Landless but not assetless: female agricultural labour on the road to better status, evidence from India

Abstract

Women's share of agricultural wage employment is rising across the Indian sub-continent. Studies examining this process of feminisation tend to be divided along the lines of an ideological debate following either the ‘poverty-push’ or the ‘demand-pull’ argument. This debate, however, has largely ignored the institution of patriarchy. In this study we revisit the debate with a focus on gender relations. The study is based on data collected from 291 households in the state of Andhra Pradesh. We find that, despite increased labour market participation, women's household status, her wages and working conditions remain acutely depressed. Women labourers with access to productive assets, however, are effectively expanding their agency within both the household and the labour market. Their experiences, we argue, have implications for transformative policies.

Keywords:

• feminisation of agriculture labour
• women's status
• Andhra Pradesh
• India

Notes

I would like to acknowledge the financial support received from the Department for International Development (award number R7617) and Newton Trust (award number INT 2.05d). I am deeply indebted to my enumerators for the time and effort they gave to this project: Achari, Chandrasekhar, Lakshmama, Narsimhlu, Lakshmi, Padma, Ravi, Rathish, and Sridevi. I am also grateful to Barbara Harriss-White, G.N. Rao, Jun Borras, Wendy Olsen and two anonymous referees for comments on earlier versions of this paper.

¹There are important exceptions like Bardhan (1985a, 1985b), Kapadia (1995), da Corta and Venkateshwarlu (1999), and Olsen and Ramana Murthy (2000). Caste-based discrimination is also widespread in India's rural labour markets (see, for instance, Thorat 2004).

²To some extent, the excellent study by da Corta and Venkateshwarlu (1999) is an exception. However, da Corta and Venkateshwarlu also fall short of undertaking a detailed empirical investigation into gender relations within the household.

³A total of 302 households were surveyed, but six de facto female and four de facto male- headed households and one income outlier were excluded from the analysis. To ensure that interviews were carried out in a congenial setting and that the data collected was reliable, all interviews were carried out by a group of two interviewers: one male and one female. Also the author was present at around one-third of all interviews. For more details on methodology and survey protocol see Chapter 2 in Horrell et al. (2008).

⁴A systematic analysis of male and female responses did not indicate a gender bias in responses, except in the case of the gender of the head of the household.

⁵The correctness of the official poverty figures is intensely debated (see, for instance, Deaton and Drèze 2002).

⁶Historically, seasonal migration has been a significant livelihood strategy in Mahabubnagar and migrant labourers from here are famously referred to as ‘Palamur labour’ (see Olsen and Ramana Murthy 2000). Estimates suggest that in some parts of the district seasonal migration is as high as 60 percent (Sajja 2003). Deshinkar and Start (2003) report similar intensities for various villages of AP. Drought of course is the main reason for such high levels of migration, but under-investment in irrigation further amplifies the problem (also see Breman 1996, Rao 2001). Estimates suggest that of the 1.413 million hectares of cultivable land available in Mahabubnagar, only a meagre 10.4 percent is irrigated (Sajja 2003). Garikipati (2008b) investigates the linkages between migration and gender relations in rural Mahabubnagar.

⁷Recent evidence suggests that although the incidence of ‘permanent attachment’ or ‘bondage’ is on the decline, the incidence of ‘non-permanent attachments’ is on the rise, especially in dry regions of the country (see Jodhka 1994, Subramaniam and Reddy 1994). The study by da Corta and Venkateshwarlu (1999), which examines unfree labour relations for the Chitoor district of AP for the period 1970–1995, highlights this phenomenon with specific reference to AP. An increase in non-permanent attachments can at least partly be attributed to the financial sector reforms in rural India which saw the cost of formal credit increase enormously in the late 1990s. This resulted in heavy indebtedness to the informal sector and, in combination with drought and rising prices for farm inputs, is held responsible for the ‘debt suicides’ by farmers in AP over the last few years (Foster 2004, Patnaik 2004).

⁸In Karimnagar, another district of the Telangana region in AP, Revathi (2008) finds that, among the forward caste, the on-farm work intensity among female labour increased while men diversified into non-farm and trading activities. Among the lower caste, households receiving remittances from male migrants saw withdrawal of their women from agricultural wage labour as a reflection of their improved status.

⁹Statutory wages as published in Government of Andhra Pradesh (2001).

¹⁰For theoretical treatment of this issue see Basu (2006), McPeak and Doss (2006).

¹¹For a historical perspective on land reform legislation in AP see Rao and Reddy (2008).

¹²Largely due to cultural reasons, women are unlikely to participate in the actual sale of cattle. This downward bias against women's control over the process of selling cattle needs to be noted when analysing the results.

¹⁴Migration incomes were not included because this creates a bias against women who do not migrate.

¹⁵Given the small proportion of female land owners, clustering assets is a common approach adopted by studies interested in the relationship between asset ownership and women's empowerment. For instance, studying women's autonomy in India, Jejeebhoy (2000) considers an index of control over economic resources which includes land but also includes other resources like jewellery and so on. Also see Jejeebhoy and Sathar (2001).

¹⁶This problem is referred to as ‘selection bias’ and is commonly encountered in impact evaluation studies. Not accounting for it may result in overestimation of the impact of the variable of interest (see Morduch 1999, Pitt et al. 1999). Selection bias occurs if the asset owners differ from non-owners in unobservable characteristics. If these characteristics are related to the empowerment measures studied here then the coefficient of ASSETS will reflect these effects and will be biased. Previous studies that have examined the relationship between land ownership and woman's status suffer from similar issues (for instance, Mason 1998, Allendorf 2007).

¹⁷Despite these efforts we acknowledge that the way we have defined this variable remains potentially problematic. For instance, one can argue that a gift from a maternal home or membership in a microfinance group may itself not be exogenous to the household's processes. For instance, gifts made by the woman's parents may themselves be determined by the status that she enjoys in her household. To solve this potential endogeneity problem we should have used valid instruments. For example, we should have collected more detailed information on the process surrounding women's asset gains and used this to endogenise ASSETS. Unfortunately, our dataset does not have suitable instruments. We therefore suggest that the results be treated with caution.

¹⁸Although we have information on years of school attended by the women labourers, lack of variation among the observations compelled us to convert this into a dichotomous variable.

¹³Hashemi et al. (1996) and Garikipati (2008c) are among other studies that have employed this method to construct women's empowerment indicators.

¹⁹It is important to note that this finding is relevant for a specific context and must not be interpreted as general evidence against the potential of increased participation in work outside the home for women's empowerment.