child labour causes and outcomes, e.g. Basu, Kaushik, ‘Child Labor: Cause, Consequence and Cure, with Remarks on International Labor Standards,’ Journal of Economic Literature, 37 (September): 1083–1119, 1999Google Scholar; migration and intergenerational contracts, e.g. Kabeer, Naila, ‘Intergenerational Contracts, Demographic Transitions and the “Quantity-Quality Tradeoff: Parents, Children and Investing in the Future’, Journal of International Development, 12: 463–482, 2000Google Scholar or within economics and related to the latter, migration as a cooperative household venture, e.g. Stark, Oded, The Migration of Labor (Cambridge, Massachusetts: Basil Blackwell, 1991)Google Scholar.

⁸ Throughout, respondent names have been made anonymous.

⁹ See Nieuwenhuys, Olga, Children's Lifeworlds: Gender, Welfare and Labour in the Developing World (London: Routledge, 1994)Google Scholar: , Iversen, ‘Autonomy in child labor migrants’; Masako Ota, Between Schooling and Work: Children in Rural Andhra Pradesh (Unpublished Ph.D. dissertation, School of Development Studies, University of East Anglia, 2002)Google Scholar; Hashim, Iman, Working with Working Children: Child Labour and the Barriers to Education in Rural Northeastern Ghana (Unpublished Ph.D. dissertation, University of Sussex, 2004)Google Scholar.

¹⁰ ‘Udupi hotel’ or just ‘hotel’ is local terminology for eating places serving tiffin (snacks) and/or South-Indian thalis (meals). For an illuminating account of their origins, see Stig Toft Madsen, Udupi Hotels: A Case of Ethnicity Based Non-Virulent Development (Paper presented at the First Nordic Conference on South-Asian Studies, 1991). See Iversen, Vegard and Raghavendra, P.S., ‘What the signboard hides: Food, caste and employability in small South-Indian eating places,’ Contributions to Indian Sociology, 41 (3): 311–341Google Scholar, for a study of the link between caste and employability in this industry.

¹¹ A feasible empirical project would thus be to test for linguistic proficiency and proxies for attitudinal and other transformations, the null hypothesis being that other things being equal, the linguistic absorptive capacity of children exceeds that of other migrants while impressionability could, in principle, generate more rapid attitudinal or other change. Similar tests could be implemented for other categories of workers, across sectors, workplace compositions and educational levels. Children may, of course, be seriously disadvantaged as well—systematic comparisons of a wide range of career and well-being outcomes (vulnerability to drug and substance abuse, to violence and physical abuse etc.) with other migrants would be of interest but was well beyond the scope and resources of the present study.

¹² For a fine review, see Gardner, Katy and Osella, Filippo, ‘Migration, modernity and social transformation in South-Asia: an overview’, Contributions to Indian Sociology, 37: v–xxvii, 2003CrossRefGoogle Scholar.

¹³ ibid; p. xiv.

¹⁴ For more on Mumbai's history of union activism, see van Wersch, Hubert, The Bombay Textile Strike 1982–83 (New Delhi: Oxford University Press, 1992)Google Scholar; Chandavarkar, Rajnarayan, The origins of industrial capitalism in India—Business strategies and the working classes in Bombay 1900–1940 (Cambridge: Cambridge University Press, 1994)Google Scholar; Pages, Carmen and Roy, Tirthankar, ‘Regulation, Enforcement and Adjudication in Indian Labor Markets: Historical Perspective, Recent Change and a Way Forward’, in ‘Meeting the Employment Challenge: Conference on Labour and Employment Issues in India’ (New Delhi: Institute for Human Development, 2006)Google Scholar.

¹⁵ The city's first Kannada Night School, run by the Mogaveera (the fisherman caste of Karnataka's Coastal belt) community, opened its doors in 1918. As far as we know, the Mogaveeras were also the first Coastal community to establish their own society in Mumbai in 1902. Such urban societies and associations, organized along caste or religious lines, would offer lodging and other important assistance to migrants (including, amongst them, some of our 90 interviewees). Caste associations were not unique to Mumbai, and Bangalore had many including the Mysore Lingayat Educational Fund Associations (est. 1905) and the Vokkaligara Sangha (1906); see Nair, Janaki, The Promise of the Metropolis—Bangalore's 20^th Century (New Delhi: Oxford University Press, 2005), p. 397Google Scholar. See Thimmaiah, G., Power, Politics and Social Justice: Backward Castes in Karnataka, (New Delhi: Sage Publications, 1993), pp. 70–72Google Scholar for a list of early twentieth-century caste associations in Bangalore.

¹⁶ Between 1901 and 1931, the percentage of migrants from the then Madras Presidency in Mumbai's migrant population increased from 1.01 to 2.45. By 1961, Mysore state accounted for 6.17% of the city's male migrants—the largest migrant groups were natives of UP-Bihar (16.1%) and Gujarat (14.96%), see Zachariah, K.C., Migrants in Greater Bombay (Bombay: Asia Publishing House, 1968), p. 52Google Scholar.

¹⁷ Fernandes's standing among Bombay's workers and others is perhaps best illustrated by his 1967 shock election defeat of the overwhelming favourite, S.K. Patil.

¹⁸ Such backlashes are not unique to Mumbai; in Bangalore disturbances have often been prompted by Kannada–Tamil political conflicts over Kaveri water (Nair, The Promise of the Metropolis, p. 258) or in a more recent instance, and dramatically, by the kidnapping of the Kannada film icon Rajkumar by the notorious (now late) sandalwood smuggler Veerappan.

¹⁹ Giddens, Anthony, The consequences of modernity (Cambridge: Polity Press, 1990)Google Scholar.

²⁰ Monastic and religious (Hindu) institutions with a distinguished history. The first four were established by Shankara in the eight century. The important mathas have significant assets and considerable political clout. See Prasad's, Leela account of the Shankara matha in Sringeri in Ethics in Everyday Hindu Life – Narration and Tradition in a South Indian Town (Ranikhet: Permanent Black, 2007)Google Scholar

²¹ Note that 1951 observations are missing.

²² Madras Presidency administrators closely watched female/male ratios, which were interpreted as reliable indicators of out-migration, see Chapter 5, Census Report, Madras Presidency, 1931. The 1931 female surplus figure for South Canara (Dakshin Kannada) District as a whole, about 1,075, was well below the figures for Udupi and Kundapura taluks.

²³ In comparison, the sex ratios in the dry-land taluks in Northern Mandya district, Krishnarajpet and Nagamangala, were 1.015 and 1.032 in 2001.

²⁴ For an account of the role of unions in the life of factory workers in Bangalore see Holmstrom, Mark, South Indian factory workers—Their life and their world (Cambridge: Cambridge University Press, 1976), pp. 65–71CrossRefGoogle Scholar.

²⁵ The sectors of employment of unskilled South-Indian migrants vary systematically with place of origin and by age. While male migrants from Andhra Pradesh clustered in spinning, weaving and as manual labourers, those from Mysore state were concentrated in services and clerical jobs, see Zachariah, Migrants in Greater Bombay, p. 315. For the under-15, male working population, occupational clustering was even starker—90% of these workers were household domestics and eating place employees in 1961 (ibid. p. 273). A similar pattern was discernible in Bangalore in the late 1990s; Iversen, ‘Autonomy in child labor migrants’.

²⁶ Vemagal Somshekar, Hotel sector (Udayama), in L.S. Shesagiri Rao (ed.): Bangaluru Darshana, Vol. 1 (Bangalore: Udaya Bhanu Kalasangha, 2005), pp. 246–260.

²⁷ Geetanjali Salian, pers. comm..

²⁸ Toft Madsen, ‘Udupi hotels’, p. 10.

²⁹ We are not aware of any systematic data on ownership of Udupi hotels disaggregated by social groups in Bangalore. There are around 12,000 small eating places in Greater Mumbai among which perhaps as many as 6,000–7,000 may be South-Indian hotels. The Bunt community may control as much as 70% of the latter eating places (Toft Madsen, Udupi hotels; Iversen and Raghavendra, ‘What the signboard hides’).

³⁰ For more detailed accounts and evidence on the caste dimensions of work in these eating places, see Toft Madsen, ‘Udupi hotels’ and Iversen and Raghavendra, ‘What the signboard hides’.

³¹ Guilmoto, Christophe Z. and Rajan, S. Irudaya, ‘District Level Estimates of Fertility from India's 2001 Census,’ Economic and Political Weekly, February 16: 665–672, 2002Google Scholar.

³² Traditional toddy-tappers.

³³ Here and later, HH is short for household.

³⁴ Srinivas, M.N., The Remembered Village (New Delhi: Oxford University Press, 1976)Google ScholarPubMed.

³⁵ Work-life histories, based on in-depth interviews with each identified child labour migrant should ideally have been collected. Due to resource limitations, we settled for a sub-sample of 55 cases or 71.4% of the total Coastal belt sample with a slight bias towards migrants residing in Mumbai, Bangalore, Udupi or Shimoga (destination area for agricultural labour migration) district at the expense of a small number of migrants in Hyderabad and individual migrants in Gujarat, Pune, Nasik, Karwar and a few other destinations. In Mandya, 35 of the 43, i.e. 81%, of the identified individuals with a history of child labour migration were traced and interviewed in depth.

³⁶ In technical terms, this is a cross-section where household members are individuals residing in the household or offspring of the household head. In the event of extended families where, say, three brothers and their wives co-reside with the parents of the brothers, the offspring of these smaller units would also be considered household members. The latter is consistent with established migration research practice, see for instance, Winters, Paul, de Janvry, Alan and Sadoulet, Elisabeth, ‘Family and Community Networks in Mexico-US Migration’, Journal of Human Resources, 36 (1): 159–184, 2001Google Scholar. While the ideal would be a longitudinal study resembling Breman, Jan, Footloose Labour—Working in India's Informal Economy (Cambridge: Cambridge University Press, 1996Google Scholar) or Breman, Jan, The Poverty Regime in Village India (New Delhi: Oxford University Press, 2007)Google Scholar, and where a sample or panel of individuals or villages are tracked over time, this was not, for obvious reasons, a feasible option for studying the history and relevant dimensions of child labour migration from our source areas. Our retrospective cross-section seems the most appropriate alternative but has some limitations, as explained below.

³⁷ One of these local informants, a returnee with more than 20 years’ migration experience from Mumbai, accompanied us during several visits to the city and introduced us to migrants from his village whom we had scheduled to interview.

³⁸ Osella, Filippo and Osella, Caroline, Social Mobility in Kerala—Modernity and Identity in Conflict (London: Pluto Press, 2000), p. 76Google Scholar.

³⁹ The estimate is of the fraction of sampled households with a child labour migration episode—hence 9.3% of the randomly sampled households in Innanje and 26% of those in Yeljith reported such episodes.

⁴⁰ Numbers in {} refer to girl migrants.

⁴¹ The extensive time-period covered means that households that (now ageing) child labour migrants left behind were dissolved a long time ago. That old individuals with a history of child labour migration were observed only in the Coastal belt suggests that child labour migration has a longer history there but could also reflect differences in return migration—hence, if there were also such early migrants in Mandya and these did not return to spend their old age in their native villages, a similar pattern to that observed would emerge.

⁴² Zachariah, K.C., Historical Study of Internal Migration in the Indian Sub-Continent 1901–31 (Bombay: Asia Publishing House, 1964)Google Scholar; Chandavarkar, The origins of industrial capitalism.

⁴³ Here, as in the following tables, some migration episodes have multiple causes with the two most important being reported. Numbers in () denote observations that are part of a multiple-cause explanation. For instance, while taking up caste occupation is never a single cause explanation, being an educational misfit usually is. And so forth.

⁴⁴ Iversen, ‘Autonomy in child labour migrants’. There is only one runaway observation in the Coastal belt sample.

⁴⁵ Note that intergenerational quarrels and tensions are also an important adult migration cause in South-Asia. See Gardner and Osella, ‘Migration, modernity and social transformations’, p. xix, and the references therein.

⁴⁶ By educational ‘misfit’ (our term) we refer to children lacking a basic interest in the education they are offered—their attendance is typically highly irregular. A school ‘failure’ (our term) will have failed at least one examination. In the Coastal belt, having to resit an examination or redo a school year is often associated with considerable loss of face and may instil a determination to discontinue education also in individuals with an initial interest in schooling. As the case of Manju attests to, a ‘misfit’ is sooner or later likely to become a ‘failure’.

⁴⁷ The taluk headquarter of Karimnagar district, Andhra Pradesh, about five hours by bus from Hyderabad.

⁴⁸ A comment on the prevalence of beatings by parents, teachers and employers is appropriate. While suffering from methodological weaknesses, the Study of Child Abuse in India (New Delhi, Ministry of Women and Child Development, 2007) provides the best available documentation to-date of the extent and nature of physical punishment and abuse of children in India. For a sample of about 12,500 children and with, at times, notable variation across states, lessons of particular relevance to the present discussion may be summarized as follows: the most common form of physical punishment reported by 60% of the child respondents is within the home and with slaps or kicking being the most common form (75%). Physical punishment is more common for the younger (5–12 years) age group and for boys. Corporal punishment in schools is widespread but more common in government schools and with distinct inter-state variation. Boys are more likely to be punished by teachers than girls (which may simply reflect that girls are more dutiful in attending to homework), but with considerable variation across states. In the workplace, younger girls working as domestic servants were more likely to be physically punished with the overall incidence being quite low (around 15%). Twenty-two and a half per cent of workers in teashops, dhabas (roadside eating places) and restaurants(occupations strongly dominated by boys—85%), reported physical abuse, with the highest incidence for the younger age group. Iversen, ‘Autonomy in child labour migrants’, p. 820, argued that girls working as domestics were more vulnerable to physical abuse by employers than boys working in South-Indian hotels in Bangalore. The girls interviewed then were in the early stages of their working careers while those interviewed this time (altogether seven) are more experienced. It would seem that while early years involve more physical punishment from employers, relations improve with the passage of time. This could reflect that the girls make fewer mistakes as they grow older and gain experience, or that the ‘human factor’ kicks in: as the examples of Kavita and Gowramma below attest to, young domestics may, over time, develop strong emotive and reciprocal bonds to members of the employer household. While boys working in hotels had exit options and could abandon abusive employers with ease back in 1998 (ibid.), domestic workers, especially in the early stages of their careers, were more vulnerable because of their lack of a similar option. A gradual strengthening of emotive ties may gradually reduce such vulnerability.

⁴⁹ Ramesh struggled with mathematical tables, English copy-writing and was also beaten for not being able to answer questions in class.

⁵⁰ Selling paan (betel leaf and areca nut plus other ingredients) for chewing and cigarettes.

⁵¹ Iversen, ‘Autonomy in child labor migrants’.

⁵² See for instance Basu, Kaushik and Van, P.H., ‘The Economics of Child Labor,’ American Economic Review, 88 (3): 412–427, 1998Google Scholar.

⁵³ See also Ota, Between Schooling and Work.

⁵⁴ This weakness of girls’ pre-migration agency in our accounts contrasts with Nieuwenhuys, Olga, ‘The domestic economy and the exploitation of children's work: the case of Kerala,’ The International Journal of Children's Rights, 3: 213–225, 1995CrossRefGoogle Scholar, observations of migration by young girls from fishing communities in Kerala to prawn-curing factories in distant Gujarat. This contrast does not, of course, rule out the possibility that girls, after migrating and over time, come to experience migration as advantageous and as furthering their interest in relation to marriage, work and more generally as brought out by the cases for the same Kavita and Gowramma below.

⁵⁵ One should avoid pushing the suggestion of new behavioural patterns too far: In its 7 November 2008 editorial honouring the 86 year-old classical singer, Bhimsen Joshi, the latest recipient of the Bharat Ratna, The Herald (Panjim) observed: ‘Born into a Brahmin family at Gadag—in what is now Karnataka—. . . he was obsessed with music even as a child, to the utter dismay of his father, a schoolteacher, who wanted young Bhimsen to become a doctor or an engineer. But the boy, deeply moved by a recording of Abdul Kharim Khan, the great founder of the Kirana Gharana, rebelled and ran away from home. He headed for [distant] Gwalior which he had heard was one of the best places to learn classical music. He was only 11 then.’

⁵⁶ In 1998 about 20% of boy migrants from Mandya were runaways.

⁵⁷ An anonymous reviewer suggested that running away could be interpreted as a response to a failed attempt to secure a cooperative outcome, while Iversen, ‘Autonomy in child labour migrants’, interpreted running away as a revolt against the content of an intergenerational contract: in either case, it would be possible to perceive running away as a signal of weakness rather than strength. Here, as in Iversen, ibid., running away signals assertiveness and having the upper hand; In 1998, 75% of the migration cases defined as ‘autonomous’ (and where runaways form a subset) involved migration where urban residence or work either directly conflicted with parental preferences or where search parties were launched to trace the migrants down (ibid., 821). The latter is reinforced by the often considerable and usually unsuccessful parental attempts to bring runaways back into the household fold as documented in the narratives in this paper.

⁵⁸ The local terminology is accurate: a ‘supplier’ takes orders and serves food, but unlike a waiter, will not clear tables; the latter is the job of ‘cleaners’.

⁵⁹ Pages and Roy, ‘Regulation, Enforcement and Adjudication’, p. 3.

⁶⁰ Geetanjali Salian, pers. comm.

⁶¹ This is illustrated by the recent migration of Kumara who left his canteen job in Mumbai after eight months because of his inability to understand customers speaking Marathi and co-workers from other states communicating in Hindi. The employer and his manager relative would speak to Kumara mostly in Tulu and Kannada but they were often not present.

⁶² Review of Education in Bombay State 1855–1955 (Government of Bombay, 1956).

⁶³ Ibid, p. 351.

⁶⁴ Ibid, p. 352.

⁶⁵ Ibid, p. 353.

⁶⁶ Toft Madsen, ‘Udupi hotels’.

⁶⁷ Note, though, that according to the 1991 Census, only 35% of Bangalore's residents declared Kannada to be their mother tongue; the corresponding figures for Tamil, Urdu and Telugu were 25%, 19% and 17%, respectively (Nair, The promise of the metropolis, p. 246).

⁶⁸ Traditional theatre form that combines dance, music, speech, costume and stage technique in a unique style (Source: www.yakshagana.com).

⁶⁹ They could also, as another respondent observes, involve discussions of ‘peculiar’ customers. The same respondent also specifically mentions that Raja, a supplier at Shivaji hotel, would secretly write stories in Kannada while on his own in the staffroom.

⁷⁰ This was Shivanand's third migration destination and the first where his father was also present.

⁷¹ Iversen and Raghavendra, ‘What the signboard hides’.

⁷² Ibid, p. 338.

⁷³ Punch, ‘Youth transitions’.

⁷⁴ Kabeer, ‘Intergenerational contracts’.

⁷⁵ Gardner and Osella, ‘Migration, modernity and social transformations’.

⁷⁶ Punch, ‘Youth transitions’, p. 124.

⁷⁷ It is worth emphasizing that the relevant decisions tally with what Naila Kabeer interprets as strategic life choices in ‘Resources, agency, achievements: reflections on the measurement of women's empowerment’, Development and Change, 30: 435–464, 1999.