I Rural- Urban

I IntroductionIt is often argued that migrants from rural areasform only a small part of current developing coun-tries' urban population growth. Still, they consti-tute a significant group for the cities' economies,and, possibly, for poverty reduction policies. Andurbanisation is expected to accelerate in countrieslike India and China, and migration is likely toincrease. Moreover, the total number of migrants ismuch larger than urbanisation and net migrationfigures suggest. In many developing countries,migration has a 'circular character'; rural-urbanmigrants do not settle permanently in cities butcontinue to maintain close links with their areas oforigin. They return regularly and after retiring, andremit substantial parts of their income.

Migrants, particularly the ones who continue tomove between rural and urban areas, are a difficultgroup to analyse and to formulate policies for.Although there is a large amount of literature onthe background of migrants, conclusions differ.There is agreement that migration is a selectiveprocess, that - as in Europe in the nineteenth cen-tury - growing cities in Third World countries arefull of young male adults (Williamson 1988: 430ff.). But this apart, contrasting pictures of migrantsexist. For example, some of the literature on migra-tion, and popular opinion, portray migrants inThird World cities as destitutes and their migrationas a last resort - a forced move from the country-side where they had no alternatives.1 Other authorssee migrants as rational actors, as individualsresponding to income incentives in their decisionsto migrate.2 Yet another strand in the literatureargues that the poorest cannot migrate.3 Someauthors have concluded that better-off villagerstend to be pulled, and worse-off villagers pushed,and that therefore town-ward migration increases

E.g., Singh 1995; Parnreiter 995; Firdausy 1994;Chapman and Prothero 1990; Breman 1985 and 1990.Much of the Indian historiography, and indeed colonialreports like the Royal Commission (1931), stress that itwas the poorest who were 'pushed' from the rural areas.

'Harriss and Todaro 1970; Stark 1991.

E.g., Chaudhury 1992, for northern India around theturn of the century; Connell et aI. 1976, for villages innorthern India during the 1960s-70s.

35

Rural-UrbanMigrationand PovertyThe Case of IndiaArjan de Haan

IDS Bulletin Vol 28 No 2 1997

inequality4 Finally, there is no consensus about whyand which women migrate, and whether theirmigration responses differ from men's (Williams1990).

Partly because of these disagreements aboutmigrants' characteristics, it is unclear how migra-tion influences urban and rural poverty If the poorare well-informed and urbanise and do not greatlydrive down the urban wage-rate or employmentprospect by so doing, the process would reducepoverty But there are doubts about whether andhow much migration contributes to poverty reduc-tion. In a recent paper, Ravallion and Datt arguethat in India the process of rural-urban migrationhas not contribued to poverty reduction. Accordingto their analysis, agricultural growth has been themain cause.5

This article looks at the relation between rural-urban migration and poverty: who migrates, fromwhich areas and income groups, how do themigrants compare to non-migrating urban groups,and how do the migrants fare over time? This showsthat there is no simple correlation between the two,that different socioeconomic groups migrate for dif-ferent reasons and that these factors change overtime. The analysis focuses on India. It draws onfield-work and interviews in an industrial area ofCalcutta in the state of West Bengal, mainlyamongst migrants from other states (Bihar, UttarPradesh, Orissa) who in many cases had moved 600km. or more to come and work in the jute andpaper industries (de Haan 1 994a). Migrants hadbeen attracted to the unskilled work in these indus-tries in earlier parts of this century; during the lastdecennia few new job have been created. The

'Lipton 1980. Fielding (this Bulletin) shows that in theUK professionals and managers tend to be highly mobile,and that there is a positive association betweengeographical mobility and upward social mobility

Ravallion and Datt t996. I would argue, however, thattheir analysis underestimates the effect of earnings inurban areas that are remitted to rural areas. If these aresubstantial, the contribution of urban economic growthcontributes more to the observed decline in rural povertythan their analysis suggest. Little information is availableabout remittances. Urban-rural remittances are estimatedto range from 10 to 13 of urban incomes in Africa, andare thought to be of the same order in Asia (Williamson1988: 432). During recent field-work in Bihar, I wasprovided with information about money orders through

36

article also draws on analysis of survey data, partic-ularly the 43rd NSS round of 1987, refering to Indiaas a whole.

The rest of this article is structured as follows.Section 2 describes the rate of urbanisation in India.Section 3 describes the process of circular migrationand the reasons for this form of migration. The fol-lowing sections look at five issues in the relation-ship between migration and poverty Section 4looks at the the role of inequality in migration andworkers' personal motivations to migrate. Section 5discusses the migrants' districts of origin, andSection 6 the background of the migrants, mainly interms of land ownership. Section 7 raises the ques-tion: how does the income of the migrant compareto the non-migrating population, both in their ruralarea of origin and in the urban environment?Section 8 discusses how migrants fare over time: arethey able to improve their income? Section 9concludes.

2 Rate of Urbanisation andMigration in IndiaIn a study on population and food prospects, Dysonexpects the 'long anticipated great shaking loose' ofthe Indian rural peasantry from the villages towardsthe towns to be pronounced during the nextdecades.6 Indeed India is today still largely a ruralsociety In 1992, 74 per cent of the total populationstill lived in rural areas, and 62 per cent of thelabour force worked in agriculture.7 Urbanisationhas progressed, but at a moderate pace. In 1901almost 26 million people, or 11 per cent of theIndian population lived in urban areas. In 1991more than 217 million lived in cities - still only 26

the post office. During August 1996, almost 46,000money orders and 37 million Rupees were sent throughSiwan districri head post office, an average nf about 18Rupees per inhabitant. This is of course only a part of thetotal remittances, since many people will carry theirmoney when going back to the village. Bank transfer isanother channel of remittances, and more commonamong better-off migrants.

'Dyson 1996: 182. The quote is from Ashis Bose.

'UNDP 1995: 176-77. The share of the labour force inagriculture declined from 73 per cent in 1965 to 62 percent in 1990-92. During that period the share in industryremained almost the same (12 per cent and 11 per centrespectively). The increase was entirely in the servicesector: 15 per cent to 27 per cent.

per cent of the total population, On average, duringthis century the Indian population growth rate wasabout 15 per cent per decade, and the growth of theurban population 26 per cent.

Historical explanations of urbanisation in Indiaoften point to the decline of old industrial townslike Murshidabad as a result of Imperial importpolicies, although new urban centres like Bombayand Calcutta emerged. After Independence, urbangrowth remained slow. During the 1960s and1970s, India's urbanisation rates were lower than inother, comparable developing countries.8 Accordingto NSS data, rural-to-urban migration as a propor-tion of total migration declined between 1963-64and 1973-74 (Mukherjee and Banerjee 1978: 31).

Population growth in Calcutta illustrates India'sslow urbanisation. During this century, CalcuttaMetropolitan Corporation, the central part ofCalcutta, has grown at less than 20 per cent perdecade, i.e. only marginally above India's naturalpopulation growth rate and slightly below the rateof urbanisation as a whole. Calcutta UrbanAgglomeration, the greater Calcutta area, has grownat 25 per cent per decade. The growth rate for bothareas has declined since the 1950s. Analysis of thepopulation growth rates of a specific industrial areawithin Calcutta shows that immigration followedthe cycles of the industry, including periods of sub-stantial net outmigration (de Haan 1994a: 140-2).

But slow urbanisation is not the result of lack ofmigration, of rural dwellers being tied to their

8Becker et al. (1992), using a general equilibrium model,show that India underwent exceptionally rapid citygrowth in the early 1960s but a slow rate of urbanisationafter that. According to Williamson (1988: 430), reactingto earlier assertions of over-urbanisation, Third Worldurbanisation has been fairly conventional: '... judged bythe standards of the First Industrial Revolution, the urbantransition associated with ongoing industrial revolutionin the Third World hardly seems exceptional.' ThirdWorld countries' urban population share rose from 17 to28 per cent between 1950 and 1975, which is almostidentical with the experience of currently developedcountries' between 1875 and 1900. English cityimmigration rates between 1776 and 1806 were not thatmuch lower than they were in the Third World after theSecond World War.

Some of the recent literature on indentured migrationalso (implictly) assumes an immobile rural population:

37

villages, as the quote from Dyson seems to imply,and as much of the older (colonial) literature hadargued.9 Historical evidence indicates that most ofthe South Asian rural population has been highlymobile.'0 Studies on the early colonial period showthat large groups of people moved over large dis-tances. In the nineteenth century large numbers ofpeople moved to work in seasonal agriculturalactivities.1' Although migration towards WestBengal is likely to have declined during the latterpart of the twentieth century following the relativeeconomic decline of the state, in 1971 and 1981still more than 2 million people had migrated fromother parts of India (Census figures).

Rates of urbanisation underestimate the number ofmigrants. So does Census migration data, since itregisters only the number of migrants present at aparticular date. Both under-report the total flow ofmigrants to and from the city, This movement hasbeen, and remains, significant, as the following sec-tion describes.

3 Circular MigrationRural-urban migration, in India as in many otherdeveloping countris, has been, and remains 'circu-lar': generally single male migrants go out for work,they maintain close links with their villages of ori-gin, they return when they can or when forced to,and they aspire to return after retirement. In thecase of Calcutta, most migrants came from 'up-country', from eastern Uttar Pradesh and westernBihar, an area about 600 km. to the west of

the literature strongly emphasises the coercive methodsused to make the villagers migrate abroad (de Haan 1995discusses the Indian labour historiography thatemphasises the segmented character of migration undercolonial rule).

'°See, e.g. Habib (1963) for references to migration underMughal rule.

''According to the 1911 Census of Uttar Pradesh, 'therewas no single family in the Banares Division which hadnot at least one member in Bengal, Assam or Bihar'(Census, 1911, Vol 15: 50). Within the province ofBengal, which encompasses the present Indian state WestBengal and Bangladesh, 'ini early 2,000,000 persons of allclasses and races were enumerated in 1921 as comingfrom outside Bengal, including other provinces in Indiaand other countries ...' (Royal Commission of Labour inIndia 1931, Vol 5: 5, pt.1)

Calcutta. They have not settled permanently inCalcutta, but continue to maintain close links withtheir villages, and they continue to speak Hindirather than adopt Bengali.

This pattern has changed surprisingly little duringthis century, contradicting many expectations.Patterns of circular migration are expected to be'dependent on the availability of short-term cash-earning opportunities, either in towns or on planta-tions. This dependence on a particular type ofoutside employment had led Zelinsky to place cir-cular migration patterns at an early stage of mod-ernization. Certainly, evidence from Africa shows anincreasing stabilisation in migration flows. BothCaidwell for West Africa and Gugler in East Africashow a reduced rate of return migration...' (Corinellet al. 1976: 9). The change in Calcuttaís sex ratiodoes indeed show signs of stabilisation,'2 but in myopinion this does not indicate a break in the patternof circular migration. As with rates of urbanisation,the sex ratio only reflects net rates of migration, andunder-estimates total flows. Net migration towardsCalcutta has declined (as a result of decliningemployment opportunities) but this is partly due toincreasing numbers returning to their villages, ormoving to other places within India.

This pattern of circular migration has been stimu-lated by various factorl (de Haan 1994a). First,transport and communication are well developed,and they were so already at the turn of the century"Migrants travel over long distances, but the areaswere relatively well connected. Most of the recruit-ing areas for Calcutta's industries are situated alongthe main west-east and north-south rail connec-tions. This implies that, compared to migration inearlier rural Europe for example, migration in ruralIndia from 1880 onwards was probably less costlyand less time consuming. During this centurytransport facilities have further improved.

In 1901, according to Census data, the sex ratio inTitagarh - the industrial area in which I carried Out myfield research - was 40 (i.e. 40 women per loo men; thefigure for Calcutta City in the same year was 49), in 1981this had become 63 (71 for Calcutta City).

In the literature there is disagreement about the howdistance and migration are related. Connell et al. (1976:16) argue that in India it is commercialisation not'nearness', that encourages migration.

"During the last decades, agricultural production in

38

A second contributing factor is the development ofrural society, as indicated by the quote from Dyson.The rural structure has continued to be dominatedby small land holdings. Rather than being a 'prole-tariat' (in the Marxist sense), many migrant workerspossessed small pieces of land in their villages oforigin. Little research has been done on the interac-tion of urban and rural society, but the two clearlyinteract: income from work outside the village isused to invest in agricultural production - in thecase of the labour migrants from Bihar probablywithout causing important changes in technology -and hence to counteract proletarianisation. Theownership of a plot of land, even if small, con-tributes to return migration.'4

Third, living conditions in urban areas have notbeen conducive to the creation of a stable workingclass. There has been a shortage of housing, popu-lation density has been high, and few amenitieshave been provided. Bad living conditions are notonly a cause of return migration; in my opinion theform of migration has also been partly responsiblefor the conditions in the town. If migrants can save,and most can, they invest their money in the village,not in the town. They have invested in improvingtheir village house - as out-migrating areas testify -and improving or maintaining agriculture on theirown land. Over time, living conditions have proba-bly improved, but at the same time congestion hasincreased, and therefore factors pushing the work-ers back continue to exist.

A fourth issue is whether the pattern of migrationhas been caused by irregularity of work. During theearlier part of this century working hours in large-scale industries changed constantly Large numbersof workers were dismissed during crises'5 and ratesof labour turnover were high. But, as I have arguedelsewhere, in general, the workers'

these areas has become less irregular, due to irrigationand improvements in agriculture. Therefore, the 'pull'from these areas, and the motives to maintain a pattern ofcircular migration may have increased. This point wasmade by Professor PP Ghosh, Asian DevelopmentResearch Institute, Patna.

"The jute mills in 1931 encountered few problems indismissing about one-fifth of their workers. In the wordsof the Labour Commissioner Gilchrist, 'they "got away"with it. Their dismissed employees melted like snow insummer' (Gilchrist 1932: 23-4).

return was not caused by ïrregularity of work.''Industrial work was hardly seasonal, and certainlythe period in which the workers went back to theirvillages was not dependent on production cycles,but on the workers' wish to return during the sum-mer season when marriages and other social occa-sions take place. Since the middle of the century,industrial labour relations and rates of labourturnover have undergone big changes (de Haan1996), but these have not significantly altered thepattern of migration. Gradually, work has becomemore secure, and it has become more difficult todismiss workers. But these changes have notchanged the pattern of migration. In the first place,improved job security applied only to a small por-tion of the labour force. Second, and more impor-tant for the argument here, I did not observe anydifference in patterns of migration between workerswith a permanent job, and those without. Bothmaintained close links with their villages, and a per-manent job may even have increased the possibilityof maintaining this. Their permanent status makesit difficult to dismiss workers, even if they overstaytheir leave. In the interviews, the workers said thatvery often they stayed longer in their village thanthe permitted one month. That people stay in thevillage longer than allowed and cheat with doctors'notes is a public secret.

The fifth and last factor which has contributed tothe pattern of circular migration is the role of thefamily. Migration was mainly by single men. Thereasons for this are complex, and different groupsshow variations in this respect, but I will not dis-cuss this in detail here (see for a longer discussionde Haan 1994b). In my opinion, the pattern of sin-gle-male migration has not been caused by employ-ers' demand for male labour. Since Independencedemand has shifted in favour of men, but the pat-tenis of single-male migration predated this, Myargument is that cultural restrictions on femalemobility outside the household - prevalent innorthern India, less so in the south -' has been the

Colonial reports argue that they left of their ownaccord, and there is little indication that they weredismissed because of seasonality of work. Evidenceregarding the workers' reasons for leaving, and thepercentage of workers that returned (from the Managers'Reports of a jute mill company) show that althoughdismissal was an important reason for workers leaving, alarge percentage left of their own accord (de Haan 1994a:Appendix 5.9).

39

main cause behind thisj7 In any case, men left thewomen behind, and this provides, perhaps, themost important reason to return. The family hasundergone little change during this century; there isno clear trend towards a nuclear family

Thus, industrialisation and urbanisation in Indiahave not broken the close links between rural andurban areas. Over a long period of time circularmigration has continued to be the dominant modeof migration. When ample employment opportuni-ties existed, workers found it easy to leave their joband return after some time. In the case of Calcuttafinding a job has become increasingly difficult, butthis has not broken the pattern of migration.Influenced by cultural norms regarding femalemobility, the predominant pattern of migration hasbeen circular, of single men leaving family and vil-lage to earn an income to maintain the family Thisaspect of the pattern of migration is crucial for thelinks between poverty and migration, as the follow-ing sections show

4 Poverty, Inequality andMigrationEconomic analyses show that inequality variouslydefined, rather than poverty may cause migration.'Our analysis of data from forty Indian villages sug-gest that high emigration from a village is intimatelyassociated with unequal distribution of resources(usually land).. '18 Along similar lines, Stark (1991:140 ff.) argues that relative deprivation plays animportant role in migration decisions. His findingsfrom Mexico show that, for international migration,relatively deprived households are more likely toengage in international migration than are better-offhouseholds. For internal migration the perceivedrisk of relative deprivation in the place of origin (thecity) also plays a role. If this risk is perceived to behigh, migration ceases to be an effective means forachieving gains with respect to relative deprivation.

'From the interviews I concluded that, in the opinion ofthe male migrants', the urban areas provided moredanger to the women's honour than rural areas.

Connell et al. 1976: 10. They emphasise that single-factor analyses of land-based determinants of migrationare inadequate. Migration may also help to relievepoverty

This link between inequality and migration doesnot, however, take into account the migrants per-sonal reasons for migration. Further, for themigrants as for other people affected by it, 'poverty'is not an objective category Although the workers'personal accounts almost invariably pointed topoverty, lack of land andlor income as the main rea-son for migrating, this does not prove a direct linkbetween absolute poverty and migration. Migrantswho in absolute and relative terms were better off,owning more land for example, also mentionedpoverty as their reason for migration. Therefore, notonly poverty, but also inequality, as well as themigrants' (and non-migrants') personal evaluationsof these 'objective factors', are determinant causes ofmigration.

5 Poor Districts and MigrationFrom which areas do the migrants come? Are thesethe most deprived districts? The evidence availableshows quite convincingly that it is not necessarilythe poorest districts from which people migrate.Colonial repórts in the late nineteenth centuryshow that the out-migrating districts were not nec-essarily the poorest. For example, Nolan wrote in1888 that Bihar's inhabitants were more healthy andthat agricultural resources had been developed at anearlier period 'so that the population increaseswithout finding an outlet'.19 In this case, it was theearlier development of these districts, and notpoverty or backwardness, that provided conditionsfor out-migration. As indicated above, the areasfrom which Calcutta attracted most of its migrantswere well connected by railways quite early (and bywater in earlier periods), and it is not very likelythat these were the poorest districts.20

The 'segmentation' of migration streams is impor-tant for understanding the link between poverty

''The soil in these districts is fertile, there is no want ofcapital for any enterprise of real promise, the people areindustrious and frugal; all the conditions of agriculturalprosperity exist except the most essential, that is, themaintenance of a due proportion between the populationand the natural resources of the country ...' (Nolan 1888;see also Chaudhury 1992). Population density is oftenseen as a cause of migration, but the causation might alsobe the reverse: high population density may be madepossible by out-migration.

40

and migration.11 Usually, because of the personalcontacts that are essential for successful migration,people from specific areas migrate to specific desti-nations ('chain migration'). In one jute mill inCalcutta 40 per cent of the workers came from thedistrict Saran in Bihar. Migration from Bihar toBengal around the turn of the century can beexplained with reference to wage differentialsbetween the two areas. But the patterns are morecomplex. Wage differentials cannot explain whypeople from Saran went to Calcutta, why peoplefrom South Bihar to coal mines or tea plantations,and why from some areas very few people migrated.The consequence of this is that, if migration is suc-cessful, it may provide cumulative advantages tocertain areas, excluding others.

6 Who Migrates?Who migrates? For the earlier part of this centurythere is some information on the caste backgroundof migrants in Bengal, since the Census of India col-lected such data (this was discontinued afterIndependence). Amongst unskilled industrialworkers, most castes were represented. In the jutemills a variety of castes was employed, includingBrahmins. Tinker, in his study of indentured migra-tion, concludes: 'The emigration from North Indiarepresented an average sample of the rural popula-tion, excluding the trading, clerical and priestlycastes - and also excluding many of the reallydowntrodden, the sweeper-folk, the lowest of theUntouchables.' Thus, in this case, it was a broadmiddle strata that migrated.

There is scarce information on the landownershipof migrants, and my own field data is not very reli-able (de Haan 1994a: Ch. 6). Most of the migrantsI interviewed said that they had migrated because ofa shortage of land. However, they were diverse in

20According to Connell et al. (1976: 17) cost of migrationmay be prohibitive in the poorest villages. They quoteevidence for migration following the nineteenth-centuryIrish famine: the migrants did not come from the poorestvillages in the West. According to Tony Fielding, the sameis true for migration in Britain in more recent periods.(see Fielding, this Bulletin.)

2 See also Hatton and Williamson (1994: 17) for adiscussion of segmentation of streams of internationalmigrants between 1850 and 1939.

other respects. Many of the migrants did not belongto the poorest section of their village.2 Migrants areboth landowners and people who work on the land.In fact, owners of small plots of land, probably themajority in the western parts of Bihar, work on theirown land, combine this with other income sources,and hire in labour in the peak seasons. It is notuncommon that land is given out to sharecroppers.Migrants stated that when they leave they keep theland because they do not want to depend on others.

Other data confirm this diversity At the end of the1940s a survey found that 59 per cent of the jutemill workers were landless, and 21 per cent ownedless than two-third of an acre (Chattopadhyay1952). The last large survey among jute workersheld around 1970 showed that 42 per cent of thefamilies did not own any land, and 29 per centowned less than 1 acre (Bhattacharya andChatterjee 1973). A crucial question, however, ishow this compares to the non-migrating populationin the areas of origin. I do not have comparableinformation for the group of jute mill workers. Buta survey carried out in Bihar (Oberai et al. 1989) inthe 1980s showed that 39 per cent of the migrants(thus not only jute workers) were landless, and that38 per cent had less than 2.5 acres. The landlessappeared to be more prone to out-migrate, but thedifferences were small: the migrants on average hadonly marginally less land than the total samplepopulation. Thus, as indicated by Connell et al.(1976: 19-2 1), the landless are not the most likelyto migrate; but most of the data shows that all stratado migrate.

Conclusions from two studies show that the rela-tionship between land ownership and migration iscontext specific. First, the study by Oberai quotedabove shows that the migration dynamics were dif-ferent in two other states where surveys were held:in Kerala the middle peasantry migrated more,while in Uttar Pradesh all the landed groups exceptfor the highest size of cultivators had a relativelyhigh propensity to out-migrate (Oberoi et al. 1989:

For example, when Rada Krishna Patro - probably thefirst migrant from Gurundi in South Orissa - left, he didpossess a few acres of land. The family has been able toincrease this land significantly, although the land hasbeen divided within the family Their well-built housesoccupy a prominent place on the village market.

41

34). Second, data on changes in inequality inPalanpur, a village in western Uttar Pradesh, showthat in 1974-75, in a period when the distributionof land was more equal than in other years, 16 outof the 37 villagers with a regular job outside thehousehold came from households in the bottomhalf of the income distribution. In the other surveyyears (1957-58, 1962-63, and 1983-84), thosewho held the well-paying jobs were also those whowere well off in total income terms. 'Where somelower castes had seized the opportunities for out-side jobs in earlier years, in 1983/84 the highercastes were more prominently represented and theoutside jobs became a source of inequality...'?3

7 Migrants' Income andExpenditureMy own field-research does not allow a comparisonof the migrants' income position with the non-migrants in the villages of origin. But the study byOberai et aI. (1989) presents some evidence. Theyconclude that the poor have a relatively higherpropensity to out-migrate from rural areas. In Bihar,Kerala, and Uttar Pradesh the bottom three decilesaccounted for a higher percentage among out-migrants than among the sample population. InBihar, 15 per cent of the out-migrants belonged tothe lowest income class, while 7 per cent of the totalsample population belonged to this income group.However, this data on household income excludedremittances; inclusion may radically change the pic-ture. Of the Bihari migrants, 72 per cent remitted tothe family, but within the lower income groups, thepercentage of remitters was higher: remittancesformed 93 per cent of the income of the Bihari out-migrant households in the lowest income group.Hence, the data is not very convincing, but it doesshow that the poorer groups migrate as well.

How do the migrants compare to other urbangroups? It is often stated that migrants do fairly wellin the urban labour markets, but this is often due totheir higher levels of education or theïr age

' Lanjouw and Stern 1989: 17. Fielding (this Bulletin)notes a difference in the mobility of the unemployed inthe UK in the two periods studied. Unemployed areaveragely mobile, but their mobility was larger in the1980s than in the 1970s.

cohorts24 For example, the comparison of migrantsand non-migrants in Bihar, Kerala, and UttarPradesh, shows that migrants clearly had more edu-cation,2 and Zachariah shows that higher employ-ment rates for migrants can be explained by thedifferences in the age cohorts.26

My own field research dealt with a specific group ina highly segmented labour market (i.e. particulargroups carrying out particular activities). Thismight imply that the migrants would not be able toenter the more lucrative jobs, and that inequalitiesmay be reinforced over time. But this is not

necessarily the case, and much depends on the timeperiod taken into consideration. At the beginning ofthis century, local labourers were predominant inthe industries, but they were, in a period when theindustry expanded rapidly, replaced by migrants.27Over time, the socioeconomic situation of theindustrial workers (in the 'organised sector') hasundergone large changes. In India, and in WestBengal in particular, labour security has increasedsince Independence, and so have wages. However,since the 1950s employment in the older industrieshas declined, and since the 1970s, the industrieshave been in almost continous crises, with regularclosures of whole factories for long periods of time.Yet, in the long run, the position of the industrialworkers has improved relative to workers outsidethe organised industry

'However, Firdausy (1994: 73-77) argues that the risingtrend of urban poverty in some parts of Indonesia wasdue to rural-to-urban migration, and that migrants havelow levels of education, Stark (1991: 371) presents analternative explanation of the high mean (and highvariance) of (international) migrants' income; this wouldbe due to relative lack of information available toemployers about migrants in the receiving country Seealso Fielding (this Bulletin) who shows that migration toBritain South East region has added both to the top andthe bottom of the social ladder.

"Oberai et al. 1989: 27 ff. According to the NationalSample Survey (18th round), in 1963-64 migrants inurban areas had a higher percentage (56 per cent) ofliteracy than the general population (48 per cent). Thismay reflect migration to obtain schooling.

Quoted in Becker et al. 1994: 114. NSS data for1963-64 shows that 45 per Cent of the male migrants inurban areas belonged to the age-group 18-34 years,while the comparable figure for all urban inhabitants wasonly 28 per cent.

42

For a quantitative analysis of differences betweenmigrants and non-migrants we can use the NationalSample Survey (NSS). The published data makes itpossible to compare migrants to the total urbanpopulation, for India as a whole. I have analysedper capita expenditure data, in terms of expendituregroups, of the 43rd NSS round (1987). This com-parison is presented in Graph 1.

In terms of expenditure, migrants were better offthan the total urban population, and hence than thenon-migrant population. The average monthly percapita expenditure of migrants was Rs. 295, while itwas Rs. 243 for the urban population as a whole.29However, we cannot draw many conclusions fromthis. First, we cannot conclude that the migrants'income is higher - although there is other evidenceto that extent - because we would have to controlfor differences in activity and dependency rates, andtake into account the money remitted to the vil-lages. Second, even inferences regarding the percapita expenditure are difficult, because of the pos-sible differences in dependency rates: a comparisonof income on this basis would assume that migrantsand non-migrants have the same dependency bur-den. But the migrants' dependency burden is notlikely to be lower than that of the non-migrants.Therefore, I conclude that the bias is not likely todistort the picture; the evidence does point in thedirection that the migrants' are better-off than thenon-migrants urban population.

77There are contradicting accounts of how and why thishappened. Many have argued that this was due to activemanagement strategies, but in my opinion (de Haan1994a: Chapter 4) local labour left voluntarily, and, in thecontext of an expanding economy at the beginning of thiscentury, they were able to find other means of livelhood.

Sarvekshana Vol XV No 4, Issue 51, April-June 1992:30. In this survey, 27,000 men and 37,000 women out ofan urban sample of 219,000 were enumerated asmigrants. People were classified as migrants if the placeof enumeration was different from their last usual place ofresidence, i.e., the place where she or he stayed for atleast 6 months prior to moving. The comparable data forthe total urban population are from Sarvekshana Vol XVNo 1, Issue 48, July-September 1991. 1 would like tothank Shakin Yaqub for assistance with this analysis.

77Also, unsurprisingly, female migrants are worse off thanthe male migrants, although it is important to takeaccount of the fact that much female migration is relatedto marriage (therefore female mobility is very high).

120

100

QQ

80o-

Total average of migrants

Total urban population

o I I * 4

less 90-110 110-135 135160 160-185 185-215 215-255 255-310 310-385 385-520 520-700 70090 & above

Monthly per Capita Expenditure Groups

20

40.0EDz

Percentage Distribution of Urban Population and Migrants byMonthly per Capita Expenditure

8 How do Migrants Fare OverTime?Finally, how do migrants fare in the city? There is aconsensus that, although migrants may start offwith a somewhat lower income than the nativeurban workers, differences are eliminated quiterapidly, and that they improve their position overtime.3° The life histories I collected show that,although the labour market is clearly segmented,there is significant mobility: sons often started anoccupation other than their fathers' job, workerstook up other jobs when they presented themselves,in factories of in the 'informal sector'. Industrialwork provided a migrant a relatively secure incomeand status. But some chose business because jutemill work would be below their status. And therewere also cases where people chose rickshawpulling: at least in the past, it gave them a betterincome, and it gave them some independence.

Vijverberg and Zeager (1994), using Tanzanian dan. Seealso Stark (1991: 29). Khundker et al. (1994: 13-14) showthat the position of migrants in urban areas of Bangladeshimproves in terms of skill level, employment status, oroccupational mobility: short-term migrants are more often

43

Overall, the mobility of the group of migrants wasdependent on the booms and slumps of the indus-try, but there were significant changes in the relativesocial and economic positions of families.

NSS data show how migrants, on an all-India level,fare over time. This information distinguishesmigrants depending on the period passed sincemigration: people who migrated less than one yearago, between one and five years, and between five andten years. The data show that the average expenditureof migrants who had been in the city for less than oneyear was slightly higher than that of the people whohad been there longer. But the differences are smalland probably not statistically significant: migrantswho had migrated less than one year ago had an aver-age monthly expenditure of Rs. 306; those whomigrated between one and five years ago Rs. 298 (i.e.3 per cent lower), and those who migrated five to tenyears ago Rs. 289 (5.5 per cent lower).

workers (as compared to owTler); 63 per cent of the migrantswho came less than five years ago were unskilled, whereasonly 35 percent of those who had been in the city more than15 years were unskilled; and long-term migrants work moreoften in production rather than sales, constmction or transport.

UptolyrEJ- 1-5yrs

--------- 5-lOyrs

90-110 110-135 135-160 160-185 185-215 215-255 255-310 310-385 385-520 520-700 700& above

Monthly per Capita Expenditure Groups

Graduation: Distribution of Migrants by Period Elapsed since Migratingby Monthly per Capita Expenditure Groups

We can only draw limited conclusion from thisdata. First, to repeat a point made earlier, the datarepresent expenditure, not income. Per capitaexpenditure will be lower when the family sizeincreases (given a fixed number of earners in onefamily). This might explain the decreasing percapita expenditure of migrants of longer duration.Second, remittances may play a role. Migrants whohave been in the city for a longer period, may sendmore money to their places of origin. Oberai et al.(1989: 49) show that Bihari migrants who migratedmore than five years ago, remitted on average morethan twice as much as those who had migrated lessthan a year ago. Third, the process of circularmigration is likely to play a role: migrants come tothe city when wages are high, and leave when lowFinally, people in the higher expenditure groups

44

may be more likely to leave the city after some time,after they have saved some money Hence, the dataon expenditure do not show that migrants improvetheir position over time, but the differences are toosmall to draw certain conclusions. Also, they needto be complemented by data on income; separateevidence on this does suggest that migrants are ableto improve.

9 ConclusionThis article has raised an old set of questions: isrural-urban migration linked to poverty, and how?But raising these old questions is justified, sincethere is no consensus about the answers. And rais-ing the issues is important for analyses and policiesregarding urban (as well as rural) poverty: for

example, what is the socioeconomic position ofmigrants compared to other urban groups, how dopolicies affect migrants, and are there specific poli-cies that can reduce the poverty of migrants withoutnegatively affecting other groups?

Before answering the questions about the linksbetween migration and poverty, two elements of themigration process need to be emphasised. First,whereas migration is a common aspect of most soci-eties (cf. Lucassen and Lucassen, forthcoming), theeffects of migration on the areas of origin and ofdestination may vary And the effects of migrationwithin one area changes over time: in a period withample job opportunities, more people will be ableto take advantage of the opportunities than whenjobs are scarce. This may explain why poormigrants from rural areas in northern India seem tohave profited more at the beginning of the century- when industries expanded rapidly - than during1960-1990, when industrial employment stag-nated.

Second, the character of migration streams isimportant. The most common pattern is single menmigrating, leaving family behind. But this is by nomeans the only pattern: in some cases singlewomen migrate, and in many cases complete fami-lies. There is debate about the reasons for specificpatterns (which has a bearing on the links betweenpoverty and migration, particularly for women).This article has looked mainly at the pattern ofmigration common in northern India. Influencedby cultural norms regarding female mobility, this ismainly by single men, who maintain close linkswith their villages of origin. This has contributed toa relatively slow pace of urbanisation, and has hadimplications - hitherto largely unexplored - fortrends in poverty in urban as well as in rural areas.

The article has discussed five sets of issues relatingto the link between poverty and migration. Many ofthese questions require further research. But thedata available does show that the correlation

45

between migration and poverty is complex. First,both poverty and inequality play a role, and so dothe migrants' own interpretation of whether theyare poor or not. Second, migrants come from a vari-ety of districts, not necessarily the poorest, and his-torical reasons often play a crucial role in this. Someareas have developed a tradition of migration, andonce certain patterns of migration exist, they do notchange easily Third, migrants come from a varietyof backgrounds, and they probably take up differentjobs. They belong to various castes, and to bothlandless and landowner groups. Although there issome evidence that the landless migrate lessbecause they cannot afford the necessary invest-ment, this seems to be context specific: in someareas they migrate less, but this is not necessarilythe case in other areas, or in other periods. Fourth,data on expenditure and income of migrants ascompared to non-migrants confirm the diversity ofmigration experiences. Although the poorest inrural areas may find it difficult to migrate, there isdata that show that in some areas the poorest domigrate. The comparison with the non-migrationpopulation in urban areas shows that migrants areslightly better off. Finally, the scarce data about howmigrants fare over time do indicate that they oftenare able to improve their position; the NSS expen-diture data do not confirm this, but the evidence isnot strong enough to contradict it either.

Thus, there is a need for more information aboutmigrants, who continue to be a very importantgroup in most societies and determine to a largeextent the urbanisation process. Studies that tracechanges over time at both ends of the migrationstreams would be particularly welcome. Migrantsare a difficult group to analyse - and therefore toformulate policies for - because of problems inCensus enumeration for example, or because theirhouseholds is divided over two or more places. Theavailable data do indicate that it is difficult to gen-eralise about migrants, and about the relationshipbetween migration and poverty

ReferencesBanerlee, N., 1989, Working women in colonial

Bengal: modernisation and marginalisation', in: K.K. Sangari and S. Vaid (eds), Recasting Women:Essays in Colonial History, New Delhi: Kali forWomen

Becker, CM. , Williamson, J,G. and Mills, ES., 1992,Indian Urbanization and Economic GrowthSince 1960, Baltimore: Johns Hopkins UP

Bhattacharya, N., and Chatterjee, AK., 1973, 'Somecharacteristics of jute industry workers in greaterCalcutta', Economic and Political Weekly: 297-308, February 19 1973

Breman, J.C., 1985, Of Peasants, Migrants andPaupers: Rural Labour Circulation and CapitalistProduction in West India, Oxford: OxfordUniversity Press

1990, 'Labour migration and rural transforma-tion in colonial Asia', Comparative Asian Studies5, Amsterdam: Free University Press,

Census of India, various years, Delhi

Chapman, M. and Prothero, R. M., 1985, 'Themes oncirculation in the Third World', in R. M. Protheroand M. Chapman (eds), Circulation in ThirdWorld Countries, London: Routledge and KeganPaul

Chattopadhyay, H., 1987, InternaI Migration inIndia: A Case Study of Bengal, Calcutta: K.PBagchi

Chattopadyay, KR, 1952, A Socio-Economic Surveyof Jute Labour, Calcutta: Calcutta University

Chaudhury, P, 1992, 'Labour migration from theUnited Provinces, 1881-1911', Studies in HistoryVol 8 No 1: 13-41

Connell, J., Dasgupta, B., Laishley, R. and Lipton, M.,Migration from Rural Areas: The Evidence fromVillage Studies, Delhi: Oxford University Press

Das Gupta, R., 1981, 'Structure of the labour marketin colonial India', Economic and Political Weekly,Special Number: 1781-1806, November

1976, 'Factory labour in eastern India: sourcesof supply, 1855-1946: Some preliminary findings',Indian Economic and Social History Review Vol13 No 3

Dyson, T., 1996, Population and Food: GlobalTrends and Future Prospects, London: Routledge

Fielding, T., 'Migration and poverty: a longitudinalstudy of the relationship between migration andsocial mobility in England and Wales', this Bulletin

Gilchrist, R.N, 1932, Indian Labour and the Land,

46

Calcutta: West Bengal Government Press

de Haan, A., 1996, 'Labour recruitment in an Indianindustry - historical roots of a labour surplus', in G.Rodgers et al. (eds), The Institutional Approach toLabour and Development, London: Frank Cass,80-101

1994a, Unsettled Settlers. Migrant Workersand Industrial Capitalism in Calcutta, Hilversum:Verloren, (also de Haan 1995)

1994b, 'Towards a single male earner: decline ofchild and female employment in an Indian indus-try', Economic and Social History in theNetherlands Vol 6: 145-67

Habib, I., 1963, The Agrarian System of MughalIndia, 1556-1 707, London: Asia Publishing House

Harriss, J., and Todaro, MP, 1970, 'Migration, unem-ployment and development: a two sector analysis',American Economic Review Vol 60: 126-42

Hatton, T.J., and Willamson, J.G. (eds), 1994,Migration and the International Labor Market1850-1939, London: Routledge

Khundker, N., Mahmud, W, Sen, B. and Ahmed MU.,1994, 'Urban poverty in Bangladesh: trends, deter-minants and policy issues', Asian DevelopmentReview Vol 12 No 1

Lanjouw, P, and Stern, N., 1989, 'Agricultural changesand inequality in Palanpur 1957-1984', LondonSchool of Economics, Development EconomicsResearch Programme, DEP No 24, October

Lipton, Michael, 1980, 'Migration from rural areas ofpoor countries: the impact on rural productivity andincome distribution', World Development Vol 8

Lucassen, J., and Lucassen, L. (eds), forthcoming,Migrations: Historical Concepts and TheoryBuilding, International and Comparative SocialHistory Series, Bern: Peter Lang AG EuropaeischerVerlag der Wissenschaften

Mukherjee, SR., and Banerjee, 5K., 1978, 'Someaspects of internai annual migration in India',Sarvekshana Vol 2 No 2: 23-34, October

National Sample Survey 1971, Eighteenth Round:February 1963-January 1964, No 183, Tables withNotes on Migration, Delhi

Oberai, AS., Prasad, PH. and Sardana, MG., 1989,Determinants and Consequences of InternalMigration in India: Studies in Bihar, Kerala andUttar Pradesh, Delhi: Oxford University Press

Parnreiter, C., 1995, 'Entwurzelung, Globalisierungund Migration. Ausgewählte Fragestellungen',

Journal für Entwicklungpolitik Vol 11 No 3:245-60

Ravallion, M., and Datt, G., 1996, 'How important toIndia's poor is the sectoral composition of economicgrowth?' The World Bank Economic Review Vol10,Noll: 1-25

Royal Commission on Labour in India, 1931, Reportof the Royal Commission on Labour in India,. 1Vol 5 pts.1 & 2, Vol 11 pts.1 & 2, Calcutta:Government of India Central Publication Branch

Sarvekshana Vol 15 Nol, Issue 48, JulySeptember1991, and Vol 15 No.4, Issue 51, AprilJune 1992

van Sehende!, W, and Faraizi, A. H., 1984, RuralLabourers in Bengal, 1880 to 1980, CASP 12,Rotterdam: Erasmus University

Sen, 5., 1992, 'Women workers in the Bengal juteindustry, 1890-1940: migration, motherhood andmilitancy', Ph.D. thesis, University of Cambridge

Singh, M., 1995, Uneven Development inAgriculture and Labour Migration: A Case of

47

Bihar and Punjab, Shimla: The Institute

Stark, 0., 1991, The Migration of Labour, Oxford:Basil Blackwell

Tinker, H., 1974, A New System of Slavery: TheExport of Indian Labour Overseas, 1830-1920,London: Oxford University Press.

UNDP, 1995, Human Development Report 1995,New York: UNDP

Vijverberg, WEM., and Zeager, L.A., 1994,'Comparing earnings profiles in urban areas of anLDC: rural-to-urban migrants vs. native workers',Journal of Development Economics Vol 45:177-99

Williams, L. B., 1990, Development, Demography,and Family Decision-Making. The Status ofWomen in Rural Java, Boulder: Westview Press

Williamson, J.G., 1988, 'Migration and Urbanisation',in H. Chenery and T.N. Srinivasan (eds), Handbookof Development Economics Vol 1, North Holland:Elsevier Publishers: 425-65