Is policy convergence required to improve women’s empowerment in agriculture? Evidence from West Bengal

ABSTRACT

Farmers’ use of agricultural technology is necessary for reducing poverty. However, women make up most poor smallholder farmers in India and are generally unable to access relevant opportunities. Tandem to supporting women farmers in technology adoption is increasing their empowerment. While women’s empowerment has been on policy agendas in India for several years, progress has been slow and results are mixed. Using primary data from West Bengal, this study finds that female farmers with access to agricultural technologies will likely have social advantages over those without, but institutional and social support is necessary to facilitate real empowerment and overall development.

KEYWORDS:

• Women’s empowerment
• convergence
• irrigation technology
• gender
• agricultural development
• West Bengal

Introduction

The role of women in agriculture is increasingly important in many parts of the developing world (Anderson et al., 2021), as is the part played by new technology in sustaining and enhancing agricultural productivity (Aldy et al., 1998). Women make up nearly 75% of farmers in India, producing up to 80% of the country’s food (Kamdar & Das, 2021). Even though supporting women in agriculture is a critical component of India’s overall development agenda, a gender gap results in female farmers often being undervalued and overlooked by some decision-makers. Women also have limited access to agricultural resources and technologies (Alex, 2013; Kamdar & Das, 2021; Rola‐Rubzen et al., 2020; World Bank et al., 2008), and yet agricultural technology is often necessary for reducing poverty because farming technologies contribute to increased production, productivity and farm incomes (Bellotti & Rochecouste, 2014; Dixon et al., 2020; Lipton, 2006; Namara et al., 2010; Thirtle & Piesse, 2007).

As ‘gender equity enhances growth and development, not only for women but for all’ (Rola‐Rubzen et al., 2020, p. 115), working towards women’s empowerment in agriculture is also espoused as wise economic decision-making (Asian Development Bank, 2013). However, empowerment is a complex concept that arguably goes beyond simply accessing technology.

According to Hennink et al. (2012), empowerment occurs through six mechanisms: knowledge, agency, opportunity, capacity-building, resources and sustainability. Overlaying these are five domains of empowerment that cover health, economic, political, resource and spiritual elements. Further complexity occurs, with empowerment potentially taking shape at the individual, community and organizational levels. There is also possible interdependence between some components, whereby an activity may create empowerment in one domain and spill over to others. Alternatively, a lack of empowerment (disempowerment) in one domain may stifle empowerment in others.

Hennink et al. (2012) further explain that identifying individual domains of empowerment and highlighting the linkages between them allows for a greater understanding of the complexity of empowerment and helps identify opportunities to strengthen targeted initiatives. Two broad interpretations can be taken from these relationships. First, if empowerment in one domain automatically spills over to others, governments can potentially economize on their efforts and simply target the ‘driving’ domain. Second, it may also be the case that government efforts that focus on a single domain may be limited, and interventions that target multiple domains concurrently may be more effective than tackling each domain separately. This second proposition is based on likely synergies and positive feedback between domains, which means the collective outcome from tackling two areas simultaneously is greater than were those areas targeted separately.

The challenge with this second proposition is establishing effective partnerships across government departments or non-government organizations. Set against this ambition are the transaction costs of having so many players collaborate that the effectiveness of individual initiatives is potentially diminished. This then raises empirical questions about when combining multiple state government efforts across different domains is least costly and most effective at lifting the status of women. Arguably, this is most likely to be the case when there are existing positive relationships between variables of interest. For example, suppose education and health are positively correlated for women. In that case, we might expect that it will be less costly (more effective) to promote both simultaneously than if they were negatively correlated. Alternatively, positive correlation could also support the view that government effort directed specifically at education, for instance, would automatically bring better health outcomes for women.

This paper specifically considers the complexity of women’s empowerment and how its achievement can be supported by the efforts of government. Using primary data from West Bengal, India, we specifically investigate women’s access to irrigation pumping technologies and its relationship to other behaviours that might enhance empowerment. The paper adds to the literature on women’s empowerment in agriculture and increases our understanding of gender in the context of agriculture for development.

The remainder of the paper is structured as follows. The next section briefly explores a range of policy approaches undertaken in India before highlighting four initiatives that can promote the empowerment of women farmers. Whilst not comprehensive, these provide some insight into the complexity of women’s empowerment generally and the range of actions already trialled. The third section introduces the notion of policy convergence and explores how it might be used in the context of enhancing empowerment for women in agriculture. The fourth section details the research methods and describes the development and deployment of a survey in West Bengal. We present the results of this survey in the fifth section and consider policy implications in the sixth section before offering some brief conclusions.

Policy settings to enhance women’s empowerment

Women’s empowerment is particularly important and sensitive in South Asia. All countries in South Asia assign equal constitutional rights to women and men, a legacy of independence from British rule in the mid-20th century. Arguably, genuine equality and empowerment have taken longer to emerge and remain a work in progress, with women’s empowerment being a specific focus of governments for the last 25 years.

In India, the first draft national policy, titled the ‘National Policy for the Empowerment of Women 1996’, was adopted as a policy with the same title in 2001. A draft revision of the policy emerged in 2016, although the finalization of that policy is yet to occur. The National Ministry of Women and Child Development (MWCD) provides a review of the Government of India’s initiatives to empower women over the last decade, and these broadly cover four domains.

First, schooling is a focus of many programmes with the ambition of providing ‘equitable access to quality education to all students’ (MWCD, 2022, p. 1). Schemes such as Samagra Shiksha, Babu Jagjivan Ram Chhatrawas Yojna and Swacch Vidyalaya Mission lie within this approach. Second, employment and training are targeted as a means of enhancing empowerment. Notably, the Mahatma Gandhi National Rural Employment Guarantee Act, 2005, which provides a guarantee of paid employment, allocates no less than one-third of jobs to women, which has significant implications for women in rural areas. In contrast, the Pradhan Mantri Kaushal Vikas Kendras scheme seeks to provide additional opportunities for women to engage in trades and areas requiring higher skill development. Third, personal safety is given a high priority, and the Nirbhaya Fund has underpinned several initiatives, including projects by the Bureau of Police Research and Development and the Emergency Response Support System. Fourth, women’s health has been targeted with a wide range of interventions undertaken in partnership with the states. Amongst others, this has resulted in additional assistance to pregnant and lactating women and enhanced support for mental health (MWCD, 2022).

MWCD (2022) references data from the National Family Health Survey as evidence of the success of these interventions. Specifically, it is noted that the number of women holding a bank account in their own name has increased by 25% in the last five years and other assets, such as mobile phones, are increasingly owned by women. A range of other changes are also attributed to these collective interventions, such as improvements in the sex ratio and declining spousal violence. While these types of changes are welcome, the attribution of outcomes to specific policies or programmes is problematic. Similarly, data drawn from aggregated national changes run the risk of disguising nuances, such as different trajectories for rural women versus their city-based contemporaries.

A formal review of the Indian government’s policies and programmes in the women and child development ‘sector’ was undertaken by the Development Monitoring and Evaluation Office (DMEO), NITI Aayog, in 2021. The review comprised quantitative analysis of secondary data, administration of a household survey across 12 jurisdictions and a large number of key informant interviews. Broadly speaking, the review found mixed results from the different programmes and policies. The review also noted that opportunities for unlocking synergies across government programmes were often not realized and consequently recommends inter-scheme and programme ‘convergence’. Specifically, ‘there is a need to adopt a multi-pronged strategy and convergent approach to achieve empowerment of women and girls’ (NITI Aayog & Development Monitoring and Evaluation Office, 2021, p. 22). Arguably, this recommendation aligns with the earlier proposition that one vehicle for improving empowerment is to tackle positively related domains simultaneously, thereby exploiting positive feedback to secure disproportionate improvements for women.

It is important to note that this recommendation is aimed at the national level and makes no specific distinction between the empowerment of rural and urban women. The recommendation for convergence in the case of women’s empowerment suggests clustering of efforts that relate to capacity building and training (e.g. Mahila Shakti Kendra), micro-credit provision (e.g. Rashtriya Mahila Kosh), accommodation for working women, gender budgeting and encouragement of entrepreneurship (e.g. Mahila e-haat). Nonetheless, this raises questions about how convergence might be approached in more targeted locations, for example, in a rural area in a specific state. More specifically, will the convergence of some approaches that seek to raise the empowerment of women lead to a noticeable improvement in women’s empowerment generally and what should be converged at that level?

In order to explore this question, we focus on four discrete areas that can help contribute to women’s empowerment in agriculture: collective institutions, claims to land, rights over water and access to technology. Although by no means exhaustive, these themes were chosen due to their prevalence in the literature (e.g. Bjornlund et al., 2019; Hennink et al., 2012; Namara et al., 2010) and are reflective of the domains of the Women’s Empowerment in Agriculture Index (Alkire, Malapit, et al., 2013). Land claims, rights to water and access to technology all relate to decision-making power in agricultural production while being part of collective institutions relates to community leadership.

Establishing collective institutions

The roles played by women and men in local power structures of user rights and control over community resources are different. Generally, women’s presence in agriculture, policymaking and local governance in developing countries is low (Demissie & Abebe, 2017). Higher literacy rates among men, greater male access to leisure time, and perceptions about leadership and technology use perpetuate this imbalance (Wallace & Wilson, 2005). When women participate in local institutions, it is often unclear whether this has real value or meaning or is merely tokenistic, both individually and in broader society (Okali, 2006).

Good governance ensures people in collective organizations are included. For example, rural women in developing countries participate in different types of groups related to agriculture, including self-help groups (SHGs), producer associations, voluntary and business groups, natural resource management groups, and credit groups. Aside from access to economic resources, these groups allow women to develop their organizational skills, self-esteem and social networks (Agarwal, 2009; Kabeer & Matin, 2004). Women’s cooperative societies can steer individuals towards a path of empowerment (Mukhopadhyay et al., 2023, p. 152); however, the degree to which women members benefit from these groups seems to vary considerably.

One example of a collective institution that supports empowerment and that is relevant to the study region is SEWA Bharat, a federation of self-employed women’s associations covering all of India. SEWA has over 800,000 women affiliates, where two-thirds are small farmers or landless labourers. SEWA’s strategy for empowering women includes improving their assets and employment opportunities and negotiating with people in positions of power on their members’ behalf (Rose, 1992). It has also created the All-India Women Farmers Association, a designed space for enhancing skills, opportunities and development of women’s leadership (Gupta et al., 2017).

SHGs are the subunits at which collective organizations operate and are a crucial component of India’s state and national government development (Bali Swain & Wallentin, 2017; Dulhunty, 2021). SHGs in India usually involve 10–20 women and act as vehicles for accessing microcredit and skills training (Finnis, 2017). In addition, SHGs are commonly characterized by the formation of strong relationships, trust and intimacy between members. Members of SHGs in West Bengal were shown to have lower loan default rates than other credit arrangements, which was attributed to members having greater trust in one another (Feigenberg et al., 2010). More generally, the strong relationships built in SHGs allow women to draw on the group’s collective strength for action initiatives (Finnis, 2017) and make positive changes in their communities (Knowles & Luke, 2018).

Overall, there is a body of evidence across South Asia that SHGs are one initiative that can raise the status of women by providing social support that gives a greater community voice to women. Whether SHGs can be combined with other initiatives to gain even more ground for women is an open question.

Women’s claims to land

Women’s land ownership and control are significant factors in enhancing empowerment (Agarwal, 2003). However, women’s direct access to and ownership of land is affected by land policies backed by customary practices and gender-biased inheritance laws (Demissie & Abebe, 2017). The small percentage of females who head households in India are often widows, divorced women or females becoming heads of home due to the out-migration of the male household head (Demissie & Abebe, 2017).

Gender-selective out-migration has created a labour constraint on the farm and households. While men leave rural areas in search of work, the women left behind are required to take up the household and agricultural activities traditionally performed by men. Although women’s roles in agriculture have increased significantly, rural women still face serious constraints while performing these roles, including the absence of land titles (Mukhopadhyay et al., 2023).

Weak property rights lead to suboptimal decisions and missed opportunities to increase productivity (Rural Development Institute [RDI], 2009). Ownership of land or legal land rights is important because it provides access to critical markets and non-market institutions (Demissie & Abebe, 2017). Land ownership is thus a step towards enhancing women’s economic and social empowerment and reducing multilayered discrimination (Mukhopadhyay et al., 2023). For instance, many government support programmes in India hinge on land ownership, such that the absence of documented ownership prohibits access to support (Kishore, 2021). For most women, lacking a land title prevents them from accessing institutional credit, crucial inputs, federal services and new agricultural technologies (Asian Development Bank, 2013). As governments move towards cash transfers and away from input subsidies, tracing farmers to land will be an increasingly critical part of such schemes, and this does not bode particularly well for women farmers.

In contrast, it has been shown that having a woman on a land title benefits the woman and the entire family (Allendorf, 2007). An example of this is the Nijo Griha, Nijo Bhumi (NGNB) programme instated by the Department of Land and Land Reform in West Bengal, which redistributes government land to landless farmers. The NGNB provides micro-plots with land titles and explicitly stipulates that land titles issued to dual-headed households must be only in the women’s name or jointly titled to the male and female heads (Santos et al., 2014). Families in this programme have been shown to have improved tenure security, credit and investments in improved agricultural inputs. Women’s decision-making over household food and farming assets and processes has also increased. The extent to which these improvements from the NGNB programme have been sequenced with other interventions is unclear.

Women and water

As with land, women in India have limited claims over water rights and management. In part, this stems from how water is tied to land under Indian law – namely, the 1882 Easement Act – regardless of the widely accepted fact that women are the principal water users in India (Khandker et al., 2020). Irrigation is considered the domain of men, and they are the main participants in irrigation programmes and policies. Typically, women’s preferences and requirements are not generally considered during the design of water-related policies. Even though men and women may share similar needs for water, their opinions and preferences in irrigation specifics may differ materially (e.g. Lountain et al., 2023), potentially leading to different outcomes from ‘universal’ decisions about water and irrigation.

Most countries worldwide are now transferring surface water irrigation management from the government to farmers. In India, this has led to the formation of local water-user associations (WUAs). Khandker et al. (2020) note that this model requires a significant emphasis on participation and empowerment and yet surface water institutions in India have a very poor representation of women. In WUAs, water rights are assigned to beneficiary ‘households’ (Gulati, 2016), which are mainly headed by men. There is also some evidence that women’s low involvement in water management is driven by a lack of familiarity with speaking in public spaces, low levels of education, lack of recognized authority in this domain, perceived limitations around the scope for participation, and disregard of their opinions (Khandker et al., 2020). This raises questions about whether women with greater confidence arising from membership of SHGs might also be more amendable to expressing their views on water issues.

While surface water irrigation is important in India, groundwater irrigation is arguably more widespread. Here, pumping technology is a prerequisite to groundwater access, and women’s interactions with technology thus have spillover effects to groundwater.

Women’s access to technology

Pattnaik et al. (2018) recognize that having the right to take water is of itself meaningless without adequate technology to use it effectively in agricultural production. In addition, most engineers design agricultural technology with men in mind (Kawarazuka, 2018). This lack of understanding of gender preferences for particular features of technology may ultimately affect adoption and use rates by specific cohorts (Rola‐Rubzen et al., 2020). Furthermore, traditional efforts to encourage adoption have often failed to account for the particular limitations on women, such as interactions between gender and other social and economic factors (Garikipati, 2009).

Female agriculturalists are also likely to receive inadequate extension training and information (Doss & Morris, 2000); for example, training for agricultural machinery and technologies often target men (Rola‐Rubzen et al., 2020). Furthermore, extension agents are usually male and may not want to interact with women farmers due to personal beliefs or social or cultural taboos (Peterman et al., 2010). Notwithstanding these challenges, when given authority, opportunity and technology, women have the potential to have far-reaching positive impacts. For example, in Uttar Pradesh, India, Paris et al. (2008) illustrated how providing women with more authority over selecting new rice varieties resulted in their increased adoption and confidence in their opinions and decision-making abilities.

Similarly, irrigation technology has tremendous potential to improve livelihoods for all farmers when placed in the hands of women. However, subsidies for accessing and using pump sets are invariably tied to land ownership, thereby preferencing men. Lountain et al. (2023) found that preferences for features of irrigation pumps differed between men and women, with women farmers in West Bengal preferring to earn income from the pump device when not in use for their agriculture, an attribute not systematically selected by the males in the same survey (Lountain et al., 2023). The point is that women accessing technologies could (1) improve their standing generally and (2) give rise to different global outcomes than those that attend the continuation of male-dominated technological access. Again, this raises questions about how a multi-pronged approach might benefit women’s empowerment.

Women’s empowerment and policy convergence

The previous section highlighted specific interventions aimed at dealing with one component of empowerment (institutions that provide self-help and support for women) or tackling a single inhibitor to increased empowerment (improving access to land, water and technology). We also noted earlier the complexity of empowerment and its multiple facets and cited the review of national approaches that call for a convergence of efforts. Those calls appear premised on the notion that convergence of efforts to achieve simultaneous uplift in multiple domains gives rise to better outcomes than tackling domains separately.

Policy convergence emerged as a topic of discussion in the published literature in the 1980s and 1990s. Initially, scholars used the notion to query the processes that drove societies to become more alike, particularly in terms of structures, processes and understanding of the performance (Kerr, 1983). Bennett (1991) reduced the discussion on broader societal convergence by focusing specifically on what made public policies converge. He identifies five areas where policy could potentially converge. These comprise policy goals (‘a coming together of intent’); policy content (‘more formal manifestations of government policy’); policy instruments (‘the institutional tools available to administer policy’); policy outcomes (‘the results – positive or negative, effective or ineffective’); and policy style (‘the processes by which policy responses are formulated’; Bennett & Howlett, 1992, p. 21). Notwithstanding these distinctions, Bennett and Howlett (1992) also acknowledged that these are not always discrete, and overlaps are likely to emerge. They also note that policy convergence is a process rather than an outcome and subsequently outline key drivers that facilitate it.

Much of the strand of literature initiated by Bennett and Howlett (1992) and others are concerned with the tendency of different nations to move towards a common policy setting in fields as diverse as agricultural risk (Popp, 2021), land-use planning (Morphet & Clifford, 2014) and higher education (Veiga et al., 2019). However, in populous states with significant and, at times, autonomous bureaucracies, as is arguably the case in most states of India, policy convergence has assumed a different but related meaning. Here, the emphasis has been on the extent to which separate arms of government intervention work collaboratively towards a common goal or outcome. Examples include policy convergence around freshwater management in rural Jharkhand (Sanga & Ranjan, 2009) and watershed development programmes in Andhra Pradesh (Chandrappagari et al., 2015). In the latter case, the watershed development programmes became a vehicle for coalescing multiple programmes and activities to focus on a shared ambition of raising the well-being of the rural poor. Interestingly, an important component of this ‘convergence’ was centred on women establishing non-farm enterprises as part of SHGs and diverting efforts around national employment schemes to produce synergies in districts where watershed development had been prioritized (Chandrappagari et al., 2015).

The case for policy convergence around women’s empowerment is intuitively appealing but warrants further investigation. On the one hand, states with limited financial resources would appear well-served to coordinate and ‘converge’ efforts to meet the best collective outcome; after all, the capability of government agencies operating in the field is not infinite and harnessing those capabilities around a single ambition appears prima facie as a sensible solution. On the other hand, potentially non-trivial coordination costs and costs of change might act against convergence. In practice, convergence may be challenging as various stakeholders may have conflicting goals. Moreover, the different bureaucracies in states such as West Bengal hold quite different skill sets and bringing those to focus on a single, shared problem is no simple task. For example, water bureaucracies are usually dominated by engineers skilled in the physics of water distribution while the creators of SHGs rely heavily on the social dynamics and institutional architecture that operate within communities. Likewise, agricultural extension officers have specific expertise in agricultural technology and forestry agents hold alternative skills. The costs of gaining consensus from these groups of experts might be so high as to ultimately render the individual interventions suboptimal. Against that background, it would be useful to know which elements of empowerment share a positive correlation as this is likely to signal lower transaction costs to achieve convergence and also open the opportunity for positive feedback between elements.

Survey approach and analytical method

The reviewed literature suggests that both access to resources and social and institutional support are necessary for the empowerment of women farmers and that neither in isolation will necessarily bring success. Against that background, we sought to understand the situation of female farmers in West Bengal, India, and whether there is a correlation between some of these variables, such as access to resources, the strength of involvement and participation in community groups and other empowerment indicators.

The data reported in this paper were collected as part of a survey investigating how different farmers access irrigation pumps in West Bengal and their underlying preference for irrigation pumping technologies (Lountain et al., 2023). The survey included a subset of questions drawn from the Women’s Empowerment in Agriculture Index (WEAI) (Alkire, Meinzen-Dick, et al., 2013).

Development of the survey tool

The survey development is described in detail by Lountain et al. (2023). The main elements included discussions with experts in India to gain an understanding of the current situation for women in agriculture in West Bengal and how different policies may contribute to or hinder their empowerment. Three semi-structured focus groups then took place with male and female farmers in West Bengal in September 2019.

The existing literature on the measurement of empowerment was reviewed, and in light of the interviews and focus groups, a subset of questions was adapted from the WEAI (Alkire, Malapit, et al., 2013; Alkire, Meinzen-Dick, et al., 2013) and included in the survey. These questions asked respondents about their household, farm and community position, intending to understand individuals’ perceived empowerment. Questions regarding family members’ involvement in household and farm decision-making in areas such as income, use of time and productive resources were included. In addition, other questions probed the ease with which women contributed to public forums and their perceived decision-making power in the home, farm and community. More specifically, these data explored participants’ voice in the community, including whether the respondent felt that someone such as themselves could generally bring about change and if they were comfortable speaking up in public to help decide on infrastructure (such as roads, small wells, water pumps and water supplies) to be built in the community. A five-point Likert question format was used for these questions.

Survey contents and pre-testing

The final survey consisted of five main parts. Part A asked questions about the respondents’ demographic and socio-economic status, while part B explored participants’ influence in their community and farm and household decision-making processes. Parts C and D comprised questions about current irrigator practices and preferences for pumping technologies. Part E of the survey sought to gauge the degree to which participants understood the survey questions. The survey was initially pretested with 30 paper-based surveys administered to farmers during October and November 2020. The purpose of this was to trial the questions and response formats. The wording of some questions was modified, but the overall format remained similar after the pretesting.

Sample selection, piloting and survey administration

A local enumerator company was employed to lead survey recruitment and data collection. The company was selected based on its track record in the region and its expertise in recruiting respondents. The project investigators trained enumerators virtually to ensure they understood the research questions and that data were entered accurately. In addition, the survey was translated into the local language to minimize variation across enumerators. Quotas were set to ensure sample representation across key characteristics. Namely, all respondents needed to be involved in agriculture, and a sampling framework was used that drew on farm households from Cooch Behar, Bankura, Hooghly, Burdwan, North 24 Pargonas and South 24 Pargonas. These regions were chosen to provide insights across a range of agricultural settings. A total of 36 villages were selected (six from each region), and enumerators surveyed no more than 15 respondents per village to add further variability within the sample. The aim was to have 10 female respondents per village, but this could not be achieved in some instances. Enumerators used local records to identify and contact participants and, where necessary, applied snowballing techniques to reach the quota in each village. Given the focus on gender and women, female enumerators were specifically employed to obtain female responses.

Recruitment and survey administration occurred via telephone as travel restrictions had been applied as part of the COVID response. Enumerators read the survey aloud and entered responses into the online Sawtooth platform while the project investigators hosted and monitored the online data collection. The online survey was piloted initially with a sample of 20 respondents to test the survey and software coding. Preliminary models were estimated using the pilot data from part D of the survey to ensure the experimental design and survey instrument were operating correctly. The final data collection took place from late February to mid-March 2021.

Analytical method

Two broad forms of analysis were expected to follow data collection. First, given the administration of the survey to both men and women in different households, it was anticipated that comparisons between these two sets of respondents would offer useful insights into the status of women relative to men and thus provide some understanding of women’s empowerment. Second, it was expected that the data could be interrogated to test empirically the case for policy convergence. Here it was expected that ordered logistic regression could be employed to explore whether specific variables related to enhancing empowerment were positively correlated. We hypothesized that positively correlated parameters could be converged at a lower cost than others.

Ordered logistic regression (also called the ordered logit model or proportional odds model) can estimate relationships between an ordinal dependent variable and a set of independent variables. An ordinal variable is a variable that is categorical and ordered, for instance, ‘poor’, ‘good’ and ‘excellent’. This approach is relevant when there are more than two outcomes where the outcomes can be ordered.

In ordered logit, an underlying score is estimated as a linear function of the independent variables and a set of cut points (STATA, 2021). The probability of observing outcome i corresponds to the likelihood that the estimated linear function, plus random error, is within the range of the cut points estimated for the outcome:

$\mathrm{Pr}(\mathrm{outcome}=\mathit{i})=\mathrm{Pr}\left({\kappa }_i-1<{\beta }_1{x}_{1\mathit{j}}+{\beta }_2{x}_{2\mathit{j}}+\cdot \cdot \cdot +{\beta }_k{x}_{\mathit{kj}}+u_j\le {\kappa }_i\right)$

In ordered logit, u_j is assumed to follow a Gumbel distribution. In either case, we estimate the coefficients ${\beta }_1,{\beta }_2,\dots ,{\beta }_k$ together with the cut-points ${\kappa }_1,{\kappa }_2,\dots ,{{\kappa }_k}_{-1}$, where k is the number of possible outcomes. ${\kappa }_0$ is taken as −∞, and κ_k is taken as +∞. All of this is a direct generalization of the ordinary two-outcome logit model.

Results and discussion

Sample profile

The total sample comprised 534 responses, with 350 females and 184 males. The sample characteristics are shown in Table 1. The mean age of the sample was 41.8 years. Of the total sample, 96.6 of the respondents worked on a small farm (less than 2 ha). A total of 36.8% of respondents managed their own farm, while 8.8% were strictly tenant farmers. Respondents who both managed their own farm and leased land from others comprised 12.6% of the sample.

Table 1. Sample profile.

Description	Female percentage of sample (n = 350)^a	Male percentage of sample (n = 184)^a	Total percentage of sample (n = 534)^a
Total sample responses	65.5	34.4	100
Managed own farm	29.14	51.65	36.84
Tenant farmer	11.71	3.30	8.83
Managed own farm and tenant farmer	11.71	14.29	12.59
Worked on a small farm (less than 2 ha)	98	94.05	96.99
Have access to a tubewell (for drinking water)	12.32	12.15	12.26
Connected to the electricity grid	99.71	100	99.81

Note: ^aRespondents could select more than one response, that is, select all that apply.

We present notable descriptive responses by gender in Table 2, followed by a breakdown of how respondents accessed water for irrigation in Table 3. The majority (75.5%) of male respondents identified as the head of their household, with only 8% of females identifying as the household head. Similarly, 83.1% of females identified as the spouse of the household head, with only 1% of males considering themselves as the spouse of the household head.

Table 2. Notable descriptive responses by gender.

Description	Female percentage of sample (n = 350)^a	Male percentage of sample (n = 184)^a	Total percentage of sample (n = 534)^a
Household head	8.00	75.54	31.27
Spouse of household head	83.14	1.09	54.87
Married	95.71	87.43	92.87
Engaged in unpaid farm/household work	39.10	Nil
Operated a pump set(s)	33.70	91.20	57.19
Owned a pump set(s)	34.83	54.14	44.76
Member of a community groups	45.38	34.25	41.56

Table 3. Survey sample breakdown of how respondents access water for irrigation by gender.

Response	Female percentage of sample (n = 344)^a	Male percentage of sample (n = 181)^a	Total percentage of sample (n = 525)^a
Own pump set(s)	39.83	54.14	44.76
Jointly owned pump set(s)	5.81	4.42	5.33
Community owned pump set(s)	6.98	10.50	8.19
Hired pump set(s)	15.12	5.52	11.81
Buy water	29.36	24.31	27.62
Canal water	2.91	1.10	2.29

Note: ^aRespondents could select more than one response, that is, select all that apply.

Regarding technology, 91.2% of males operated an irrigation pump set, while only 33.7% of females did (the breakdown of pump types is presented in Table 3). This gender split may reflect the reasons explored above, including traditional views of men as farmers and women as farmers’ wives. As discussed, this perception often prevents women from consideration in the design of agricultural technology (Mukhopadhyay et al., 2023, p. 111). Likewise, extension training is often withheld from women, resulting in a ‘discriminatory distribution of knowledge’ (p. 65). Rola‐Rubzen et al. (2020) also note how in some societies, machines and equipment are seen as men’s domain and can often reflect a symbolic power, although arguably, access to machines goes beyond symbolism.

Similarly, while zero males reported that they participated in unpaid farm or household work, 39.1% of female respondents did. This also reflects findings from the literature, as many women in India are involved in farming, but much of their work is informal or unpaid (Kamdar & Das, 2021; Quisumbing et al., 2014). Moreover, women’s unpaid family duties also often take them away from work that generates an income (Huyer, 2016), further preventing them from improving their situations.

Community, farm and household decision-making

Table 4 provides insights into involvement in the community, farm and household decision-making, and who made these decisions. It focuses on the contrasting opinions of males and females in reference to their counterparts within the household and highlights some interesting contrasts between the responses of females and males surveyed in separate households. Table 4 does not include responses that did not fit this focus, that is, ‘male outside the household’. For example, when asked if they felt they could change things in the community where they live, 59% of female respondents answered ‘No, not at all’, compared with 41.8% of males. Likewise, when asked if they felt comfortable speaking up in public to help decide on community infrastructure, 40.1% of females answered ‘No, not at all comfortable’, compared with 21% of males choosing this response.

Table 4. Community, farm and household decision-making by gender.

Decision-making questions	Response	Female percentage of sample (n = 350)^a	Male percentage of sample (n = 184)^a	Total percentage of sample (n = 534)
Do you feel that a [man/woman] like yourself can generally change things in the community where you live if s/he wants to?	No, not at all	59.31	42.54	53.58
	Yes, but with great difficulty	29.80	42.54	34.15
	Yes, but with a little difficulty	9.46	14.36	11.13
	Yes, fairly easily	0.86	0.55	0.75
	Yes, very easily	0.57	Nil	0.38
Do you feel comfortable speaking up in public to help decide on infrastructure (like roads, small wells, water pumps, water supplies) to be built in your community?	No, not at all	40.11	20.99	33.58
	Yes, but with difficulty	40.97	45.86	42.64
	Yes, comfortable	18.91	33.15	23.77
Are you a member of any group or organization in your community?	Yes	45.38	34.25	41.56
Who in your household makes decisions related to productive resources (i.e., use of land, livestock, water, technology, etc.)?	Both self and spouse	42.77	22.65	35.86
	Male in household	37.86	40.33	38.71
	Female in household	6.94	0.55	4.75
Who in your household makes decisions related to farm income and spending?	Both self and spouse	48.27	20.44	38.71
	Male in household	35.26	39.23	36.62
	Female in household	6.36	1.10	4.55
Who in your household makes decisions related to household income and spending?	Both self and spouse	60.69	38.67	53.13
	Male in household	23.70	24.86	24.09
	Female in household	6.07	2.21	4.74
Who in your household makes decisions related to how your time is spent?	Both self and spouse	55.65	4.97	38.21
	Male in household	3.48	91.16	33.69
	Female in household	39.71	Nil	26.03

Note: ^aChi-squared tests comparing responses from women and men all report p ≤ 0.2.

However, when asked if they were members of a community organization or group, 45.3% of women answered ‘Yes’, compared with 34.3% of males. Arguably, this response reflects the high and targeted activity of SHGs in the sampled region (Khandker et al., 2020).

When asked about who makes decisions on the farm and in the household, responses by gender varied considerably. While a large portion of females answered, ‘Both self and spouse’ (42.8%) about decision-making around productive resources and money, males more often answered ‘Self’ (40.3%).

Ordered logistic regression model results

The data were further analysed using an ordered logistic regression model. To reiterate, the overarching purpose of this part of the analysis is to identify those elements of empowerment that are positively correlated, as these are presumed to be more conducive to a convergent policy approach.

For this portion of the analysis, the data were separated by gender, and only female survey responses were reported in detail (equivalent analyses for the male respondents are reported in Appendices A and B). The aim was to investigate relationships between empowerment indicators and specific dependent variables. Of the entire dataset, potential dependent variables for analysis were selected based on whether the findings were considered relevant from a policy perspective. Some variables had to be removed from the analysis because the dataset was too small to produce statistically significant findings. Subsequently, the relationship between each of these variables and two of the empowerment questions (dependent variables) from the survey was investigated with two-way frequency counts. In this case, the questions underpinning the two empowerment variables are: (1) ‘Do you feel that a [man/woman] like yourself can generally change things in the community where you live if s/he wants to?’ and (2) ‘Do you feel comfortable speaking up in public to help decide on infrastructure (like roads, small wells, water pumps, water supplies) to be built in your community?’, both adapted from the WEAI (Alkire, Malapit, et al., 2013; Alkire, Meinzen-Dick, et al., 2013). The variables tested included age, relationship to the head of household, marital status, principal occupation, present home occupancy status and current irrigation pump ownership and use.

Two-way frequency counts revealed which variables had relationships to the empowerment indicators. An ordered logistic regression model to test the significance of these relationships followed. Variables without relationships to empowerment indicators were omitted from further analysis. Variables (independent variables) with statistically significant relationships with the two empowerment variables are reported and discussed below. As noted in Table 3, 59.6% of females answered ‘No, not at all’ in response to the question related to instigating change, and 40.4% of females answered ‘No, not at all’ in response to the question about speaking publicly to influence infrastructure.

As also noted in Table 2, of the females surveyed, 34.8% reported that they accessed irrigation technology by owning a pump set, 33.7% operated a pump set and 45.3% were members of a community group. As access to technology and being a part of a collective institution can be used as proxy drivers of empowerment, an ordered logistic regression was used to test the correlation between pump operation and participation in community groups and the responses to the questions about perceived capacity to speak and influence communities.

An ordered logistic regression model (model 1) is presented in Table 5. In this model, the variables of owning and operating a pump set are combined because the dataset was too small to produce statistically significant findings separately. Model 1 indicates a positive and statistically significant (p < 0.05) correlation between women who own and operate pumps and women speaking up in public to help decide on infrastructure in their community. At the same time, this was not true for men (results available in Appendices A and B). Thus, while these relationships are not necessarily causal, it is likely that women who use pumps also feel more confident speaking up in public about them. Likewise, women in community groups have a positive and statistically significant (p < 0.01) correlation with feeling confident about speaking up in public to help decide on infrastructure in their community, but this was also not the case for men. Women who bought water for irrigation were also more likely to feel more comfortable speaking publicly about community infrastructure (p < 0.1).

Table 5. Ordered logistic regression: women speaking up about public infrastructure (model 1).

	Coefficient	SE	Z-statistic
Women own and operate pump set(s)	0.689**	(0.305)	2.46
Women who hire pump set(s)	−0.540	(0.329)	−1.64
Women who buy water	0.464*	(0.258)	1.80
Women who use canal water	−0.918	(0.720)	−1.27
Women part of community groups	0.861***	(0.214)	4.01
Cut points
${\kappa }_1$	0.086	(0.193)
${\kappa }_2$	2.11	(0.230)
Model statistics
LL	−341.57
Observations	343
LR Chi^2	35.17
Prob > Chi^2	<0.001
McFadden’s pseudo-R^2	0.049

Note: ***p < 0.01, **p < 0.05, *p < 0.1.

Table 6 reports another ordered logistic regression model (model 2) for perceptions of ability to instigate change in their community. The results presented in Table 6 are analogous to those reported in Table 4. The model indicates a positive and statistically significant relationship between women who own and operate a pump (p < 0.05) and an increased perception that they have the capacity to make a change. Similarly, a positive and significant relationship exists between women who are members of a community group (p < 0.01) and an increased perception of instigating change. Interestingly, neither of these relationships was true for the men surveyed (for the results, see Appendices A and B).

Table 6. Ordered logistic regression: women’s perceived ability of making changes in their community (model 2).

	Coefficient	SE	Z-statistic
Women own and operate pump set(s)	0.508	(0.316)	1.61
Women who hire pump set(s)	−0.371	(0.356)	−1.04
Women who buy water	0.108	(0.280)	0.39
Women who use canal water	−1.542	(1.089)	−1.41
Women part of community groups	1.142***	(0.230)	4.97
Cut points
${\kappa }_1$	0.939	(0.214)
${\kappa }_2$	2.79	(0.267)
${\kappa }_3$	4.97	(0.497)
${\kappa }_4$	5.89	(0.739)
Model statistics
LL	−314.90
Observations	343
LR Chi^2	34.46
Prob > Chi^2	<0.001
McFadden’s pseudo-R^2	0.053

Note: ***p < 0.01, **p < 0.05, *p < 0.1.

Discussion

In this project, we explored the complexity of women’s empowerment and how its achievement may be better pursued. Our contention was that those activities aimed at increasing empowerment that are positively correlated will be better candidates for a convergent approach relative to those that are not positively correlated. This also provided an opportunity to test the contention offered by NITI Aayog & Development Monitoring and Evaluation Office (2021) that convergence should be considered as a means of enhancing the effectiveness of policy interventions. The NITI Aayog & Development Monitoring and Evaluation Office (2021) base their recommendation for convergence on the view that synergies across domains can deliver economies of scale in empowerment. The reviewed literature suggests that both access to resources and social and institutional support are key drivers for the empowerment of women farmers and that neither in isolation will necessarily bring success. Against that background, we sought to better understand the situation of female farmers in West Bengal and how the minutia of women’s access to and use of irrigation pumping technologies and the detail of their participation in collective institutions related to their perceptions of empowerment.

The comparisons by gender (Tables 2 and 4) are consistent with well-documented disparities between men and women regarding a lack of female agency and empowerment in the region (World Bank et al., 2008). Strikingly, no men reported participating in unpaid work, while over a third of women did. In addition, women usually perceived decisions within their household and farm as being made jointly, while men mostly perceived themselves as the sole decision-maker. Regarding technology, over 90% of males operated an irrigation pump set, while only 33.7% of females did. This figure might appear high, given the earlier commentary around water being predominantly ‘men’s business’, but wider population adjustments with out-migration of men may account for this and signal an emerging trend. It is also worth noting that many subsidies related to pumping technologies are linked to land ownership, which thus favours men, regardless of out-migration. To the knowledge of the authors, modifying these arrangements to offer targeted support to women is not on the agenda but may be worth contemplating.

The strong participation of women in community groups clearly outstrips the participation of men. Rola‐Rubzen et al. (2020) highlight the critical role of empowerment through social networks in enabling inclusive technology in agriculture. They note that women’s social networks are crucial to encouraging technology adoption and, arguably, long-term use. The findings presented in Tables 5 and 6 support this. Of the female farmers in West Bengal surveyed, women who operated irrigation pumps and were members of community groups were more likely to feel confident speaking up in public about infrastructures such as roads, small wells, water pumps and water supplies (Table 5). In addition, these women were generally more confident that they could instigate change in their community (Table 6). These correlations support similar links in the literature between technology access and decision-making power (e.g. Agarwal, 2009; Finnis, 2017; Huda & Kaur, 2011; Kabeer, 2011; Kabeer & Matin, 2004; Paris et al., 2008; Sanyal, 2015). Importantly, in a policy context, this implies that providing women with access to technology and providing social support through SHGs will have the added spillover benefit of ensuring women’s preferences around local infrastructure and governance will be more accessible.

Based on our findings, a convergent policy approach that targets increased access to technology for women – particularly pumping technologies – and the development of capacity within collective institutions would appear to offer significant promise in terms of enhancing women’s empowerment in the study region. Our results show that while access to technology alone translated into feelings of general empowerment, participation in cooperative institutions showed a strong and positive relationship between women participating in public social behaviours and having greater feelings of empowerment overall. Thus, while resource access is one key for empowerment, other social supports combined with that resource can deliver better outcomes. The results suggest that these actions could potentially operate in tandem, and coordinating across them to support convergence around empowerment should be a serious consideration.

At the outset, we noted that there had been a call for increased convergence of policies to support women’s empowerment at the national level in India. However, this call was broad scale and offered relatively limited direction for convergence to proceed at a subnational scale. Our results draw on state-level information and provide some specific direction for convergence at that level. We argue that support for women to gain access to pumping technologies combined with an expansion of SHG activities has the potential to realize increased women’s empowerment in rural settings in West Bengal. This contrasts with the national recommendation by drawing attention to the potential role of access to pumping technologies in this state.

Conclusions

In this paper, we explored the complexity of women’s empowerment and how its achievement may result in greater success if pursued through a convergence of efforts by the government. The motivation for this was the national call for convergence across programmes aimed at enhancing the delivery of programmes to support women and children in India. Empowerment is clearly a complex goal, and policies targeting individual domains may not always be effective. However, especially in a populous country with many rural poor, government resources and capacity can be stretched, and there may be logic in taking a convergent approach to economize on delivery.

Using primary data from West Bengal, India, we investigated women’s access to irrigation pumping technologies and their relationship to a range of other social behaviours. We found that women’s access to technology and their belonging to a collective institution correlated with a greater likelihood of speaking out in public about community infrastructure decisions and perceptions of instigating change in their wider community. Therefore, our results support a convergence of policy efforts may be more beneficial and efficient than pursuing targets separately.

There are important caveats to this finding. The sample is restricted to a single state in India, albeit one of the most populous and agriculturally important. However, India has wide disparities and variabilities, and what works in one state may not apply to others. Care also needs to be taken around other generalizations – convergence may be appropriate with some programmes and initiatives and not others. There are also constraints around the focus of this project, as there are multiple technologies that could potentially impact agricultural productivity in the region beyond those related to irrigation pumping. The data are also limited to a subset of empowerment indicators, and the results may differ with a broader survey. Nonetheless, there is sufficient evidence within this work to buttress a more comprehensive discussion about convergent efforts to empower women at the subnational level.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Australian Centre for International Agricultural Research [LWR/2018/104].

Appendix A

Table A1. Ordered logistic regression: men speaking up about infrastructures.

	Coefficient	SE	Z-statistic
Men operate a pump	0.826	(0.618)	1.34
Men who hire a pump	−0.751	(0.908)	−0.83
Men who buy water	0.165	(0.660)	0.25
Men who use canal water	−1.006	(0.708)	−1.42
Men part of community groups	0.139	(0.298)	0.47
Cut points
$k_1$	−0.757	(0.608)
$k_2$	1.389	(0.615)
Model statistics
LL	−189.21
Observations	185
LR Chi^2	9.89
Prob > Chi^2	0.079
McFadden’s pseudo-R^2	0.026

Appendix B

Table B1. Ordered logistic regression: men’s perceived ability of making changes in their community.

	Coefficient	SE	Z-statistic
Men operate a pump	01.2248*	(0.650)	1.88
Men who hire a pump	−0.276	(0.928)	−0.30
Men who buy water	0.493	(0.698)	0.71
Men who use canal water	1.768	(1.314)	1.34
Men part of community groups	0.378	(0.301)	1.25
Cut points
$k_1$	0.714	(0.643)
$k_2$	2.822	(0.675)
$k_3$	6.373	(1.193)
Model statistics
LL	185.715
Observations	185
LR Chi^2	14.20
Prob > Chi^2	0.014
McFadden’s pseudo-R^2	0.037