Seasonal migration of rural labor in India - Springer Link

Population Research and Policy Review 18: 473–489, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.

473

Seasonal migration of rural labor in India Y. HABERFELD1, R. K. MENARIA2, B. B. SAHOO3 & R. N. VYAS4 1 Tel-Aviv University, Israel; 2 Society for Development Research and Action, Udaipur, India; 3 Indian Statistical Institute, New Delhi, India; 4 M.L. Sukhadia University, Udaipur, India

Abstract. The impact of seasonal migration has been overlooked by students of migration. A unique data set collected in Dungarpur – one of the less developed districts of India – allows us to closely examine both the determinants and impact of seasonal migration. Detailed information was gathered from all members of 624 households, thus enabling analyses at both individual and household levels. The findings indicate that seasonal migration among rural laborers is wide-spread. Rural households in India use migrant labor offered by their members to improve their well-being by both reducing the impacts of inferior conditions and by raising household’s income levels. Migrant labor is a compensating mechanism used by households to reduce their disadvantageous position. Migrant households are characterized by lower education levels, lower levels of income from agriculture, and by an inferior geographical location. However, those households sending migrant labor are found to have higher income levels than those not sending migrant labor. Income from migrant labor accounts for almost 60% of total annual income of households sending at least one migrant laborer. Such findings are in accordance with explanations derived from the ‘new economics of migration’. We can thus learn that migration-related decisions should not evaluated only on the basis of utility maximization of individual migrants, but also on the basis of risk reducing by households. Keywords: Migrant labor, Rural households, India

Introduction In recent years international migration has emerged as a major issue of investigation in the economic and sociological literature (e.g. Borjas 1992). The massive waves of immigrants moving from one country to another have been explained by various types of theories (for a review, see Massey et al. 1993). In the same time, explanations specific to internal migration have been overlooked. Those studies designed to examine processes of internal migration made use of explanations developed originally in order to explain international migration. Gabriel & Schmitz (1995), for example, offered the positive self-selection explanation to explain migration of white men from one American city to another. Stark and his colleagues have offered a ‘new economics of migration’ (e.g. Stark & Bloom 1985). Two issues in their approach that do not appear in the

474

Y. HABERFELD, R. K. MENARIA, B. B. SAHOO & R. N. VYAS

neoclassical explanations are particularly important to the study of seasonal migration in rural India. First, they argue that the relevant unit of analysis in migration studies is the household rather than the individual migrant because migration decision are made collectively by household members and not separately by individuals. Second, they suggest that migration decisions in rural, less developed areas are made not only in order to maximize income, but also to minimize risks. Government and private insurance programs available to residents of developed countries are either absent or very costly to residents of rural areas in the developing countries. As a result, households in the less developed areas minimize risks of various market failures such as drought, flood, or drop in prices by sending some of their members to work away from home as migrant workers. Such diversity of their labor supply guarantees a relatively stable flow of income to the households. Those scholars that examined internal migration processes in India (e.g. Dayal 1959; Prabhakara 1986), were mainly interested in population growth and redistribution. According to a National Commission on Rural Labor report (1991), there were in 1981 about 6 million Indians that left their homes seeking employment somewhere else in India. Most seasonal migrants belong to the lowest classes, namely Scheduled Tribes and Casts, tend to be relatively young, and with low education levels. The main reason for seasonal migration is scarcity of land. It has been established that a small land-to-man ratio is a major factor in rural laborers’ decision to migrate to other areas. Another factor affecting people’s decision to migrate is inter-regional disparities in economic growth caused either by industrial or by agricultural (e.g. the Green Revolution) development. It was suggested that migrant laborers help to raise their household’s standard of living (Rao 1986). However, most students of Indian labor markets have not paid much attention to migration of rural labor. Starting with the pioneering work of Davis (1951), it has been repeatedly argued that the Indian population is relatively immobile due to the prevalence of the cast system, the practice of marriage at an early age, the importance of life within the household, the diversity of Indian languages and cultures, the low education levels, and the predominance of agriculture in the economy. More recently, several factors have been offered in order to explain why massive migration from one market to another is not expected to occur: First, there is risk of open unemployment in moving out of agriculture . . . The second reason is the cost of travel . . . . Third, there is a cost of information. (Acharya 1989: 138) The present study questions this expectation. It is designed first to describe seasonal migration of rural labor in India, and second to analyze both its determinants and economic outcomes.

SEASONAL MIGRATION OF RURAL LABOR IN INDIA

475

Area profile. India is divided into states, and each state is divided into districts. This study was based in the Dungarpur district, situated in the southern hilly part of the state of Rajasthan. It is the smallest district of the state, constituting only 1.1% of the state area and 2% of the state population. The total population of the district is about one million. The density of population was 232 persons/km2 . in 1991, rising from 181 in 1981. The sex ratio in 1981 and 1991 was 1.045 and 0.995 respectively. The decennial population growth between 1971–1981 and 1981–1991 was 28.78% and 28.07% respectively, which is higher than the national average. Dungarpur is one of the less developed districts in Rajasthan. Its rate of urbanization is low, no industry can be found there, and the agriculture in Dungarpur is inefficient as a result of a low mechanization level. There are 846 villages in the district. In 1991, about 10% percent of these villages had populations of 200 persons or less; 27% had populations of between 200 and 499; 30% had between 500 and 999; 21% had between 1,000 and 1,999; 10.5% had 2,000–4,999, and only four villages (0.5%) had populations of between 5,000–10,000 persons. There are three towns in the district, of which one has more than 5,000 people and two have populations ranging from 20,000 to 49,999 (India 1991). The district can be divided into two regions: a southern region, which is somewhat more developed and has a better road system than the northern region. The 1991 literacy rate in the district was 30.5%. While for males it was 46%, for females it was only 15%. These figures are much lower than those at the national level. The 1991 literacy rates in India for the entire population, for males and for females were 52.1%, 63.9%, and 39.4% respectively. In rural areas of the district, literacy rate was somewhat lower and stood at 27%. Among males it was 42% and among females it was 12% (Bose 1991; India 1991). Labor force participation rate in the rural areas was 52% for men and 40% for women as compared to 52% and 22% respectively in the indian labor market. Agriculture provides the main source of livelihood to 70% of the population in the district. The main assets of the households include a small patch of land and heads of livestock. The total number of land holdings in the district was 136,748, out of which the majority (57%) were marginal holdings (less than 1 hectare); 20% were small (between 1–2 hectares), 17% were semi-medium (between 2–4 hectares), 6% were medium (between 4–10 hectares) and barely 0.34% were large (more than 10 hectares). The climate of the area is characterized by a prolonged dry season between September and June, and unpredictable monsoons between July and August. Normal rainfall in the district is quite scanty. Annual rainfall in the district is

476


729 mm. of which 82% falls during the months of July through September. The area irrigated, as a percent of the gross area cropped in the district, was approximately 17% in 1992–1993. This low level of water resources, and the relatively high cost associated with irrigation restrict land cultivation, and necessitate alternative ways for generating income. Two-thirds of the district population belong to scheduled tribes. This proportion is higher in rural areas and stands at 70%, while in the few urban areas of the district, scheduled tribes account for only 16% of the total population (India 1991). The tribal communities are essentially forest dwellers, traditionally living in isolated groups. Their economic conditions have continuously been deteriorating because of depleting forest cover, tribal population growth, pressure by other populations invading their environments, and changes in consumption patterns resulting from interactions with the incoming populations. Under welfare-oriented constitutional provisions, preferential treatment has been provided for these tribal communities since 1950. For this purpose, tribal communities were enlisted in a schedule of the Constitution, resulting in their identification as Scheduled Tribes (ST’s). For example, 7.25% of all government jobs are reserved for ST members (for a detailed description of the social system in India, see Gallanter 1984; Dunn 1993). The households belonging to ST’s live in huts located far away from one another. Decisions within the tribal society are made at the household level (Ghanshyam 1991). The head of the household, in consultation with other members, takes vital decisions like farming operations, marketing of agricultural and livestock produce, division of labor, etc. (Unni 1996). Their asset holdings, mainly land, are extremely low and they combine farming, livestock rearing, collection of forest produce and manual labor in order to survive. ST’s have unique values, cultural norms and social organization. In recent years, tribes in the region have been influenced by a social reformist movement called Bhagatism. Its followers are called Bhagats, which means ‘those worshipping the God’. They follow certain rules including non-alcoholism, non-smoking and worship in groups. Bhagats take extra care of their personal hygiene and cleanliness, a custom which is not evidenced much in other tribes. The Bhagat movement is found only among tribes and not in other social groups. Although the social composition of the district is predominantly tribal, other groups such as upper castes, other backward castes, and scheduled castes, are also found in the district. In the wake of the land reforms initiated in the late 1950s, transfer of land from tribal to non-tribal people has been discontinued. The tribes follow a system of equal distribution of assets between all male inheritors. This continuous sub-division of land assets, coupled with their low asset holdings to


477

begin with, have resulted in the average size of a holding being very small. Yet, among tribe people landlessness is very rare. The Dungarpur district is located on the border of the Gujarat State. The tribal communities located in this area have close links with Gujarat because of spatial proximity and economic factors. Gujarat has relatively well-developed industrial and agricultural sectors. These sectors provide ample employment opportunities for the tribal population as manual laborers. During lean agricultural periods, tribes people from Dungarpur migrate to different places in Gujarat to seek employment (National Commission on Rural Labor 1991). During the frequent drought years in this district, migration is resorted to as a coping mechanism for survival. Thus, the migration from Dungarpur towards border districts of Gujarat is associated with a relatively low cost as a result of several factors, such as familiarity, family relations, proximity and prospects of short-term employment opportunities.

Methods Data. The data used in the present study were collected at the end of 1996. This survey was carried out primarily as part of an evaluation study designed to examine the effects of a large-scale socio-economic project on the lives of rural households. The project was originally designed to raise agricultural productivity through social and economic intervention at the individual and community levels. The villagers participating in the project were encouraged to actively search for income generating avenues and to maintain these new income levels. The project was launched in the Dungarpur district in Rajasthan, India. The district is divided into five blocks. The sampled villages participating in the project were randomly selected from the population of all participating villages in each block. Overall, eight villages were drawn into the sample. Since the primary purpose of the data collection was to evaluate the impact of the project, eight more villages from the population of non-participant villages were added to the sample. These villages were matched with the participating sampled villages in terms of block, geo-physical conditions, size and socio-economic structure. Each village was then divided into three spatial segments based on geographical neighborhoods (‘Phalas’) and 13 households were randomly selected from each such village segment. All these sampling procedures yielded a sample of 624 households. Overall, 540 individuals belonging to 348 of the sampled households migrated from these 16 villages during 1995–1996, mostly for short periods of time. Models. The present study tries to shed some light on three questions. First,

478


how prevalent is seasonal migration and who are the seasonal migrants? In order to answer that, estimates of the rate of migration and a demographic profile of the migrants is provided. In addition, employment characteristics of the migrants in their target labor market are described. These characteristics include destination, duration of employment, type of employment and earnings. The second question to be discussed involves the determinants affecting the decision made by households to send some of their members to work outside their locality. The unit of analysis here is the household. It is hypothesized that the following factors affect such a decision: the demand for labor by the household (D); the supply of labor by the household (S); demographic structure of the household (Z); economic security provided to the household by stable, well-paying jobs held by some of its members (W ); and household income (I ). Thus, the estimated logistic model is: mj = B 0 Dj + C 0 Sj + G0 Zj + O 0 Wj + H 0 Ij ,

(1)

where mj is an indicator of whether the j -th household sent at least one of its members to work as migrant workers in the year preceding the survey, D, S, Z, W and I are vectors of determinants, and B, C, G, O and H their respective vectors of coefficients. The demand for labor by households (D) is measured by the size of cultivated land (in hectares), size of irrigated land, and by the number of units belonging to three groups of livestock owned by the household – cattle, sheep and goats, and poultry. The supply of labor (S) is captured by the number of household members of prime-working age (15–60 years old), number of household members of other ages, number of members aged 10–25 years who attend school, and by the number of men of prime-working age. Household demographic characteristics (Z) include education level within the household as measured by the number of members with post-primary school education, social group as measured by dummy variables indicating ST and SC, whether it is a Bhagat household and a dummy variable indicating the less developed geographical region. In addition, participation in the socio-economic project is introduced as a control. W contains indicators of whether at least one household member is employed in the relatively well-paying, secured government sector, and whether one member works in a high paying job (of more than 1500 Rs per month). Finally, household income (I ) is acquired by income from salaries, earnings from other sources, and monetary value of agricultural crops and livestock outputs. In addition, a separate analysis is conducted only for those households providing migrant labor. An OLS model is estimated as follows: m∗j = B 0 Dj + C 0 Sj + G0 Zj + O 0 Wj + H 0 Ij + qyj ,

(2)


479

where m∗j indicates the number of months of migrant labor provided by the j -th household, y denotes the daily wage rate paid to the migrant laborers belonging to that household, and q is its coefficient. All other vectors are similar to those of Equation (1). The third question addressed in this study is the effect of migrant labor on household income. For that purpose, the following OLS model is estimated for all households: Ij = A0 Wj + K 0 Lj + M 0 Sj + N 0 Zj + pm∗j ,

(3)

where Ij is the estimated total (natural logarithm of) income of the j -th household; W is a vector of earnings stability variables; L is a vector of agricultural variables generating income; S is a vector of labor supply indicators; and Z contains demographic attributes of the household. A, K, M and N are vectors of coefficients. m∗ denotes months of migrant labor and p is its coefficient. The dependent variable is composed of household wages and salaries, monetary value of the agricultural production, and monetary value of the livestock’s output (it is assumed that households can sell them). W contains indicators of whether at least one household member is a government employee, and whether one member works in a high-paying job. The group of agricultural variables include size of cultivated land, size of irrigated land, area under high yielding variety of seeds (which indicates higher level of agricultural technology), and the number of units belonging to the three different types of livestock. Labor supply (S) is measured, again, by the number of household members at prime-working age, number of household members of other ages, number of members aged 10–25 who attend school, and by the number of men of prime-working age. The number of household members with post-primary education is broken down in this model into two components: number of men with primary school education or higher, and the number of women with primary school education or higher. Household demographic characteristics include social group, Bhagat status, region, and participation in the project. Finally, migrant labor is measured by the total number of months of migrant labor provided by the household.

Results The immigrants. Most workers migrating from the district of Dungarpur are seasonal migrants. They leave their villages usually during October– November after harvesting rain crops and, in a case of irrigated land, sowing winter crops, and return home before the next summer season. In between,

480


immigrants visit their homes intermittently for family and local obligations. The average duration of an immigrant away from home is approximately 5.6 months. The variation on this average is relatively small and stands at 3.1 months (see Table 1 for descriptive statistics). The large majority of migrant workers (88%) head for neighboring, more affluent districts. About 91% of the seasonal migrants are men. Their average age is 26.4 years (with a standard deviation of approximately 10), and their education levels are higher than those found among other Dungarpur villagers. On average, migrant laborers have more than 3 years of schooling. Only 32 percent of the sampled immigrants were illiterate. About 30 percent of them were literate, yet did not complete the primary education level, which in India stands at 5 years of schooling. The largest group among the seasonal migrants (38%) is that of people with post-primary school education. This profile indicates that the migrant workers are probably positively self-selected. Non-immigrants of the age of 15 years or above, are both older and less educated than the immigrants. Their average age is 35 years (SD = 15.9) and they completed, on average, 2.5 years of schooling (SD = 4.0). Even if we restrict the age of non-immigrants to include only those between the ages of 15–55 years, still their average years of schooling is lower than that of the immigrants (2.8 vs 3.2). 89% of the immigrants belong to ST’s. Only 9% belong to the nonscheduled social system. These proportions are representative of the population from which they come. Similarly, 24% of the immigrants are considered to be Bhagats, as compared with 29% Bhagats found in the entire sample. Migrant workers from Dungarpur are concentrated in a small number of jobs. The largest group among them (27.5%) is found in manual, unskilled jobs. The other large groups are employed in construction (26%), hotels and tea shops (10%), factories (9.5%) and agriculture (9%). The migrants daily wage rate (about 44 Rs) does not vary much (SD = 25) and is much higher than the daily earnings in their homes. In the sampled villages, daily wages averaged that year 25–30 Rs. Migrant vs non-migrant households: Description. More than one half of the sampled households provided migrant labor. The average household from which migrant labor is provided is different on every dimension, from the average household not providing migrant labor (see Table 2). About 60% percent of migrant households’ annual income is accumulated through wages for migrant labor, while the main source of income for non-migrant households are local salaries paid to its members. The second most important source of income for all sampled families is agriculture. However, while for migrant households, income from agriculture constitutes less than 20% of its total in-


481

Table 1. Season immigrant workers from Dungarpur: Descriptive statistics Variable Percent men Age Education Years of schooling Percent illiterate Percent with 1–4 years schooling Percent with 5 or more years

Mean 91 26.4 3.2 32 30 38

SD

9.9 3.5

Social groups ST SC Non-scheduled backward classes Percent Bhagats

89 1.5 9 24

Migrants’ destination Within district Outside district Outside state Outside country Duration of migration (months)

9.0 2.5 88.0 0.5 5.6

3.1

Type of employment Unskilled labor Construction Hotels and teashops Factory workers Agriculture Services Domestic help Skilled labor Other Daily earnings (Rs)

27.5 26 10 9.5 9 6 5.5 3 3.5 43.5

24.8

N

540

482


come, it is more than one-third of non-migrant households’ total income. This is in part because non-migrant households have, on average, more assets than their migrant counterparts. Overall, however, migrant households’ income is significantly higher than that of non-migrant households. Comparing the two types of households on the labor supply dimension yields mixed results. On the one hand, quantity of labor available to migrant households is larger than that available to non-migrant households, by approximately one-half of a person at prime working age. On the other hand, average education of non-migrant households’ labor is higher than that of migrant households. The differing labor quality between the two types of households is further reflected in their employment opportunities. Members of non-migrant households hold better jobs than members of migrant households. Finally, the two types of households differ also in their social characteristics. Migrant households tend to belong more to ST’s, to live in the less developed region of the district, and to stay out of the Bhagat community. Migrant vs non-migrant households: Determinants. The results of a logistic model (Equation (1)) designed to estimate the determinants of households’ decision to make at least one of its members a seasonal migrant, are described in Table 3. These results reveal that such a decision is affected by several factors. First, larger labor supply by the household is associated with being a migrant household. Second, higher levels of education within the household tend to lower the probability of a household to provide migrant labor. Third, living in the less developed region raises this probability. Fourth, being a Bhagat household lowers this probability. Finally, higher income from agriculture and a more livestock-intensive household, as measured by poultry, tends to also lower the probability of a household to provide migrant labor. Determinants of migrant labor supply by households. The amount of migrant labor supply (in months) is analyzed for migrant households only – as described in Equation (2). The results of this analysis are presented in Table 4. The daily wage rate offered to migrant workers coming from Dungarpur has a major impact on the amount of labor supplied by them. On average, an increase of one rupee in their wage offer is associated with an increase of more than a day in migrant labor (b = 0.044). In addition, living in the less developed region is associated with a sharp increase of migrant labor supply. However, the most important determinants of the amount of migrant labor supplied by the households studied are their labor resources. All five variables measuring quantity and skill level of labor resources affect significantly the dependent variable. Quantity of household labor tends to raise the amount of

483


Table 2. Migrant and non-migrant households: Descriptive statistics Variable

Months of migrant labor Income (annual) Total In (total) Salaries (100 Rs) Other earnings (100 Rs) Income from agriculture (100 Rs) Income from livestock offspring (100 Rs) Income from migrant labor (100 Rs) Agricultural assets Cultivated holdings (hectares) Irrigated holdings (hectares) Cattle Sheep and goats Poultry Percent cultivated area with high yielding variety of seeds Labor supply Number of people of prime working age Number of people of other ages Number of men of prime working age Number of people aged 10–25 attending school Number of people with post-primary school education Number of men with post-primary school education Employment Percent in which at least one person has a secure job Percent in which at least one person has a high-paying job Social characteristics Percent scheduled tribes and castes Percent Bhagats Percent in the less developed region Percent participating in project N

Migrant households Mean SD

Non-migrant households Mean SD

8.71

7.62

–

20,483 9.654 20.64 12.10 38.19 13.22 120.68

18,770 0.727 92.96 19.39 44.84 21.88 134.75

17,328 9.140 86.60 11.12 64.05 11.50 –

22,942 1.150 196.82 20.28 104.17 12.52

0.68 0.34 4.66 4.06 1.66

0.87 0.65 2.60 5.51 3.80

0.92 0.55 4.88 3.53 2.12

1.27 1.03 2.80 5.33 4.83

4.06

11.94

6.78

15.57

4.16 2.32 2.19 0.5 0.81

1.88 1.67 1.16 0.78 1.02

3.64 2.32 1.83 0.80 1.38

1.74 1.72 1.02 0.98 1.63

0.67

0.81

1.05

1.19

0.09

0.27

0.04

0.19

0.90 0.25 0.79 0.54

0.85 0.34 0.70 0.45 350

274

All differences in means other than ‘number of people at other ages’ are significant at p < 0.05 level.

484


Table 3. Determinants of a migrant household: Results of a logistic analysis Variable

Coefficient

SE

Agricultural assets Cultivated holdings Irrigated holdings Cattle Sheep and goats Poultry

0.232 −0.258 0.019 0.005 −0.059∗∗

0.298 0.356 0.41 0.19 0.24

Labor supply Number of people of prime working age Number of people of other ages Number of people aged 10–25 attending school Number of men of prime working age Number of people with post-primary school education

0.199∗ 0.152∗∗ −0.177 0.463∗∗ −0.426∗∗∗

0.118 0.060 0.130 0.189 0.108

Social characteristics Social group Bhagat Region Project

0.297 −0.661∗∗∗ 0.614∗∗∗ 0.291

0.311 0.213 0.229 0.194

Employment A person with a secure job A person with a high-paying job

−0.647 −0.950

0.401 0.645

Income (annual) Salaries (100 Rs) Other earnings (100 Rs) Income from agriculture (100 Rs) Constant Pearson χ 2 d.f.

−0.0003 −0.007 −0.006∗∗ −1.378 642.7a 604

0.033 0.005 0.003

∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01. a The Pearson χ 2 statistic is not distributed as chi-square because it is derived

from ungrouped data. As a result, a test of goodness-of-fit cannot be performed.

migrant labor supplied by the household, and the skill level of household’s potential labor force as measured by the education of household members, is negatively related to migrant labor. The number of males of prime working age contributes a lot to migrant labor provided by the household. Each such member is associated with a rise of almost three months of migrant labor.

485


Table 4. Determinants of migrant labor supply (in months): Results of an OLS regression analysis Variable

Coefficient

SE

Agricultural assets Cultivated holdings Irrigated holdings Cattle Sheep and goats Poultry

0.50 −0.40 0.21 0.04 −0.34∗∗∗

1.16 1.55 0.16 0.07 0.10

Labor supply Number of people of prime working age Number of people of other ages Number of people aged 10–25 attending school Number of men of prime working age Number of people with post-primary school education

0.80∗ 0.61∗∗∗ −1.32∗∗ 2.72∗∗∗ −0.79∗∗∗

0.42 0.21 0.51 0.66 0.41


0.74 −0.72 2.23∗∗ −1.13

1.24 0.82 0.93 0.41

Employment A person with a secure job A person with a high-paying job

0.43 −2.73

1.83 3.07

Income (annual) Salaries (100 Rs) Other earnings (100 Rs) Income from agriculture (100 Rs) Daily migrant wage Constant N R2 (adjusted)

−0.0006 0.001 −0.014 0.044∗∗∗ −4.36 350 0.33

0.007 0.019 0.010 0.015

∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01.

486


Again, number of poultry units affects migrant labor supply negatively. Trying to understand this effect, we estimated two more equations. The second equation did not contain the poultry variable. The results obtained by this equation were very similar to those described here. A third equation in which all (standardized) livestock indicators were combined into one index, and all (standardized) land assets were combined into a second index was then estimated. Both indices were not found to be significantly different from zero. Thus, we assume that the poultry variable captures some latent, possibly livestock-intensive production processes which are not captured by the other livestock variables. Other agricultural assets, as well as income and employment variables, do not affect the amount of migrant labor supplied by the household. The effect of migrant earnings on household income. The final analysis is designed to examine the importance of various determinants of the rural households’ income. As expected, agricultural assets owned by the household affects income significantly. Every hectare of cultivated land adds, on average, 16% to the household income. Livestock owned by the household contributes as well to the family well-being. However, the most important factors affecting tribal household’s income are stable and high-paying jobs. Since government jobs are characterized as both stable and high-paying, holding such jobs helps the most in raising income levels of the villagers. Villagers living in the Northern, less-developed region of Dungarpur, suffer a 10% loss in their income as compared to their southern counterparts. Finally, and despite a temporary decline in labor supply for local production as a result of the migration process, migrant labor provided by tribal households contributes significantly to annual income. Every month of migrant labor is associated, on average, with a 7% increase in annual income. Put differently, two months of migrant labor are almost equivalent to the contribution of one hectare of cultivated land (which is the average land holding of a migrant household) to the family’s annual income.

Discussion The impact of seasonal migration on the social and economic well-being of residents of the less developed regions is enormous. The first important finding is that seasonal migration among rural laborers is wide-spread. A majority of the households studied sent one or more of their members to work away from home for several months a year. Those households sending migrant labor were found to have higher income levels than those not sending migrant labor. Furthermore, within the group of migrant-labor households, income

487


Table 5. Determinants of total household (ln) income: Results of an OLS regression analysis Variable Agricultural assets Cultivated holdings Irrigated holdings Cattle Sheep and goats Poultry Percent high yielding variety seeds

Coefficient 0.158∗∗∗ 0.34 0.062∗∗∗ 0.016∗∗∗ 0.009 0.002

SE

0.061 0.077 0.010 0.005 0.006 0.002

Labor supply Number of people of prime working age Number of people of other ages Number of people aged 10–25 attending school Number of men of prime working age Number of people with post-primary school education Number of women with post-primary school education

0.013 0.015 0.004 −0.040 0.050 0.052

0.030 0.015 0.034 0.050 0.033 0.052


−0.106 0.004 −0.104∗ −0.004

0.081 0.055 0.060 0.050

Employment A person with a secure job A person with a high-paying job Months of migrant labor Constant N R2 (adjusted)

0.669∗∗∗ 0.980∗∗∗ 0.073∗∗∗ 8.442 624 0.63

0.103 0.121 0.004

∗ p < 0.10; ∗∗∗ p < 0.01.

from migrant labor accounts for almost 60% of their total annual income. Put differently, the above-average income level of migrant-labor households is attributed primarily to the earnings brought in by the migrant members of the household. Migrant labor is a compensating mechanism used by disadvantageous households. These households are characterized by lower education levels, lower levels of income from agriculture, and by an inferior geographical

488


location. Such findings are in accordance with explanations derived from the ‘new economics of migration’. We can thus learn that migration-related decisions should not be evaluated only on the basis of utility maximization of individual migrants, but also on the basis of risk reducing by households. In sum, rural households in India use migrant labor offered by their members to improve their well-being by both raising household’s income levels, and reducing impacts of inferior conditions. An interesting question left unanswered is whether to develop social policies aimed at reducing seasonal migration by offering rural households alternative income sources – as proposed in some of the less developed regions. We found migration to be a major economic force in the region studied. It serves, together with agriculture, as the primary income source of the region. Its importance becomes greater as socio-economic conditions deteriorate. Unless alternative policies benefit the poor, less developed, less socially connected households, seasonal migration should be left as a free market mechanism helping those with a lower starting point to improve their well-being. Our conclusions at this stage are limited to the studied district. The next necessary step should be to conduct studies similar to the one presented here in other backward regions in India and in other countries as well. If the findings of these studies will turn out to be the same as ours, then it would be possible to generalize our conclusions beyond the borders of the district of Dungarpur.

Acknowledgments The data collection was financed by the Swedish International Development Agency, New Delhi for evaluating the impact on livelihood and awareness of the PAHAL project.

References Acharya, S. (1989). Agricultural wages in India: A disaggregated analysis, Indian Journal of Agricultural Economics 44: 121–139. Borjas, G. (1992). Immigration research in the 1980s: A turbulent decade, in: D. Lewin, O. S. Mitchell & P. D. Sherer (eds.), Research frontiers in industrial relations and human resources (pp. 417–446). Madison, WI: IRRA. Bose, A. (1991). Population of India: 1991 Census Results and Methodology. Delhi: B.R. Publishing Corporation. Davis, K. (1951). The population of India and Pakistan, Princeton NJ: Princeton University Press.


489

Dayal, P. (1959). Population growth and rural urban migration in India, National Geographical Journal of India 5. Dunn, D. (1993). Gender inequality in education and employment in the scheduled castes and tribes of India, Population Research and Policy Review 12: 53–70. Gabriel, P. E. & Schmitz, S. (1995). Favorable self-selection and the internal migration of young white males in the United States, The Journal of Human Resources 30: 460–471. Gallanter, M. (1984). Competing inequalities: law and the backward classes in India. Berkeley, CA: University of California Press. Ghanshyam, S. (1991). Study group on migrant labour. In: National Commission on Rural Labour, Report 2. India (census) (1991). District Census Handbook – Dungarpur, Government of India. Massey, D. S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A. & Taylor, E. J. (1993). Theories of international migration: A review and appraisal, Population and Development Review 19: 431–466. National Commission on Rural Labour (1991). Report. Prabhakara, N. R. (1986). Internal migration and population redistribution in India. New Delhi: Concept Publishing Company. Rao, M. S. A. (1986). Some aspects of sociology of migration in India, in: M. S. A. Rao (ed.), Studies in migration. Delhi: Manohar. Stark, O. & Bloom, D. E. (1985). The new economics of labor migration, American Economic Review 75: 173–178. Unni, J. (1996). Diversification of economic activities and non-agricultural employment in rural Gujarat, Economic and Political Weekly, 17 August.

Address for correspondence: Yitchak Haberfeld, Department of Labor Studies, Social Sciences, Tel-Aviv University, Tel-Aviv 69978 Israel Fax: 972-3-6407300; E-mail: [email protected]