FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL: A VILLAGE LEVEL ANALYSIS

The Indian Journal of Labour Economics, Vol. 48, No. 3, 2005 FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL: A VILLAGE LEVEL ANALYSIS Sucharita Sinha* This paper examines rural female work participation rates (FWPRs) in four districts of West Bengal. Rather than studying the diverse levels of FWPRs within a broad geographical area, factors associated with differing levels of FWPRs in different regions may be better analysed by a detailed analysis of smaller and socio-economically more homogeneous units. Such a targeted approach as followed in this paper is best suited to understand the level of FWPRs in a region. I. BACKGROUND AND INTRODUCTION While participation in the labour market is not by itself a sufficient condition to ensure an increase in the bargaining power of women and a substantial decision-making role for women within families, ample evidence in the literature (see, among others, Mencher and Saradamoni, 1982; Bardhan, 1985; Nagaraj, 1989; and Bennet, 1992) shows that the ability to earn an income by active participation in the labour market is a vital prerequisite for the economic independence of women. The analysis in this paper is based on data from the Census of India. Methodological biases such as changes in definitions of work and workers in Censuses of different years, problems related to data collection, the respondent enumerator biases arising from the lack of trained enumerators and exclusion of a gamut of activities women are commonly engaged in, from the definition of work 1, are common criticisms of secondary data sources such as the Census of India (see Anker, 1983; Banerjee, 1989; Bardhan, 1977; Deere and de Leal, 1982; Krishnaraj, 1990; and Visaria, 1997). However, the Census of India is the single largest source of a large volume of data in the country and comparative analysis of such data across regions could yield interesting results, given the absence of regional concentration of the above-mentioned lacunae. Further, despite the inadequacy of figures, trends indicated by secondary data sources have often been corroborated by village level micro studies. Finally, while the Census of India, does under enumerate women workers on account of its exclusion of the vast bulk of women engaged in productive activities within the household, the analysis in this paper attempts to understand the level of visible female work participation as it is this kind of participation which is accompanied by monetary compensation important for the social emancipation of women. The Census definition of work provides a fairly good estimate of women engaged in paid work. A study of Table 1 showing the rural FWPRs between 1961 and 2001, across 15 most populous states of India reveals FWPRs to be persistently and widely different across states. While a wide gamut of demographic (such as the average age of the female population, literacy of the overall and the female population, marital status of females and the presence of femaleheaded households), social (such as the presence of a large or small Scheduled Caste (SC) and Scheduled Tribe (ST) population and the religious composition of the population), household (such as the level of household income and male earnings, landholding and the extent of (often * Research Scholar, Department of Economics, University of California, Riverside. The author is extremely grateful to Madhura Swaminathan and V.K. Ramachandran for guiding her through writing this paper.

564 THE INDIAN JOURNAL OF LABOUR ECONOMICS unpaid) female participation in economic activities within the household), regional (such as the availability of employment opportunities, the level of regional prosperity, cropping patterns in the region) factors have been posited in the literature to be associated with and explain the pattern of rural FWPRs across Indian states, none of these factors are mutually exclusive and can independently explain the level of rural FWPRs across Indian states (see Jose, 1989). Female work participation in the labour market or the lack of it cannot be attributed to any specific factor across the country but is instead the outcome of a complex.interaction of the general context with the specific factors affecting women (Nagaraj, 1989). 2 While it might a priori seem that a multivariate regression kind of analysis would be best suited to analyse rural FWPRs across Indian states, previous such work (see Sundaram, 1989; Dholakia and Dholakia, 1978; Gulati, 1975; and Nath, 1970) has indicated the absence of a single set of significant factors associated with the pattern of rural FWPRs across Indian states. The analysis of rural FWPRs within a broad socio-economic and cultural canvass as the whole country thus appears a difficult task and a smaller unit of analysis might be well suited to the task at hand. Motivated with the idea of specificity of factors associated with rural FWPRs in a region, this study is an attempt to analyse rural FWPRs in four districts of rural West Bengal, namely Haora, Puruliya, Murshidabad and Maldah from the 1991 Primary Census Abstract and Village Directories. 3 While no specific sampling techniques guide the choice of these districts, (which are not representative for the state), the following discussion analyses the different levels of rural FWPRs within these districts and thus highlight their essential differences, to become the key to understanding the different pattern of rural FWPRs in each of them. A comparative analysis of the four areas of the state highlights the importance of a targeted approach to understand and analyse rural FWPRs across the country. Table 1 Rural Female Work Participation Rates in 15 States and India, 1961-2001 State 1961 1971 1981 1991 2001 India 31.39 13.36 23.18 26.67 31 Andhra Pradesh 46.00 27.37 40.03 42.48 43 Assam 32.41 5.58 N/a 21.61 22 Bihar 28.49 9.31 14.65 16.26 20 Gujarat 34.15 12.07 26.85 35.60 39 Haryana N/a 2.29 12.29 12.62 34 Karnataka 36.79 15.77 30.66 36.60 40 Kerala 20.88 14.08 17.72 16.86 16 Madhya Pradesh 48.60 20.75 35.78 39.26 41 Maharashtra 46.74 24.39 40.85 46.05 47 Orissa 27.38 6.83 21.09 22.62 27 Punjab 16.50 0.72 6.90 4.37 23 Rajasthan 40.82 9.27 24.99 33.25 41 Tamil Nadu 37.11 17.63 33.55 38.50 41 Uttar Pradesh 19.90 7.27 9.04 14.16 19 West Bengal 10.62 4.58 8.89 13.07 21 Note: The 1961 Census had Haryana as part of Punjab; The 1981 Census was not held in Assam; The 1991 Census was not held in Jammu and Kashmir; N/A indicates data is not available; Further, in this table, we have considered Indian states with a population more than 10 million, only. We have not considered Assam and Haryana in certain cases because of the lack of data and have excluded the north eastern states as they show a pattern of FWPRs different from the rest of the country. Source: Calculated from the Census of India, Various Volumes.

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 565 West Bengal is a densely populated state, characterised by rural poverty. The rural economy of the state is mainly agrarian with a concentration on jute and paddy production. Industrial growth in the state is relatively stagnant in contrast to the massive growth of agricultural productivity (see Lieten, 1996). There are two main reasons motivating a study of rural FWPRs within the state of West Bengal. First, rural FWPRs have persistently been relatively low and consistently below the Indian average in rural West Bengal in comparison to other Indian states (see Table 1). In contrast to states like Andhra Pradesh, Madhya Pradesh and Maharashtra where rural FWPRs are above or in the vicinity of 40 per cent, the FWPRs in West Bengal have been 11 per cent on an average between 1961 and 2001. 4 Other states with comparable rural FWPRs are Bihar, Uttar Pradesh, and Kerala. This low level of rural FWPRs has persisted despite the implementation of land reform measures and the establishment of panchayats, and despite the absence of any major social opposition to women s work in the countryside (Lieten, 1996). Secondly, this low level of rural FWPRs contrasts with the dynamics of MWPRs, which have risen between 1961 (53.47 per cent) and 2001 (54 per cent) and are infact above the average Indian MWPR (52 per cent). This raises a suspicion that dearth of job opportunities cannot explain low rural FWPRs in the state and the role of traditional regional biases would have to taken into account in order to understand FWPRs in the state. An observation of Table 2 shows the dispersion in rural FWPRs even within rural West Bengal. While districts like Puruliya and Darjiling have had FWPRs consistently considerably above the Indian average, extremely low FWPRs in districts like Haora and Nadia have resulted in low average rural FWPRs in the state. Thus, there exists considerable variability in the rural FWPRs even within a smaller geographical area within a state where cultural and climatic differences are less pronounced. II. THE DATA We have used data from the State Primary Census Abstract (PCA), 1991 of West Bengal and the Village Directories (VDIR), 1991 of each district for this paper. While the PCA gives Table 2 Female Work Participation Rates, West Bengal, Rural, 1961-2001 Districts 1961 1971 1981 1991 2001 Barddhaman 9.77 5.63 8.47 11.22 19.69 Puruliya 40.46 9.83 26.89 37.44 39.13 Bankura 19.50 8.19 15.94 20.22 33.45 Medinipur 11.81 5.15 10.53 18.53 24.29 Hugli 7.99 5.15 8.44 9.32 18.69 Haora 2.27 0.89 1.89 3.90 10.01 24 Parganas 2.80 1.26 3.21 6.06 11.95 Nadia 3.18 1.57 3.09 4.24 13.02 Murshidabad 6.10 2.42 6.14 9.61 14.43 Maldah 11.71 3.38 9.66 17.56 29.50 West Dinajpur 6.70 2.75 8.87 14.02 25.90 Darjiling 35.70 25.88 26.37 24.82 25.32 Jalpaiguri 20.65 10.66 14.74 17.24 26.05 Koch Bihar 3.74 1.49 4.12 9.26 23.13 Birbhum 9.64 4.38 9.02 13.35 20.16 Note: The figures for 24 Parganas in 1991 and West Dinajpur in 2001 have been obtained by adding North and South 24 Parganas and Uttar and Dakshin Dinajpur respectively. Source: Calculated from the Census of India, Various Volumes.

566 THE INDIAN JOURNAL OF LABOUR ECONOMICS information on population characteristics of the villages such as the number of workers, their occupational classification, the number of SC/STs and the number of literates, the VDIRs give information on the infrastructural variables in villages. Besides the qualitative aspects of the data set, there were problems of matching the data on villages obtained from the PCA and VDIRs in order to construct the full data set on each village. Absence of villages, which figured in one data set, from the other resulted in the selection of only the common villages, which were there in the PCA as well as the VDIR. Secondly, while the village location code in the PCA was an eighteen digit code consisting of the two digit district code, the four digit police station code, two four digit community development (CD) block codes and the four digit village number, the code in the VDIR consisted of the district, the police station and the village number only. There arose a problem when some villages belonging to different CD blocks had the same village number and were hence coded identically in the VDIR. Such villages could not be traced to an appropriate CD block and we hence deleted them from the data set. The third problem with the data was the occurrence of some undefined dummy variables for some of the villages in some districts. This was the case when there was a zero or a blank instead of the specified one or two in order to indicate the availability or the non-availability of a variable. This problem was particularly prominent in the case of Puruliya among the districts that we have chosen. In this case the presence of zero or a blank cell was considered to be an indicator for the non-availability of the particular variable. There was also a problem related to zero or negligible population in some villages. 5 Since even a small number of workers in such villages (in case of negligible population) would mean a large per cent of the population, including such villages would bias the WPR estimates to a considerable extent. Such villages had to be removed and we retained only such villages in the analysis, which had a population of more than 20 persons. Among the districts that we have considered, Puruliya, Haora, Maldah and Murshidabad had 2686, 745, 1799 and 2220 villages respectively in the Primary Census Abstract and 2684, 741, 1801 and 2220 villages in the Village Directory. The final data set for Puruliya consisted of 2424 villages (90 per cent), 733 villages (98 per cent) for Haora, 1633 (91 per cent) for Maldah and 1910 (86 per cent) for Murshidabad. III. SOCIO-ECONOMIC CHARACTERISTICS OF THE FOUR DISTRICTS The focus of our analysis is a deeper regional investigation, in contrast to generalised analysis, in order to understand the association between the nature of the socio-economic conditions specific to any given region and why women seem to participate more in rural labour markets under some conditions vis-à-vis others. In this section we attempt a description of the socio-economic characteristics of the four districts of study. The socio-economic conditions vary greatly among the districts and a complex mesh of factors determine the socio-economic fabric even within a small geographical expanse of a single district and are vitally connected to the resultant FWPRs. 1. District Background Puruliya is a district is situated in the dry areas in the western part of the state with climatic characteristics similar to the Chota Nagpur plateau area of Bihar. The district is one of the poorest in the state and despite the land redistribution after the land reforms initiated in 1977, there is still a large population of landless agricultural labourers. In contrast, Murshidabad, situated in the midlands of West Bengal, synonymous with the silk industry of Bengal, and is

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 567 a relatively prosperous district. Besides sericulture, the region also specialises in the production of jute and mangoes. Situated in the south central part of West Bengal, Maldah, though famous for its commercial crop of mangoes, lacks the use of sophisticated agricultural techniques used in more advanced regions and plantation methods often involve single cropping in contrast to agriculturally prosperous districts like Barddhaman and Medinipur. Besides mango, agriculture in Maldah is mainly dependent on paddy cultivation. Lastly, Haora is the most urbanised of all districts being studied and both; the proximity to Kolkata (the capital of West Bengal) and the predominance of factories in this region perhaps contributes to the relatively low degree of agricultural activities in the district. Haora has traditionally been famous as an industrial centre and is also known for jute mills (see Ghosh, 1989). Though no specific concerns motivate the choice of the four districts, they are each distinct with respect to the geographical and socio-economic conditions and are hence expected to provide us with a wide spectrum of conditions for understanding the nature of rural FWPRs within each of them. 2. Infrastructure Table 3 gives the details of the infrastructural variables across the four districts. The Table shows the difference in the availability of infrastructural facilities across the four districts under study and shows Haora and Murshidabad to be more infrastructurally developed regions in contrast to the two other districts. We have listed the number of primary schools in each district, as they are the most frequently occurring educational institutions. 6 Medical facilities include hospitals, maternity and child welfare centres, maternity homes, child welfare centres, primary health centres, health centres, primary health sub centres, dispensary, family planning centres, tuberculosis centres, nursing homes, community health workers, registered private practitioners, subsidiary medical practitioners and other medical centres. While only 19.4 per cent and 11.45 per cent of the total villages studied, had any medical facilities in Puruliya and Maldah, the corresponding percentages were 59.35 per cent and 44.97 per cent in Haora and Murshidabad respectively. A second important criterion for measuring the level of infrastructural development was the availability of power. While 87.99 per cent of the villages in Haora and 76.81 per cent of the villages in Murshidabad had any power supply, these percentages were 57.75 per cent in Maldah and a mere 25.58 per cent in Puruliya. Poor climatic conditions in Puruliya emphasise the need for irrigation and though the Ganga and Mahananda rivers water Maldah, crisis results Table 3 Infrastructural Variables within Four Districts, West Bengal, Rural, 1991 Name of district Puruliya Haora Murshidabad Maldah Villages selected for study 2424 733 1910 1633 Villages with educational institutions 2172 704 1562 1155 Primary schools 2157 700 1538 1123 Villages with medical institutions 471 435 859 187 Villages with power supply* 620 645 1467 943 Power supply for agriculture* 0 46 573 347 Power supply for all purposes 104 192 228 43 Average area irrigated per village (%) 18.74 33.31 45.26 23.89 Note: *All villages with power supply may not have power supply for agriculture.

568 THE INDIAN JOURNAL OF LABOUR ECONOMICS in the district due to the absence of irrigation in the villages. We however, find that the average irrigated area per village to be extremely low in Puruliya and this reflects the infrastructural backwardness of the district given that the rural economy is mainly dependent on agriculture. The average percentage of area irrigated in Murshidabad and Haora is more than the corresponding percentages for Maldah and Puruliya. 3. Work Participation Rates Table 4 gives the summary statistics on the male and female work participation rates within the four districts under study. MWPRs are the highest on an average in Maldah followed by Puruliya, Murshidabad and Haora. The extent of variation of MWPRs as measured by the coefficient of variation (C.V.) is low, reiterating the conclusion that male participation in the rural labour markets is nearly uniform across these districts. An observation of FWPRs within the rural areas of districts in the Table indicates the difference in the participation rates of women in the labour markets across the rural areas of districts. Average rural FWPRs in Puruliya and Maldah are considerably higher than those of Haora and Maldah. Further, the coefficient of variation for female workers is considerably high even within the districts, suggesting that FWPRs differ within the districts as well. Expectedly, the intra-district variation in the percentage of total female workers is higher in the case of districts where FWPRs are below the state average such as Murshidabad and Haora than those where FWPRs are above the state average such as Maldah and Puruliya. Table 4 Summary Statistics for FWPRs and MWPRs for Four Districts of West Bengal, Rural, 1991 Male total workers Female total workers District Mean C.V. Mean C.V. Maldah 54.91 13.48 21.73 90.19 Haora 51.28 9.22 3.95 140.16 Murshidabad 52.72 10.33 9.43 146.15 Puruliya 53.34 14.18 40.91 50.09 Note: C.V. indicates the coefficient of variation; WPR is defined as the percentage of workers in the total population;total workers include main plus marginal workers. 4. Distribution of Female Population by Social Class Much of the literature analysing FWPRs in different parts of India has highlighted the importance of caste and ethnicity in explaining the differential rural FWPRs as norms and biases associated with caste and ethnicity (primarily tribal or non tribal) condition the participation of women in the rural labour markets. Table 5 shows the mean and the coefficient of variation of the female population by social groups within the four districts. While the mean and the coefficient of variation of the SC population seems to be almost same across the different districts, the mean of the ST population is definitely higher and the C.V. is lower within those districts where rural FWPRs are above the state average compared to those which are not. While the average percentage of ST females is 30.33 per cent within Puruliya (Puruliya has the second largest population of tribals in West Bengal) and 18.38 per cent within Maldah, this percentage is 0.21 per cent and

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 569 Table 5 Summary Statistics of the Female Population of Different Social Classes within Four Districts, West Bengal, Rural, 1991 SC ST SC/ST NSC/ST NST Statistic Mean C.V. Mean C.V. Mean C.V. Mean C.V. Mean C.V. Maldah 26.37 112.09 18.38 155.63 44.75 86.57 55.25 70.11 81.62 35.05 Haora 25.14 104.58 0.21 591.83 25.36 103.86 74.64 35.28 99.79 1.29 Murshidabad 18.93 126.69 3.56 354.86 22.48 121.64 77.52 35.28 96.44 13.08 Puruliya 17.35 135.61 30.33 115.29 47.68 71.81 52.32 65.44 69.67 50.20 Note: SC/ST indicates Scheduled Castes and Tribes; SC and ST indicate Scheduled Castes and Scheduled Tribes; NSC/ST and NST indicates non-scheduled Castes and Tribes and non-scheduled Tribes; Figures denote the percentage of the economic group to the total population. 3.56 per cent in Haora and Murshidabad respectively. The combined average population of SCs and STs is lower and the variation is higher within villages of Haora and Murshidabad compared to those of Puruliya and Maldah. Consequently the non-sc/st population is higher and the variation is also lower in Haora and Murshidabad compared to Puruliya and Maldah. While Haora and Murshidabad have 74.64 per cent and 77.52 per cent of non-sc/sts in the female population, Maldah and Puruliya have only 55.25 per cent and 52.32 per cent of non- SC/STs in the female population. 5. Literates among Male and Female Population An observation of Table 6 shows average female literacy to be higher within Haora and Murshidabad compared to Puruliya and Maldah. Further the c.v. within districts is also lower for Haora and Murshidabad compared to the other two districts, suggesting that there are more villages where female literacy is close to the average percentage and hence indicates a more uniform distribution of female literates across villages. Expectedly, the mean per cent of literates is higher and the C.V. within districts is lower for the males compared to their female counterparts within all districts. Table 6 Literacy Statistics across Four districts, West Bengal, Rural, 1991 Male Female District Mean C.V. Mean C.V. Maldah 31.91 42.35 14.93 70.71 Haora 59.61 19.58 40.50 29.96 Murshidabad 35.32 38.21 20.89 52.86 Puruliya 46.54 30.43 14.13 72.83 6. Occupational Classification of Women Workers Contrasts within the two sets of districts are also apparent from the occupational classification of female workers. While the percentage of women engaged in non- agricultural activities is considerably higher in the case of those districts with lower than average FWPRs, that in the agricultural activities is higher in the case of those districts with higher than average FWPRs (see Table 7).

570 THE INDIAN JOURNAL OF LABOUR ECONOMICS Table 7 Summary Statistics for Female Workers Engaged in Agricultural and Non-agricultural Occupations in Four Districts, West Bengal, Rural, 1991 AGL NAGL District Mean C.V. Mean C.V. Maldah 68.23 55.07 27.00 129.11 Haora 80.85 36.89 13.44 502.06 Murshidabad 28.93 124.97 60.86 66.40 Puruliya 25.37 112.70 35.51 208.53 Note: AGL and NAGL indicate agriculture and non-agriculture respectively; The table includes main workers only. IV. ANALYSIS OF RURAL FWPRs ACROSS FOUR DISTRICTS Having described the areas of study and highlighted the differences in the socio-economic scenarios, this section of the paper, attempts to explain the pattern of rural FWPRs within districts by using correlation and regression analysis. While, results from the correlation analysis in the next subsection helps to see the clear differences in the different correlates of FWPRs within the districts with relatively lower and higher FWPRs, the regression analysis attempts to explain the FWPRs within each district and thus identify contrasting areas of policy intervention in each district. 1. Correlation Analysis Table 8 gives the correlation coefficients between rural FWPRs and some other variables within the four districts. MWPRs are a rough indicator of the extent of employment opportunities available in the rural areas of districts as men usually being the chief breadwinners in the family are part of the work force whenever employment is available. We find that with the exception of Murshidabad, a strong positive correlation exists within all other districts between MWPRs and the corresponding FWPRs. 7 Higher MWPRs and FWPRs exist in the same villages as those may be villages offering greater job opportunities in all districts. Table 8 shows the percentage of SCs among the female population to have a strong positive correlation with the FWPRs for the districts, which had FWPRs above the state average. There is on the contrary no strong relation between FWPRs and the percentage of female SCs in the population within districts where FWPRs are below the state average such as Haora and Murshidabad. FWPRs among SCs in West Bengal are considerably lower than that Table 8 Correlation Coefficients between FWPRs and Other Variables within Four Districts, West Bengal, Rural, 1991 Districts SC ST SC/ST Non-SC/ST Non-ST FLIT MWPRs FAGL FNAGL Puruliya -.22* -0.22* -0.08* -0.08* -0.22* -0.31* 0.52* 0.45* 0.03 Haora 0 0.14* 0.01-0.01-0.14* 0.05 0.26 0.10* -0.10* Maldah 0.09* 0.56* 0.48* -0.48* -0.56* -0.37* 0.46* 0.36* -0.26* Murshidabad 0.03 0.36* 0.19* 0.19* -0.36* -0.24* 0.01 0.09* 0.07* Note: SC, ST indicate Scheduled Castes and Scheduled Tribes; FLIT indicates per cent of female literates; FAGL and FNAGL indicates the percentage of female workers in agriculture and non-agriculture; *indicates that the variable is significant at the 5 per cent level of significance.

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 571 in most Indian states and traditional biases against female manual labour could explain this phenomenon (see Sinha, 2001). The strong association between percentage of SCs in the female population and FWPRs in districts where FWPRs are above the average indicates the possibility of traditional biases against women s work to be possibly weaker in these districts. The strong positive correlation between proportion of female STs in the population and the FWPRs however holds for all districts. The nature of the tribal society, which is relatively free from the inhibitions and status consciousness affecting the behaviour of caste Hindu women and even SC women, maybe inducing higher FWPRs among tribal women. There exists a strong inverse correlation between FWPRs and the non-sc/st and also the non-st population within all districts. This result is true for all districts except Haora where the former correlation is not significant. The difference in the nature of association between FWPRs and the SC and the ST population in the districts with higher and lower FWPRs, perhaps indicates a growing trend of sanskritisation 8 among SCs in the more developed districts, causing them perhaps to embrace traditional biases against female work. We find a strong inverse relation between the percentage of female literates and FWPRs for all districts except Haora, suggesting that greater literacy may be deterring the participation of women in rural labour markets. The inverse relation between literacy and rural FWPRs may be explained in two ways. First, higher literacy may reduce FWPRs in rural areas by fostering a need for working in more lucrative occupations. Further, higher literacy among younger girls may imply greater time devoted to studies and consequent reduction in the time spent in working. Table 8 shows the absence of a clear dichotomy across the four districts regarding the association of rural FWPRs within districts and the percentage of female workers engaged in agricultural and non-agricultural occupations. We find that while employment in agricultural work is always strongly positively correlated with the rural FWPRs, the association of rural FWPRs with female workers engaged in non-agricultural occupations does not follow a uniform pattern across districts. While rural FWPRs are lower in areas with high percentage of nonagricultural female workers in Maldah and Haora and the correlation between overall rural FWPRs and the percentage of women engaged in non-agricultural activities is positive in Murshidabad. 9 Though such a positive relation holds for Puruliya too, the relation is very weak. Table 9 does not indicate a uniform relationship between social group and percentage of women workers in agricultural and non-agricultural activities across districts. The relatively greater involvement of ST women in agricultural activities is evident from the strong positive correlation between the proportion of STs in the female population and the percentage of women engaged in agricultural activities within all districts. Another important feature is the inverse relation between the percentage of non-sc/sts in the female population and the percentage of women engaged in agricultural activities within all districts, suggesting the low Table 9 Correlation Between the Percent of Female Workers in Agriculture and Non-agriculture and Proportion of Female Population Across Social Groups for Four Districts, West Bengal, Rural, 1991 Maldah Murshidabad Haora Puruliya Statistic AGL NAGL AGL NAGL AGL NAGL AGL NAGL SC 0.20* -0.19* 0.16* -0.18* -.02.01 0.13* -.02 ST 0.32* -0.35* 0.37* -0.31* 0.07*.04 0.16*.08 Note: *indicates that the correlation is significant; The table includes main workers only.

572 THE INDIAN JOURNAL OF LABOUR ECONOMICS Table 10 Correlation between FWPRs and Proletarianisation in Four Districts, West Bengal, Rural, 1991 Maldah Murshidabad Puruliya Haora 0.36* 0.20* -0.06* -0.18* Note: *indicates that the correlation is significant; The table includes main workers only. employment of non-sc/st women in agricultural activities. 10 There however appears to be a positive relationship between the percentage of non-sc/st and non-st women and the percentage of women engaged in non-agriculture for all districts except Puruliya, where this association is inverse. Participation in agricultural activities forms the largest occupational category for women workers across Indian states (see Duvvury, 1989). These results indicate the differences in the nature of work done by SC/ST (particularly ST) and non-sc/st (particularly non-st) women within West Bengal and also posits the hypothesis that traditional biases against agricultural participation of non-st (and non-sc/st in some cases) women coupled with the absence of non agricultural opportunities maybe important reasons explaining the low and varied nature of FWPRs within the state of West Bengal. Agricultural labourers form the lowest rung in the rural economic scale (see Ahluwalia, 1978). The strong association between proletarianisation 11 and the percentage of female workers in the population is found in all the four districts (Table 10). This finding corroborates the positive association between greater poverty and FWPRs in rural areas of Indian states in the case of Maldah and Puruliya. The association is however found to be inverse in Haora and Murshidabad. These results thus show that the positive relation between proletarianisation and rural FWPRs, true at the all-india level, holds within the districts where the rural FWPRs are above the average for the state. High rural FWPRs are on the other hand associated with low proletarianisation in the districts with below average rural FWPRs. Women, particularly in the rural areas of the country often have to bear the double burden of household tasks along with the work outside the household and the literature on FWPRs in India points out that the pressure of domestic duties could be an important factor deterring the steady participation of women in the labour market and restrict them to being marginal workers with intermittent work force participation. Table 11 shows that main and marginal rural FWPRs 12 are significantly correlated within all districts we are studying. 13 The inverse relation between female main and marginal WPRs found by Nagaraj (1989) in the case of Tamil Nadu and India from the 1981 Census data, was not however corroborated in our district level analysis of West Bengal (see Sinha, 2001). The village level data confirms such a relationship for the two districts of West Bengal where rural FWPRs Table 11 Correlation between Female and Main and Marginal Work Participation Rates for Four Districts, West Bengal, Rural, 1991 Maldah Haora Murshidabad Puruliya Product moment correlation -0.09* 0.10* 0.09* -.45* Rank correlation 0.03 0.23* 0.29* -0.43* Note: *indicates that the correlation is significant at the 5 per cent level.

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 573 are above the state average. In case of Haora and Murshidabad, the two districts with rural FWPR below the state average, main and marginal rural FWPRs are however positively related. 2. Regression Analysis We now attempt to identify the factors explaining FWPRs within the rural areas of the four districts considered in this analysis, by OLS regression of the FWPR (the dependent variable) on a matrix of explanatory variables (x). The regression analysis would help us to understand the partial effect of any explanatory factor (variable) within a framework where all the variables proposed to be associated with FWPRs are being considered in a multivariate framework. The following is the model that we have estimated: FWPR = a+bx+e The matrix of explanatory variables, x, have been divided into two groups of variables which relate to the characteristics of the population and variables which indicate the level of development of the particular region under consideration. The percentage of Scheduled Castes among the female population (FSC), the percentage of Scheduled Tribes among the female population (FST), the percentage of literates among the female population (FLIT) and the percentage of workers among the male population (MWPR) comprise the former set of variables. The total number of primary schools in the village (SCH); a dummy variable indicating the presence or absence of any kind of medical facility (MEDINST); a dummy variable indicating availability of power supply for agricultural usage (PAGL) and the percentage area irrigated (IRR) comprise the variables which indicate the level of economic development of the village. 14 Table 12 gives the summary of the regressions results for each district. The summary results show the percentage of STs in the female population to have a significant positive effect on rural FWPRs in all districts studied. Given the strong positive association between FWPRs and the percentage of STs in the female population at the district level and the high FWPRs among the ST population, this result was as expected. The percentage of SCs in the female population is a significant variable affecting rural FWPRs in districts where rural FWPRs are above the state average. We find that though this percentage affects rural FWPRs in a positive way in Maldah, the effect is negative in the case of Table 12 Summary of Regression Results: Dependent Variable-FWPR District Puruliya Maldah Murshidabad Haora Independent variable Intercept -19.51* -15.27* 20.42* -12.05* FST 0.03* 0.28* 0.36* 0.60* FSC -0.07* 0.07* 0.02 0.00 FLIT -0.31* -0.25* -0.24* 0.01 MWPR 1.21* 0.60* -0.10*** 0.30* SCH 0.35 1.52 0.35 0.02 MEDINST 1.47 2.22 0.58 0.23 PAGL/PALL -3.37*** -3.71* -4.56* 0.18 IRR -0.01-0.01-0.03* 0.01 R 2 0.31 0.39 0.21 0.09 Adjusted R 2 0.31 0.38 0.21 0.08 Note: We have used PALL instead of PAGL for Puruliya; Figures in brackets indicate the standard errors for the estimates; *, *** indicate 1 per cent and 10 per cent level of significance respectively.

574 THE INDIAN JOURNAL OF LABOUR ECONOMICS Puruliya. The effect of literacy on rural FWPRs is significantly negative in all districts except Haora. The inverse association between literacy and FWPRs has often beem found in the Indian context, and is to be interpreted with caution and the knowledge that higher FWPRs are often a reflection of greater distress in the country. Male work participation rates have a significant positive effect on rural FWPRs in all districts studied except Murshidabad, where this effect is inverse. However, the estimate while significant at the 1 per cent level in Haora, Puruliya and Maldah, is only significant at the 10 per cent level for Murshidabad. We may infer from this that opportunities for employment as broadly indicated by the MWPRs affects rural FWPRs within districts strongly. While this may induce higher rural FWPRs in districts, which are not infrastructurally well developed, high, MWPRs may reduce rural FWPRs in relatively infrastructurally advanced districts such as Murshidabad. The latter finding corroborates the inverse association between between FWPRs and district prosperity alluded to earlier. The number of primary schools and the presence of medical facilities do not have a significant effect on rural FWPRs in any of the districts. A strong inverse effect of electricity for agricultural use on rural FWPRs is found for all districts except Haora. Further, a significant inverse relation between percentage of irrigated land and rural FWPRs is also found in Murshidabad. We thus find limited effect of infrastructural variables in explaining FWPRs in rural Haora. For all other districts the variables indicating infrastructural development in agriculture seems to be having an inverse effect on rural FWPRs. The values of the R 2 for all models suggests that our model specification may be better in the case of districts with higher than average rural FWPRs, than those where the rural FWPRs are below the state average. This may also indicate that the role of factors other than those specified may be stronger in the case of those districts where rural FWPRs are below the state average, hence leading to the hypothesis that the role of traditional aversions towards female participation in the rural labour markets may be stronger in the case of those districts where rural FWPRs are below the state average. The only factor which seems to be uniformly associated with high rural FWPRs at the village level is a high proportion of STs in the female population. While the presence of a large percentage of male workers induces higher rural FWPRs within three of the four districts, the effect is reversed in Murshidabad. The SC population does not affect rural FWPRs in the two districts where rural FWPRs are below the state average; however, while a high SC population induces higher rural FWPRs in Maldah, the effect is reversed in Puruliya. Other than Haora, high percentage of literates among the female population implies lower rural FWPRs, in all other districts. The effect of variables indicating the level of infrastructural development in agriculture on rural FWPRs also differs across districts. While better agricultural development does not seem to have any effect on FWPRs in rural Haora, they induce lower FWPRs in Puruliya, Maldah and Murshidabad. V. CONCLUSION The role of non-economic factors in explaining rural FWPRs may be stronger in districts with lower FWPRs in contrast to those with higher FWPRs 15, among the four districts of study. Data on infrastructural variables suggests that Puruliya and Maldah, with high rural FWPRs were relatively more infrastructurally backward compared to Haora and Murshidabad. The relatively high level of urbanisation, dissemination of literacy and infrastructural development

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 575 in Haora and Murshidabad coupled with the high percentage of non-sc/st population perhaps makes the pattern of FWPRs in the rural labour markets of these districts markedly different from the relatively backward districts of Puruliya and Maldah. This might be indicative of the fact that an important factor explaining the varied FWPRs in rural West Bengal may be the level of district development and high rural FWPRs are characteristic of backward districts. A study of MWPRs and FWPRs within districts showed considerable variation in the latter variable particularly within those districts with rural FWPRs lower than the state average. There were higher average number of SC/STs, lower number of non-sc/sts and lower number of literates in Puruliya and Bankura, relative to Haora and Murshidabad. High rural FWPRs are associated with high MWPRs (in all districts except Haora) and low levels of female literacy (in all districts except Haora). The relation between main and marginal rural FWPRs was positive in Haora and Murshidabad and inverse in the other districts, suggesting that domestic work may be impeding rural FWPRs in certain parts of districts with high FWPRs and not those with low FWPRs. While the strong positive correlation between rural FWPRs and the percentage of STs in the female population is true for all districts, the significant positive association with rural FWPRs with the female SC population holds only for districts with high rural FWPRs. While high rural FWPRs were found to be correlated with high percentage of women workers engaged in agriculture, no such correlation was found in the case of FWPRs and workers engaged in non-agriculture except Murshidabad. While ST women may be more engaged in agricultural activities, non-sc/sts within districts desist from participating in agricultural activities. Non-agricultural workers among the females are however related positively with the non-sc/st population within districts. Regressing FWPRs on a set of population and infrastructural variables, we found that most of the former variables affect rural FWPRs strongly. The presence of a high percentage of STs in the female population strongly affects rural FWPRs in all districts. Low literacy levels and high levels of MWPRs also imply high rural FWPRs in all districts except Haora. Our regression however yields mixed effects of variables indicating the level of agricultural development on rural FWPRs. These range from the absence of impact in Haora to inverse impacts in Murshidabad Maldah and Puruliya. Three interesting implications may be drawn from our village level analysis of rural FWPRs within four districts of rural West Bengal. First, while it is generally difficult to generalise the association of rural FWPRs with any relevant variable at the all-india or even the state level, it is possible to find associations between some variables and FWPRs within groups of villages in a district, though the nature of association differs across districts. This shows the strong influence of region specific factors on FWPRs. The role of non-economic factors (often not lending themselves to quantification and thus difficult to incorporate in statistical analysis) in explaining rural FWPRs may be stronger in districts with lower FWPRs in contrast to those with higher FWPRs. The regression models for Puruliya and Maldah had a better fit than Haora and Murshidabad. While working as agricultural labourers forms the largest occupation for women in the rural areas of most Indian states (see Duvvury, 1989), the sectoral composition of the West Bengal female labor force shows relatively higher participation of women in non-agricultural activities. A study of the occupational classification of female workers within West Bengal shows an almost clear dichotomy between districts with having high employment of women workers in agricultural and non agricultural activities, suggesting that the absence of

576 THE INDIAN JOURNAL OF LABOUR ECONOMICS opportunities for non agricultural employment could be a possible cause for low rural FWPRs in those districts, which in turn maybe lowering the average FWPRs for the state. Institutional biases and a tradition of non- participation in agriculture (see Sinha, 2001) might thus be reasons contributing to a low rural FWPR in the state. Absence of such bias among the STs results in higher than average FWPRs in the areas of the state with high ST populations. Thirdly, association of rural FWPRs with variables such as the extent of proletarianisation and also the association between the percentage of female main and marginal workers follow different patterns for the districts with relatively high and low rural FWPRs. The essentially different nature of rural FWPRs in regions with relatively low and relatively high female participation in labour markets suggests a targeted approach for studying rural FWPRs, as essential for an improvement in independence and decision-making power vested in women within and outside the family. Notes 1. The Census of India 1991 defines work as any economically productive activity whichis physical or mental in nature. This also includes the supervision or the direction of work. This definition excludes from its purview the vast bulk of invisible non-marketed productive activities a big section of the Indian women population are commonly engaged in. 2. The general context is the economic, social and the ecological factors which affect the whole society and economy such as the cropping pattern, the level of SC/STs in the population or the level of income. The specific factors are those that affect women alone alone such as as the pressures of housework. 3. 1991 is the last year for which complete village level data (included in the village directories) has been published by the Census of India. Such data is not yet available for the 2001 Census. 4. Much of the apparently dramatic increase in the FWPRs in rural West Bengal in 2001 comes from an increase in the marginal work participation ie greater participation of workers who had worked for less than 180 days in the previous year. According the West Bengal Human Development Report, 2004 much of this increase could be apparent rather than real, on account of better enumeration and recognition of women workers in the 2001 Census of India. 5. Revenue collection in West Bengal in the colonial period was based on land taxes. Land tax was collected by dividing the total area under cultivation into villages for facilitating the collection of revenue. Hence there could be villages with zero or negligible inhabitants. It is probable that the same administrative classification was yet retained in West Bengal, during the Census data collection. 6. Primary School, Middle School, High School, PU College, Graduate College, Adult Literacy Center, Industrial School, Training School and Other Schools comprised the entire set of educational institutions. 7. Though the relation between the two variables is positive in the case of Murshidabad, the correlation is not significant. 8. The term sanskritisation coined by M.N. Srinivas refers to the tendency of the SC populations in prosperous regions to emulate the behaviour of the non SC poplation in order to rise up the social scale. 9. Women have traditionally been engaged in the silk industry in rural Murshidabad though the role has been diminishing across years. 10. This is also true of the association between the percentage of non-sts and percentage of women workers engaged in agriculture. 11. Proletarianisation has been defined in the literature as the percentage of agricultural labourers among the main workers (Nagaraj, 1989). 12. The Census of India defines main workers as those that have been working for 180 days or above in the previous year while marginal workers are those that have been working at anytime of year. 13. The rank correlation between main and marginal Rural FWPR is not significant in the case of Maldah. 14. Since very few villages in Puruliya had power supply for agriculture, a dummy variable indicating availability of power supply for all purposes (PALL) has been used in the case of Puruliya. 15. The regression models for Puruliya and Maldah had a better fit than Haora and Murshidabad.

FEMALE WORK PARTICIPATION RATES IN RURAL WEST BENGAL 577 References Ahluwalia, Bina (1978), Rural Poverty and Agricultural Performance in India, Journal of Development Studies, Vol. 14, No. 3, April. Anker, Richard (1983), Female Labour Force Participation in Developing Countries: A Critique of Current Definitions and Data Collection Methods, International Labour Review, Vol. 122, No. 6, December. Banerjee, Nirmala (1989), Trends in Women s Employment 1971-81: Some Macro Level Observations, Economic and Political Weekly, Vol. 24, No. 17, April. Bardhan, Kalpana (1977), Rural Employment, Wages and Labour Markets in India: A Survey of Research, Economic and Political Weekly, Vol. 12, No. 26. (1985), Women s Work, Welfare and Status: Forces of Tradition and Change in India, Economic and Political Weekly, Vol. 22, Nos. 50, 51 and 52, December. Bennett, L. (1992), Women, Poverty and Productivity in India, Seminar Paper No. 43, Economic Development Institute, The World Bank, Washington, D.C. Deere, Carmen Diana and de Leal, Magdalena Leon (1982), Women in Andean Agriculture, International Labour Organisation, Geneva. Dholakia, Bakul, H. and Dholakia, Ravindra, H. (1978), Inter-state Variation in Female Labour Force Participation Rates in India, The Indian Journal of Labor Economics, Vol. 20, No. 4, January. Duvvury, Nata (1989), Work Participation of Women in India: A Study with Special Reference to Female Agricultural Labourers, 1961 to 1981, in Jose, A.V. (ed.). Ghosh, Arun (1989), West Bengal Landscapes, K.P. Bagchi and Company, New Delhi/Calcutta. GoWB (2004), West Bengal Human Development Report, Development and Planning Department, Government of West Bengal. Gulati, Leela (1975), Female Work Participation: A Study of Inter-state Differences, Economic and Political Weekly, Vol. 10, Nos. 1 and 2, January. Jose, A.V. (ed.) (1989), Limited Options: Women Workers in Rural India, Asian Regional Team for Employment Promotion and World Employment Programme, International Labour Office. Krishnaraj, Maithreyi (1990), Women s Work in Indian Census: Beginnings of Change, Economic and Political Weekly, Vol. 25, Nos. 48 and 49. Lieten, G.K. (1996), Continuity and Change in Rural West Bengal, Sage Publications, New Delhi/New Bury Park, London. Mencher, Joan and Saradamoni, P. (1982), Muddy Feet, Dirty Hands: Rice Production and Female Agricultural Labour, Economic and Political Weekly, Vol. 17, No. 52. Nagaraj, K, (1989), Female Workers in Rural Tamil Nadu: A Preliminary Study, in Jose, A.V. (ed.), op.cit. Nath, Kamala (1970), Female Work Participation and Economic Development, Economic and Political Weekly, Vol. 15, No. 31, May. Sinha, Sucharita (2001), Female Work Participation in Rural India: A Case Study of Rural West Bengal, 1961 to 1991, M.Phil. Dissertation, Indira Gandhi Institute of Development Research, Mumbai. Sundaram, K. (1989), Inter-state Variations in Participation Rates of Women in India: An Analysis, in Jose, A.V. (ed.), op.cit. Visaria, Praveen (1997), Women in the Indian Work Force: Trends and Differentials, Artha Vijnana, Vol. 39, No. 5. Read our Publications! The Institute of Employment Rights A think tank for the labour movement Our aim is to provide a wide variety of high quality publications which we hope will stimulate debate and analysis about employment law policies and legal developments in industrial relations. The results of the work of the Institute are published in booklets available for sale or through annual subscription. The Institute also provides short articles (free of legal jargon) for trade union journals and other popular publications. It organises seminars on topics of particular importance and holds occasional lectures. The Institute of Employment Rights, 177 Abbeville Road, London SW4 9RL, UK, 020 7498 6919, Fax: 020 7498 9080, Email: office@ier.org.uk

578 THE INDIAN JOURNAL OF LABOUR ECONOMICS