‘Inclusivity’ in Welfare-Oriented Microfinance: Lessons From Kudumbashree in Kerala,India

Introduction

Microfinance is increasingly recognised as a significant component of poverty eradication strategies in developing nations, occupying a central position in contemporary development discourse (Dichter, 2007; Oommen, 2008). Research highlights its pivotal role in financial inclusion, poverty alleviation, and women’s empowerment, particularly among socially and economically disadvantaged groups (e.g., Abel, 2020; Ismail et al., 2024; Lacalle-Calderon et al., 2018; Maity, 2019). Case studies further suggest that adopting more people-centric approaches can improve the effectiveness of microfinance programmes in reducing social exclusion (e.g., Bahng, 2013; Maity, 2023). Nevertheless, concerns persist about microfinance’s ability to reach the most vulnerable populations (e.g., Coleman, 2006; Hulme & Mosley, 1996; Lonborg & Rasmussen, 2014; Navajas et al., 2000). This concern stems from the recognition that the impoverished, who constitute the primary beneficiaries of microfinance, are not a homogenous group. Rather, variations and fluctuations in household well-being arise due to the dynamic nature of poverty and disparities in the socio-economic capital possessed by different households (Montgomery, 1996; Rahman & Razzaque, 2000).

While empirical research on microfinance has expanded globally, there remains a lack of comprehensive, objective, and independent academic inquiries on why certain eligible households choose not to participate, particularly in diverse contexts. Often, the predominant focus on measuring the impact of microfinance programmes on clients has arguably side-lined research into the exploration of the process of participation (Mahmud, 2000). Since participation aligns with household livelihood strategies, especially for those facing poverty, understanding these mechanisms is crucial (Mahmud, 2000). Furthermore, the lack of representative quantitative data to examine exclusion at various stages of programme implementation, coupled with insufficient qualitative insights to investigate the reasons for this exclusion, has constrained the scope of available studies.

Recognising this gap in the literature and leveraging our prior research experience in microfinance, this paper aims to provide robust empirical evidence on the factors and reasons driving exclusion in microfinance programmes, while offering a strengthened conceptual framework to understand these dynamics. To this end, we examine non-participation among eligible households in Kudumbashree, a welfare-oriented microfinance initiative implemented by the Government of Kerala, India.

Specifically, this paper attempts to answer the following question: despite the financial and non-financial benefits provided by the programme and the government’s concerted efforts to encompass all poor households, why do certain women from rural poor households in Kerala abstain from participating in Kudumbashree, and what distinctive characteristics define this group? Although primarily empirical in nature, this paper also employs the lens of social exclusion to investigate if any particular population group—defined by social category, religion, and economic status in the Kerala context—faces institutional barriers in participating the programme. Additionally, it examines how attitudinal factors, particularly attitudes towards debt, influence participation decisions. By combining quantitative and qualitative insights, this study sheds light on the paradoxical processes of exclusion that could be embedded even in welfare-oriented microfinance programmes or similar social protection initiatives, specifically designed to address multiple forms of socio-economic exclusion, marginalisation, and vulnerabilities. On a broader conceptual level, we argue that institutional exclusion from social protection schemes—where participation is voluntary and conditional—can occur in implicit ways that are often overlooked.

Our choice of the social exclusion framework as the analytical lens of this paper is driven by its dynamic, multidimensional, and relational nature, which encompass a broader spectrum of social challenges (Berghman, 1995; Mathieson et al., 2008). While welfare policies and social protection programmes are widely recognised as crucial mechanisms for alleviating social exclusion (Kidd, 2014; Room, 1995), prior research has highlighted the risk of institutional exclusion potentially rooted within these policies. For instance, Daly (2006) and Kidd (2014) argue that flaws in policy design or implementation may unintentionally reinforce existing patterns of exclusion and discrimination, thereby perpetuating disadvantage. Consequently, in alignment with the European Commission (2008) and Levitas (2006), we acknowledge the importance of rigorous evaluation and monitoring of social inclusion policies. However, we believe that such evaluations should also incorporate ‘inclusiveness’ as a core dimension, for which the social exclusion framework can serve as a valuable analytical tool.

This study is particularly significant in the Indian context. Despite India being one of the world’s largest microfinance markets, limited academic research has examined the determinants and causes of non-participation in microfinance programmes. Most existing studies on the issue originate from Bangladesh or Africa (e.g., Evans et al., 1999; Islam, 2011; Karim & Osada, 1998; Lonborg & Rasmussen, 2014; Pearlman, 2014). Our approach diverges from these studies in several ways. First, unlike many studies that adopt either a quantitative or qualitative lens, providing only a partial understanding, ours integrates both quantitative and qualitative data, offering a more nuanced perspective on the phenomenon of non-participation. This dual approach enables us to discern whether specific communities experience exclusion from the programme and if the reasons for non-participation differ across population groups.

Second, existing literature has not explored how eligible participants’ debt attitudes influence their decision to join or leave a microfinance programme. In contrast, our study incorporates a debt attitude scale in quantitative modelling, recognising its importance given Kudumbashree’s primary role as a source of affordable and accessible credit.

Third, most prior studies focus on specific commercial microfinance programmes run by non-governmental organisations (NGOs) or financial institutions, thereby limiting their scope. Conversely, Kudumbashree’s broader coverage allows for a more comprehensive inquiry into the factors influencing non-participation. By identifying three distinct groups—current members, non-joiners, and dropouts—from a representative sample survey among eligible participants, this study offers a more systematic understanding of the determinants and causes of non-participation. Besides, as an independent study examining a government-sponsored microfinance programme and utilising data drawn from a representative survey collected as part of a larger research initiative, the response bias in our study would be considerably lower compared with studies focused on programmes implemented by specific organisations.

Therefore, Kudumbashree serves as a valuable case study for examining indirect exclusionary processes and unintentional institutional barriers within welfare-oriented microfinance initiatives, despite the many benefits they offer. Understanding these barriers is highly relevant for policy, not only to enhance the outcomes of the Kudumbashree programme in Kerala but also to inform the design and implementation of similar programmes in other developing countries, ensuring that all intended beneficiaries can access the full range of benefits.

The remainder of this paper is organised as follows: Section 2 furnishes the context and relevance of this study and provides a brief overview of the features of the Kudumbashree programme. Section 3 describes the data and methodology employed. Sections 4 and 5 sequentially present the insights derived from the analysis of quantitative and qualitative data. Section 6 synthesises the combined understanding emerging from quantitative and qualitative data, highlighting their policy and theoretical implications. Section 7 concludes the study.

Context and Relevance

Originating in Bangladesh during the 1970s, the concept of microfinance has undergone significant evolution, adapting to diverse contexts and cultures with the proliferation of varied approaches, structures, and operational modes. In India, the Self-Help Group-Bank Linkage Programme (SHG-BLP) is a prominent variant (Ghosh, 2013; Kalpana, 2017), although regional adaptations exist across the country. The Kudumbashree programme launched by the Kerala government in 1998 exemplifies this diversity. With 4.6 million women mobilised into 317 thousand neighbourhood groups (NHGs), ¹ this programme is considered the largest SHG-BLP in India, and arguably one of the world’s largest female networks. While broadly adhering to the prevalent SHG-BLP model, this state-sponsored, welfare-oriented, women-centric, poverty alleviation programme, grounded in the concept of microfinance, distinguishes itself with its well-defined decentralised structure ² and ‘credit-plus’ approach ³ (Oommen, 2008; Williams et al., 2011). ⁴ With its distinctive features, Kudumbashree has attracted significant academic interest, particularly in relation to gender dynamics (e.g., Agarwal, 2020; Devika & Thampi, 2007; Rajagopal, 2020). However, an underexplored area pertains to the factors influencing enrolment and dropout from the programme.

Motivations for understanding exclusion from microfinance programmes vary based on their goals, structure, and operating context. In commercial microfinance programmes, enhancing participation and minimising dropouts is crucial not only for the well-being of beneficiaries but also for the financial sustainability of the programme itself (Bardsley et al., 2015; Karim & Osada, 1998). Conversely, in non-profit programmes like Kudumbashree, specifically designed for the socio-economic upliftment of vulnerable households, non-participation implies that eligible beneficiaries are excluded from crucial welfare benefits. ⁵ Moreover, given the diverse models, and operational modes adopted by different microfinance programmes, the elements and processes of exclusion embedded within each may vary. Therefore, it is crucial to study the exclusionary processes across various programmes, each operating within distinct structures, objectives, and contexts.

Accordingly, Kudumbashree serves as a valuable case study due to its three distinct features. First, while previous studies highlight high interest rates in microfinance programmes as a barrier to participation (Panda, 2012; Rahman et al., 2014), Kudumbashree, as a government-subsidised programme, offers loans at lower rates than commercial banks. Second, research supports the effectiveness of a ‘credit-plus’ approach, in enhancing participation and retention in microfinance programmes (Muneer, 2020; Zaman, 1996). Kudumbashree, functioning as a welfare-oriented and poverty eradication programme, offers a myriad of non-financial services to its members. Third, unlike many microfinance initiatives, Kudumbashree explicitly aims for universal coverage of all economically vulnerable households (Section 7.13 of Kudumbashree bylaw, 2021), with the government undertaking targeted campaigns encouraging enrolment. To support this goal, Kudumbashree’s bylaw (2021) uniquely empowers NHGs to waive institutional requirements—such as regular thrift payments and mandatory meeting attendance—based on members’ specific financial and personal circumstances, thereby striving to enhance the programme’s accessibility for everyone.

However, despite all these concerted efforts, the sample survey conducted among rural poor households in Kerala in 2023, from which the data for this paper were drawn, revealed that 30 percent of these households remained excluded from the programme; some have never joined, and others have dropped out. Therefore, an inquiry into why certain eligible impoverished households opt not to participate in Kudumbashree can offer valuable insights into how, even within an almost economically homogeneous group, some households become disadvantaged in their ability to participate in a government-designed programme specifically aimed at providing benefits and addressing the multiple vulnerabilities faced by this group.

In this regard, the social exclusion framework proves useful, as it broadens the focus from individual and household level factors to encompass community-level dynamics and social relations (Atkinson, 1998; Daly, 2006). While no uniform definition for the concept of social exclusion exists—since it varies according to national and ideological contexts (Silver, 1994)—it is generally understood as the process through which individuals or population groups become either wholly or partially detached from the organisations and communities of which the society is composed and from the associated rights and obligations (Kidd, 2014; Room, 1995). Extending beyond the poverty-driven focus on the lack of disposable income or material resources, social exclusion incorporates a broader array of dimensions, including dysfunctions or imperfections within societal systems that disadvantage certain groups more than others (Barnes, 2019; Berghman, 1995).

However, Saith (2001), cautions against the blind application of conceptual definitions or measurement frameworks of social exclusion developed in the context of advanced economies to the realities of developing countries. Given the limited presence of a fully formed welfare state in many developing regions, Saith (2001) suggests an alternative approach; examining social exclusion within the context of social security schemes that have evolved in response to specific local circumstances and unique challenges in these settings. Thus, by examining inequalities in the risk of exclusion and distinguishing population groups based on the cumulative severity of the challenges they face (Paugam, 1995), social exclusion framework can aid in assessing whether a programme’s design and implementation intentionally or unintentionally exclude any specific population segments from its scope. It also facilitates an understanding of the dynamics and processes behind such exclusion.

Finally, previous studies have identified several exclusionary elements and mechanisms within the design and functioning of microfinance programmes that may exclude vulnerable communities, particularly people with disabilities, from participating (Beisland & Mersland, 2014; Bwire et al., 2009; Cramm & Finkenflügel, 2008; Hulme, 1999; Sarker, 2024). Building on these insights and considering the unique characteristics of the microfinance programme under study, this paper seeks to develop a broader framework for understanding the exclusionary processes within social protection schemes that are designed for voluntary participation but require beneficiaries to meet certain conditions.

Data and Methodology

The data for this paper were drawn from a household survey conducted as part of a larger project investigating indebtedness patterns among rural poor households in Kerala. To ensure representativeness, we identified 765 rural poor households ⁶ using a multistage cluster sampling technique ⁷ and successfully completed 729 interviews. Data collection took place between November 2022 and May 2023, carried out by 10 professionally trained investigators with prior experience in household surveys. A dedicated section of the questionnaire explored experiences of participation and non-participation in Kudumbashree. Following a thorough review for missing data and inconsistencies, a final sample of 678 households was selected for analysis.

As Kudumbashree is a programme designed to target households, our survey classified households into three categories based on their participation status; those currently participating, those that participated earlier but later dropped out, and those that have never participated (Table 1).

OPEN IN VIEWER

Table 1.

Distribution of Sample Households Based on Their Participation Status in Kudumbashree.

Participation Status	Frequency	Percentage
Currently participating	476	70.2
Dropouts	123	18.1
Never participated	79	11.7
Total	678	100.0

Information regarding households currently engaged in the programme or those that had previously participated but later discontinued was obtained directly from women who were either active or former participants. For households that had never participated, data were collected from the female member with the highest financial decision-making authority within the household. Our study combined quantitative and qualitative data from the same household survey. Given the multidimensional nature of social exclusion, this mixed-methods approach serves as a powerful tool, enabling both an examination of the characteristics of the excluded populations and an exploration of the processes driving their exclusion (Mathieson et al., 2008; Moulaert, 1995).

The quantitative component employed two separate probit models to identify measurable characteristics of women or households that influence the likelihood of participation and dropout. This analysis aimed to ascertain if there is a systematic exclusion of any specific population group—defined by social category, religion, and economic status—from the programme. The quantitative analysis also examined if the debt attitude and debt literacy of the respondents significantly influence their participation in microfinance. To this end, we adapted the Debt Attitude Scale developed by Lea et al. (1995) and the Debt Literacy Scale introduced by Disney and Gathergood (2011) to suit the local context of Kerala (see Appendix A). Using survey data, we conducted a confirmatory factor analysis to construct a variable reflecting debt attitude (see Appendix B). The construct meets all established reliability and validity thresholds as per Hair et al. (2019) and MacKenzie et al. (2011).

The qualitative component explored the reasons behind non-participation. In the survey questionnaire, respondents who had never participated or dropped out of the programme were asked to elaborate on their reasons for non-enrolment or discontinuation. Using NVivo, we systematically categorised these responses. From the identified patterns, we also sought to determine if any specific reason predominantly contributed to the non-participation of particular population groups. Additionally, we conducted 27 focus group discussions with current Kudumbashree members and entrepreneurs, alongside in-depth interviews with grassroots-level Kudumbashree leaders in the surveyed panchayats, as well as district- and state-level programme officials. Where relevant, we supplemented our analysis with insights gathered from these discussions.

Quantitative Analysis

We employed two separate probit models to analyse the factors influencing non-participation in Kudumbashree. The first model assessed the probability of a household joining the programme. For those who had joined, the second model analysed the factors affecting dropout from the programme. To account for potential selection bias in the second stage, we initially considered the Heckman probit model. However, the results indicated no significant selection bias, leading us to opt for two separate probit models.

The independent variables in the models were classified into two categories: individual-level characteristics of the respondents and household-level attributes. Descriptive statistics for these variables are reported in Table 2. Table 3 presents the results from the probit models. Given that a respondent’s current age may not reflect the age at which they left the programme, we excluded age as a variable in the second model. Instead, recognising the potential influence of entrepreneurial engagement on dropout probability, we introduced an enterprise variable in the second model.

OPEN IN VIEWER

Table 2.

Descriptive Statistics of Independent Variables.

Measure	Description	Item	Obs	Mean	Std. Dev	Min	Max
Individual-level characteristics
Age			678	50.11	13.01	20	85
Employment		Self-employed	678	0.02	-	0	1
		Daily wage labour	678	0.42	-	0	1
		Job with stable monthly income	678	0.05	-	0	1
		No income earning activity*	678	0.51	-	0	1
Income	Income of the respondent in a normal month (INR)		678	2172.91	3068.15	0	20,000
Education	Educational levels were categorised into seven distinct tiers and assigned numerical values ranging from 1 to 7, with lower values representing lower educational attainment and higher values denoting higher levels of education		678	3.27	1.22	1	7
Marital status		Married*	678	0.77	-	0	1
		Unmarried	678	0.03	-	0	1
		Widow/divorced	678	0.20	-	0	1
Financial decision-making power	Financial decision-making power enjoyed by the respondent within the household		678	1.48	0.50	1	2
Enterprise	Whether the respondent has ever engaged in an enterprise under Kudumbashree		678	0.23		0	1
Attitude to debt	Measured by debt attitude scale (Appendix A)		678	0.00	1.00	−2.04	2.31
Debt literacy	Measured by counting the correct answers given by the respondent to the three questions provided in Appendix A		678	1.05	0.94	0	3
Household-level characteristics
Religion		Hindu*	678	0.79	-	0	1
		Christian	678	0.11	-	0	1
		Muslim	678	0.10	-	0	1
Social category	Population in India is officially classified into four groups based on their social positioning considering historical, socio-economic and educational factors; scheduled castes (SCs), scheduled tribes (STs), other backward castes (OBCs) and general. SCs and STs are constitutionally recognised as the most disadvantaged social groups, who have historically faced severe social discrimination and oppression. The general category includes the social groups that have historically been considered the most privileged in terms of opportunities, wealth and education	SC	678	0.20	-	0	1
		ST	678	0.14	-	0	1
		OBC*	678	0.50	-	0	1
		General	678	0.16	-	0	1
Economic category	Captures the economic differentiation within low-income groups based on the colour of the PDS card. In Kerala although both yellow and pink cardholders are considered as low-income households, those holding yellow cards are considered as extremely poor	Yellow PDS card	678	0.28	-	0	1
		Pink PDS card*	678	0.72	-	0	1
HH Educational attainment	The highest level of education attained by any member of the household. Educational levels were categorised into seven distinct tiers and assigned numerical values ranging from 1 to 7, with lower values representing lower educational attainment and higher values denoting higher levels of education	-	678	4.56	1.32	1	7
HH Employment ratio	The proportion of employed household members to the total household size	-	678	0.46	-	0	1
HH Income	Total income of the household in a normal month (INR)	-	678	15,920.33	10,441.90	200	62,110
HH Stable income count	Number of household members with stable jobs and regular income	-	678	0.17	0.45	0	3
Housing tenure		Own property with title deed	678	0.81	-	0	1
		Do not own property with title deed*	678	0.19	-	0	1

OPEN IN VIEWER

Table 3.

Probit Models Exploring Non-participation in the Kudumbashree Programme.

Model	I		II
Dependent Variable	Probability of Joining (Current Participants or dropouts = 1; Never participated = 0) (N = 678)		Probability of Dropping out (Dropouts = 1; Current participants = 0) (N = 599)
Self-employed	1.853	(1.749)	−0.828	(1.477)
Daily wage labour	2.194	(1.676)	−0.586	(1.393)
Job with stable monthly income	1.821	(1.875)	−0.862	(1.563)
Ln income	−0.221	(0.203)	0.019	(0.169)
Education	0.086	(0.083)	−0.054	(0.070)
Unmarried	0.184	(0.448)	−0.178	(0.388)
Widow/divorced	0.053	(0.188)	−0.117	(0.178)
Financial decision-making power	0.169	(0.155)	0.236	(0.139)*
Attitude to debt	0.111	(0.069)	−0.284	(0.067)***
Debt literacy	0.073	(0.084)	−0.086	(0.076)
Christian	−0.513	(0.258)**	0.549	(0.242)**
Muslim	−0.161	(0.259)	0.675	(0.227)***
SC	0.219	(0.226)	0.300	(0.182)*
ST	−0.639	(0.261)**	0.599	(0.234)**
General	−0.138	(0.239)	0.231	(0.218)
Yellow PDS card	−0.216	(0.190)	0.289	(0.169)*
HH educational attainment	0.011	(0.067)	−0.138	(0.064)**
HH employment ratio	0.105	(0.362)	0.206	(0.334)
Ln HH income	0.113	(0.111)	−0.037	(0.107)
HH stable income count	0.218	(0.229)	0.004	(0.169)
Own property with title deed	0.205	(0.172)	0.201	(0.174)
Enterprise			−0.299	(0.160)*
_Cons	0.487	(1.155)	0.358	(0.893)
LR chi squared		65.80***		94.79***

Note. Standard errors in parentheses, ***p < .01; **p < .05; *p < .1.

As mentioned earlier, to investigate non-participation through the framework of social exclusion, we examined whether certain population groups among eligible beneficiaries—defined by social group, religion, and economic status—exhibit a higher likelihood of either non-enrolment or dropout from the programme. The results suggest that ST women, considered as one of the most underprivileged social groups in Kerala, face exclusion at both the entry and operational stages of the programme, as indicated by their higher probability of non-enrolment and dropping out.

Religion also appears to significantly influence participation. Women from the Christian community are less likely to join the programme compared with their Hindu counterparts. Additionally, the probability of dropout is higher among women from both Christian and Muslim communities than among Hindus. Conversely, economic status, measured by the PDS card colour, seems to have a more limited impact on participation.

Furthermore, respondents’ debt attitudes significantly influence the likelihood of dropout, with those holding more favourable views towards debt being less likely to discontinue participation. To explore whether the relationship between debt attitudes and dropout probability varies across different population groups—specifically, whether debt attitudes are shaped by distinct perceptions, such as confidence or fear regarding repayment ability held by different socio-economic and religious groups—we incorporated interaction terms between debt attitudes and these categories (Table 4). The results indicate that adding these interaction terms does not substantially alter the outcomes. Specifically, no significant differences were found in the influence of debt attitudes on dropout probability across various socio-economic and religious groups, suggesting that debt attitudes have a minimal influence in driving variations in dropout rates among these groups.

OPEN IN VIEWER

Table 4.

Probit Model Exploring Dropouts After Incorporating Interaction Terms.

Dependent Variable	Probability of Dropping out (Dropouts = 1; Current participants = 0) (N = 599)
Self-employed	−0.786	(1.485)
Daily wage labour	−0.616	(1.399)
Job with stable monthly income	−0.898	(1.566)
Ln income	0.023	(0.170)
Education	−0.054	(0.070)
Unmarried	−0.112	(0.388)
Widow/divorced	−0.100	(0.181)
Financial decision-making power	−0.254	(0.142)*
Attitude to debt	−0.279	(0.119)**
Debt literacy	−0.095	(0.077)
Christian	0.572	(0.246)**
Muslim	0.756	(0.245)***
SC	0.292	(0.184)
ST	0.540	(0.237)**
General	0.207	(0.227)
Yellow PDS card	0.325	(0.170)*
HH educational attainment	−0.135	(0.065)**
HH employment ratio	0.171	(0.336)
Ln HH income	−0.028	(0.107)
HH stable income count	0.037	(0.171)
Own property with title deed	0.187	(0.177)
Enterprise	−0.296	(0.162)*
Christian * attitude to debt	0.000	(0.237)
Muslim * attitude to debt	0.124	(0.228)
SC * attitude to debt	−0.069	(0.185)
ST * attitude to debt	−0.433	(0.245)*
General * attitude to debt	−0.137	(0.225)
Yellow PDS card * attitude to debt	0.244	(0.172)
_Cons	0.292	(0.898)
LR chi squared		99.40***

Finally, a negative relationship was observed between the highest educational level within a household, and the probability of women dropping out. This may be because higher education equips family members to recognise and value the benefits of participation, making them more likely to encourage women in their households to remain in the programme.

To verify the robustness of these results, we employed a multinomial logistic regression model with participation status as the dependent variable (results are presented in Appendix C). The dependent variable comprised three outcomes: current participants, non-joiners, and dropouts. To explore potential exclusion from Kudumbashree, the currently participating group was used as the reference category. By calculating the relative risk ratio (RRR), we assessed how the likelihood of being in the non-joiners or dropout groups changes relative to being in the currently participating group, given a one-unit change in the explanatory variables. The results confirmed the relationships observed in the probit models, while also underscoring the significant role of favourable debt attitudes in encouraging continued participation in microfinance programmes. Specifically, individuals with more favourable debt attitudes are less likely to be in the non-joiners or dropout groups compared with the current participants.

Qualitative Insights

The objective of the qualitative data analysis was to elucidate the process of exclusion by examining the reasons underlying certain women’s decisions to abstain from Kudumbashree, despite its provision of numerous welfare benefits. Applying the social exclusion lens, the analysis focused on identifying whether specific reasons predominantly influenced the non-participation of particular population groups.

As previously mentioned, survey respondents who were not currently participating in the programme were asked to explain their reasons for non-enrolment or dropping out. These reasons were collected through open-ended questions in the survey questionnaire.

The reported reasons were analysed and coded by the first author of this paper using NVivo to identify recurring patterns. The analysis followed the six-step guide proposed by Braun and Clarke (2006). Initially, each reason was openly coded, resulting in 49 codes for non-enrolment and 104 codes for dropout. These codes were then refined and grouped into broader categories, leading to the identification of four meta-themes for non-enrolment and three meta-themes for dropout, as illustrated in Table 5 and Figure 1 respectively.

OPEN IN VIEWER

Table 5.

Cited Reasons for Non-enrolment in Kudumbashree.

Category	Description	Causes	Explanation
Self exclusion (39)	Self exclusion driven by non-programme related barriersThe respondent herself decided not to join the programme due to personal reasons	Motivational barriers (24)	• Respondent lacked the confidence to go outside, visit banks or take up group responsibilities
			• The respondent lacked interest in participating in public events or attending meetings on a regular basis
		Personal circumstances reflecting limited individual-level capabilities (15) a Memberships in other SHGs (2)	Respondent was not interested in joining Kudumbashree due to factors such as age, health issues, care responsibilities, time constraints arising from family responsibilities or other job commitments
Explicit institutional exclusion (11)	The respondent wanted to join, but could not join due to certain institutional constraints	Absence of NHGs (10)	• No NHGs nearby
			• Existing NHGs are full to capacity b
			• Don’t have sufficient members to form a new NHG c
		Group exclusion (3)	Members of nearby existing NHGs did not allow the respondent to join the group presuming that she might not have the financial means to regularly contribute to thrift
Implicit institutional exclusion (27)	Self exclusion driven by programme related constraints. The respondent herself decided not to join the programme due to certain institutional requirements	Obligation to pay weekly thrift (22)	Respondent did not have sufficient financial means due to her physical inability to engage in work, high health expenditure within the family, other debt obligations or structural factors such as lack of job availability and the absence of a stable job or regular income
		Requirement to attend weekly meetings (11)	Respondent was unable to attend meetings due to health issues or time constraints stemming from care responsibilities, household chores or other job commitments
Family resistance (4)	The respondent wanted to join the programme, but could not join since the family members did not approve		• Family did not want the respondent to attend public events
			• Family did not want to increase the debt burden
			• Family did not want to pay regular thrift
			• Family perceived Kudumbashree as a ‘gossip group’ for women

OPEN IN VIEWER

Figure 1.

Types of discontinuation from the Kudumbashree programme.

Previous research conducted in non-Indian contexts has sought to classify the reasons for non-enrolment in microfinance programmes into programme related barriers and client related barriers (e.g., Evans et al., 1995, 1999; Hulme, 1999). However, based on the pattern emerged from our data, we recognised that simply categorising the reasons for non-joining into programme-related barriers and client-related barriers, or self-exclusion and institutional exclusion, would not comprehensively capture the complete narrative of exclusion. There were instances where the decision for self-exclusion stemmed from the perceived inability to fulfil the programme-related requirements, amounting to an indirect form of institutional exclusion as articulated by a 60-year-old respondent belonging to the ST category.

There is enough work at home. Will not be able to attend meetings amidst all these. And don’t I have to pay thrift every week? From where I will give it when there is no job?

Consequently, in line with Wright (1997), we recognised that distinction between ‘self’ and ‘organisation-driven’ exclusion is largely arbitrary, as the programme designs at times force the potential beneficiaries, particularly coming from socially and economically disadvantaged backgrounds, to ‘self-exclude’. Therefore, we introduced a third category to account for responses indicating self-exclusion driven by programme-related constraints (Table 5).

Table 5 reveals that women who had never participated in the programme mostly attributed their non-engagement to self-exclusion driven by factors unrelated to the programme, chiefly tied to their lack of motivation. Although instances of direct institutional exclusion appeared relatively fewer, programme-related factors frequently influenced the decision to opt for self-exclusion. By aggregating these two categories—explicit and implicit institutional exclusion—the number of individuals attributing their exclusion to institutional factors is almost equal to those citing self-exclusion driven by non-programme-related factors. From a broader perspective of social exclusion, it is possible that motivational barriers, such as lack of self-confidence or hesitancy to attend public events, are shaped by the social capital possessed by each population group, their environments, and their upbringing (de Haan, 2000). However, a thorough investigation into these aspects is beyond the scope of this study.

Figure 1 illustrates three distinct pathways through which participants disengage from the programme: dismissal of an NHG by authorities, the voluntary dissolution of an NHG by its members, and individual beneficiaries withdrawing their membership. Our survey revealed a preponderance of dropouts falling into the third category. Consequently, we further explored the reasons driving their decision to exit the programme (Table 6). The two most frequently cited reasons were the respondents’ limited capabilities to engage in the programme, due to their personal circumstances, and the institutional requirements—such as the obligation to make regular thrift payments or consistently attend meetings. Additionally, it was observed that the intra-group dynamics, often considered a defining feature of microfinance programmes, occasionally forced participants to withdraw their membership.

OPEN IN VIEWER

Table 6.

Cited Reasons for Withdrawing Membership From Kudumbashree.

Categories a	Description	Causes
Limited individual-level capability (55)	Had to leave the programme due to personal circumstances	Age, health issues, care responsibilities, difficulty in loan repayments, migration, pregnancy, time constraints, other job commitments, financial difficultiesetc.
Institutional requirements (39)	Had to leave the programme due to the respondent’s inability to fulfil institutional requirements	• Requirement to attend weekly meetings
		• Obligation to pay weekly thrift
		• Compulsion to attend programmes or events organised under Kudumbashree
Group pressure (16)	Had to leave the programme either at the compulsion of other group members or due to the respondent’s inability to withstand harassment or pressure from fellow group members	• Respondent’s inability to attend regular meetings due to health issues or family responsibilities
		• Respondent’s financial incapacity to make regular thrift payments due to unavailability of jobs, debt related obligations or health issues preventing engagement in work
Intra-group conflicts (12)	Opted to voluntarily withdraw from the programme due to issues with fellow group members	Referred contributory factors were discrimination, cheating, lack of mutual trust, financial fraud, favouritism, and free riding
Loss of motivation (6)	Left the programme due to a reduced motivation to participate	• Respondent lost interest in continuing with the programme
		• Respondent experienced boredom within the group as there was no one in her age group within the NHG.
Family resistance (4)	Had to withdraw from the programme due to family resistance

Consistent with the findings of previous studies (e.g., Bardsley et al., 2015; Wright, 1997), our analysis reveals the complexity and multi-dimensional nature of factors contributing to dropout from a microfinance programme. The decision to withdraw often cannot be attributed to a single event or cause; rather, it is the culmination of a series of events. In some instances, multiple causes accrue hierarchically, with one event triggering another. At other times, several events may occur concurrently. In such scenarios, identifying the immediate reason for dropping out becomes challenging as it is intertwined with other contributory events. This complexity makes it difficult to pinpoint a single root cause and formulate targeted interventions, as exemplified in the words of a 55-year-old daily wage labourer from Central Kerala.

In our NHG, I have the worst financial situation. When there was no work I couldn’t manage to pay the thrift regularly. So, everyone together started blaming me in meetings. Kudumbashree’s policy is that it is not compulsory for those who have no money to pay thrift. But those in the NHG won’t agree, right? Thus, when I reached a position that I couldn’t endure that torture anymore, I quit.

In this example, the apparent reason behind the respondent’s decision appears to be the pressure from the NHG. However, there were underlying factors that had driven her to this decision. She had to endure the group criticisms due to the non-repayment of thrift, an institutional requirement under the programme. She was unable to fulfil this requirement due to her already precarious financial position, compounded by lack of regular job availability or a fixed income flow—issues commonly faced by rural poor households in developing countries. Therefore, in this case, the respondent’s limited capability, as reflected by her financial difficulties, hindered her ability to meet institutional requirements, resulting in group pressure, ultimately influencing her decision to drop out from the programme. Aligning with the views of Sen (2000), such forms of self-withdrawal, driven by the ‘public shame’ tied to limited capabilities, represent an extreme form of social exclusion, even though it may not always be explicitly recognised as institutional exclusion.

We further aimed to delineate the predominant reasons contributing to the non-participation of specific population groups, focusing primarily on discerning whether institutional barriers or requirements systematically exclude particular population segments, categorised by religious and socio-economic categories (Tables 7 and 8).

OPEN IN VIEWER

Table 7.

Population Group-wise Reasons for Non-enrolment in Kudumbashree (In Percent).

Reasons for Non-enrolment	Religion			Social Category				Economic Category
	Hindu	Christian	Muslim	SC	ST	OBC	General	Very Poor	Poor
Explicit Institutional Exclusion	16.1	7.1	11.1	-	26.1	12.5	5.9	14.3	13.6
Absence of NHGs	14.3	7.1	11.1	-	21.7	12.5	5.9	11.4	13.6
Group exclusion	3.6	-	11.1	-	4.3	6.3	-	2.9	4.5
Implicit institutional exclusion	35.7	28.6	33.3	42.9	39.1	40.6	11.8	37.1	31.8
Obligation to pay weekly thrift	32.1	14.3	22.2	42.9	34.8	31.3	5.9	34.3	22.7
Requirement to attend weekly meetings	12.5	14.3	22.2	14.3	13.0	18.8	5.9	14.3	13.6
Self-exclusion	46.4	64.3	44.4	71.4	34.8	40.6	76.5	45.7	52.3
Membership in other SHGs	1.8	7.1	-	-	4.3	-	5.9	2.9	2.3
Motivational barriers	33.9	28.6	11.1	57.1	26.1	18.8	47.1	34.3	27.3
Personal circumstances reflecting limited individual-level capabilities	14.3	28.6	33.3	28.6	4.3	21.9	29.4	11.4	25.0
Family resistance	5.4	-	11.1	-	-	9.4	5.9	2.9	6.8
Base (Number)	56	14	9	7	23	32	17	35	44

OPEN IN VIEWER

Table 8.

Population Group-wise Reasons for Discontinuing From Kudumbashree (In Percent).

Reasons for Dropout	Religion			Social Category				Economic Category
	Hindu	Christian	Muslim	SC	ST	OBC	General	Very Poor	Poor
Dismissal of NHG	13.8	6.3	0.0	21.7	11.1	7.8	4.5	12.8	9.2
Dissolution of NHG	13.8	18.8	15.0	8.7	25.9	9.8	18.2	14.9	14.5
Individual client dropout	72.4	75.0	85.0	69.6	63.0	82.4	77.3	72.3	76.3
Institutional requirements	32.2	25.0	35.0	21.7	37.0	29.4	40.9	31.9	31.6
Compulsion to attend programmes	3.4	0.0	5.0	0.0	3.7	3.9	4.5	4.3	2.6
Obligation to pay weekly thrift	21.8	12.5	15.0	17.4	33.3	13.7	18.2	23.4	17.1
Requirement to attend weekly meetings	12.6	12.5	15.0	17.4	3.7	13.7	18.2	12.8	13.2
Group pressure	13.8	6.3	15.0	4.3	18.5	17.6	4.5	12.8	13.2
Financial incapacity to pay regular thrift	10.3	0.0	5.0	0.0	18.5	7.8	4.5	10.6	6.6
Inability to attend regular meetings	1.1	0.0	10.0	0.0	0.0	5.9	0.0	0.0	3.9
Loan related	2.3	6.3	0.0	4.3	0.0	3.9	0.0	2.1	2.6
Intra-group conflicts	11.5	0.0	10.0	8.7	7.4	13.7	4.5	14.9	6.6
Limited individual-level capability	41.4	56.3	45.0	47.8	25.9	47.1	54.5	38.3	47.4
Loss of motivation	3.4	12.5	5.0	0.0	7.4	7.8	0.0	4.3	5.3
Family resistance	2.3	6.3	5.0	4.3	3.7	2.0	4.5	4.3	2.6
Base (Number)	87	16	20	23	27	51	22	47	76

While the quantitative analysis suggests a lower likelihood of Christian communities enrolling in the programme, Table 7 does not highlight any disproportionate institutional barriers faced by this group. Rather, members of the Christian community are more prone to voluntary self-exclusion due to personal circumstances or a lack of interest and motivation. In terms of dropout, the quantitative data indicate that women from Christian or Muslim communities are more likely to leave the programme compared with those from the Hindu religion. However, Table 8 reveals that these two groups face different reasons for discontinuing participation. Among the three religious groups, the Christian community had the lowest percentage of women reporting institutional requirements influencing their decision to drop out. While the difference with the Hindu community is not substantial, a slightly higher proportion of Muslim respondents mentioned institutional requirements as a reason for their withdrawal. Additionally, the data indicate that individuals from the Hindu or Muslim faiths are more susceptible to group pressure or intra-group conflicts compared with their Christian counterparts. This suggests that, relative to the Christian community, a higher proportion of Hindus and Muslims are likely to experience exclusion from the programme due to institutional barriers.

Furthermore, compared with the General category, vulnerable social groups including SC, ST or OBC communities were more likely to report explicit or implicit institutional barriers as reasons for non-enrolment. Although the difference is not substantially wider, a higher proportion of respondents holding yellow PDS cards (indicating the very poor category) mentioned explicit and implicit institutional exclusion compared with those with pink PDS cards. These observations suggest that socially and economically vulnerable groups are more likely to encounter institutional barriers at the entry level, or they may exhibit greater hesitancy to join the programme due to the associated institutional requirements.

Compared with those from the OBC or General categories, a higher proportion of individuals from the SC or ST communities reported dismissal of NHGs by officials as a reason for discontinuing their participation. However, there is no apparent indication of other institutional barriers disproportionately driving the dropout decisions of these most vulnerable social groups. In fact, a larger proportion of respondents from the General category cited institutional requirements as reasons for dropping out, compared with those from the SC/ST/OBC categories.

Similarly, while a slightly higher share of respondents from the very poor category mentioned the dismissal of NHGs as the reason for terminating their participation, no significant difference was found between the yellow and pink PDS card holders regarding the proportion of respondents reporting difficulty in fulfilling institutional requirements. This pattern implies that once enrolled in the programme, the institutional constraints or requirements of Kudumbashree do not significantly discriminate against or disproportionately force out socially or economically vulnerable groups.

Discussion and Implications

Examining the case of non-participation of eligible households in Kudumbashree, a unique welfare-oriented microfinance initiative in Kerala that centres on women’s empowerment and poverty eradication, this paper seeks to contribute to the literature on the inclusivity of microfinance programmes. More broadly, by employing the social exclusion framework, it aims to understand how exclusionary processes can arise even within social protection schemes specifically designed to combat social exclusion or promote social integration. Through its ‘credit-plus’ approach and targeted focus on poor women in Kerala, Kudumbashree aims to address multidimensional deprivations, including gender inequality, poverty, and credit inaccessibility. As such, it provides a valuable case study for exploring exclusion within an anti-exclusion framework.

The concept of social exclusion serves as a fitting analytical framework for our study, as it not only focuses on the characteristics of the excluded groups but also examines the processes that drive exclusion, extending the discussions beyond mere descriptions of individuals’ circumstances (de Haan, 2000; Room, 1995). Additionally, the social exclusion framework, with its adaptability to different local contexts, enables context-specific analyses and policy recommendations (de Haan, 2000). In this study, both quantitative and qualitative data from a survey of 678 rural poor households in Kerala were analysed to comprehensively understand the state and process of exclusion within Kudumbashree.

This section discusses the insights gained from the combined analysis of both quantitative and qualitative data, along with the specific policy suggestions and broader theoretical implications emerging from them.

A key objective of the study was to investigate whether certain population groups are systematically excluded from the programme, with a particular focus on identifying institutional barriers. To achieve this, respondents were categorised based on their socio-economic and religious characteristics. The complementary use of quantitative and qualitative methods offers a dual perspective in this regard, addressing distinct yet interconnected research questions.

The quantitative analysis identifies which population groups are not participating in the programme by examining their socio-economic and religious characteristics. It reveals that women belonging to the ST category, one of the most disadvantaged social groups in Kerala, are more likely to be excluded from the programme, either due to a lower likelihood of joining or a higher propensity to drop out. Religion also significantly influences programme participation. However, within the broader category of low-income households, no significant differences in participation were observed across economic subgroups.

The qualitative analysis uncovers the nature of obstacles different groups face in participating the programme. Findings indicate that, compared with women from the General category, vulnerable social groups—including SC, ST and OBC—are more likely to cite explicit or implicit institutional barriers as reasons for non-enrolment. Similarly, a slightly higher proportion of respondents from the very poor category reported both explicit and implicit institutional exclusion compared with those from relatively economically well-off groups. These observations imply that socially and economically disadvantaged groups are more prone to encountering institutional obstacles at the entry stage, or may be more hesitant to enrol due to the programme’s requirements.

Regarding dropout reasons, a higher proportion of women from the SC and ST communities, as well as those from the very poor category, reported dismissal from NHGs by officials as a reason for discontinuing their participation in the programme, compared with their socially and economically better-off counterparts. However, there was no strong evidence indicating that other institutional requirements—such as weekly meetings or thrift contributions—disproportionately influence the dropout decisions of these vulnerable groups. This suggests that, once joined, the programme’s institutional constraints do not systematically disadvantage socially or economically vulnerable participants.

While the quantitative analysis suggests that religion influences women’s non-participation in the programme, the qualitative data indicate that this is not always due to institutional exclusion. For instance, although Christian women are less likely to enrol and more likely to drop out, qualitative findings do not reveal any disproportionate institutional barriers affecting this group. Given the relatively better socio-economic standing of Christians in Kerala—both compared with other religious groups in the state and to Christians elsewhere in India (Harriss-White, 2002; Zachariah, 2017)— despite their numerical minority status, this pattern is understandable. In contrast, both quantitative and qualitative findings point to a higher likelihood of systematic exclusion faced by Muslim communities.

Another objective of the study was to examine the influence of respondents’ debt attitudes on programme participation. The quantitative analysis reveals a significant association between debt attitudes and continued participation in Kudumbashree. This observation underscores the notion that, despite the array of other benefits the programme offers, Kudumbashree is predominantly perceived as a microcredit initiative and a pivotal credit source for rural poor households in Kerala. This perception highlights the significance of financial incentives in enhancing the retention rate in social protection schemes. However, none of the non-joiners or dropouts explicitly cited aversion to borrowing when qualitatively explaining their reasons for non-participation. This could imply that respondents tend to prioritise material constraints over attitudinal factors when explaining their decisions in response to a single overarching question. In-depth interviews with follow-up questions could provide deeper insights into the role of debt attitudes in shaping participation in microfinance programmes.

Conclusion

By exploring the factors influencing non-participation in the Kudumbashree programme in Kerala, India, this paper enriches the empirical understanding of the inclusiveness of microfinance programmes in developing countries. Designed to combat multiple deprivation and exclusion, detachment from such programmes can further exacerbate existing vulnerabilities and deepen social exclusion, as noted by Sen (2000) and Walker (1995). Additionally, the study identifies a positive link between respondents’ debt attitudes and their likelihood of continued participation, underscoring the role of financial incentives in improving retention rates in social protection schemes.

On a broader conceptual level, this study illustrates the paradox that policies aimed at reducing social exclusion may unintentionally reinforce it. The paper argues that, exclusion in such programmes—where participation is defined as voluntary, and comes with certain conditions for beneficiaries to meet—can manifest implicitly through the interplay of three factors: design-stage conditions, their enforcement during implementation, and individual-level constraints in meeting these requirements.

Finally, this paper acknowledges the critical need to deepen the empirical understanding of non-participation in microfinance programmes across diverse contexts, given the limited existing research. As microfinance initiatives vary in structure and operation, each may embed distinct exclusionary forces. To address this, the study advocates for a comprehensive approach that integrates both quantitative and qualitative data, while considering the unique design and objectives of each programme. Ensuring inclusivity and eliminating barriers for vulnerable groups is not just an aspirational goal for the concept of microfinance; they are essential steps toward realising its core mission.