Should gender be a determinant factor for granting crowdfunded microloans?

This study is motivated by increasing women’s empowerment in developing countries linked to the UN Sustainable Development Goals (SDGs), which has boosted microloans to women and to groups where women are the majority. It asks whether men and women differ in microloan repayment behavior, information that is valuable to MicroFinance Institutions (MFIs). Using Kiva data covering borrowers in 56 countries (key SDG implementation areas), the study examines microloans—a tool for low-income individuals and SMEs that contributes to poverty reduction and aligns with social economy and green microfinance objectives. Crowdfunding and digital platforms can securitize and scale microloans, expand outreach, and connect social investors with borrowers, potentially reshaping group lending dynamics and enhancing ESG-aligned development. The paper tests whether borrower gender relates to core microloan features: amount, term, number of lenders, time to contact/fundraising (recruitment period), and repayment system, with environmental concerns as a related area.

The review spans microfinance, crowdfunding, and their overlap, plus gender issues. Microfinance literature over four decades covers MFIs in developing and developed contexts, with key strands on group lending, sustainability, and digitalization. Crowdfunding research examines definitions, models, investor behavior, and platform dynamics (including P2P lending). The overlap (crowdfunded microloans) indicates that platforms like Kiva aggregate small contributions to fund micro-entrepreneurs via MFIs, potentially improving outreach and sustainability while introducing network effects and big data analytics. Gender-focused studies often find higher repayment rates and greater socio-economic impact when lending to women; microfinance may reduce gender inequality and enhance women’s empowerment. In crowdfunding, emerging evidence suggests female micro-entrepreneurs may benefit, especially with prosocial framing and homophily effects, and equity crowdfunding can favor firms with gender-diverse leadership. Behavioral research notes women’s greater risk aversion, and studies show lenders sometimes prioritize personal characteristics (including gender) over business features in crowdlending markets. Overall, the literature supports investigating gender as a determinant in crowdfunded microloans and suggests potential ESG (“green and pink”) synergies where women’s initiatives align with environmental goals.

The study employs multinomial logistic regression to assess whether borrower gender is associated with microloan features. The dependent variable Y (gender category) has three unordered classes: 0 = male or men-only group; 1 = mixed-gender group; 2 = female or women-only group. Predictors are: X1 (loan amount), X2 (loan term), X3 (number of lenders), X4 (repayment system: 0 = monthly, 1 = bullet, 2 = irregular), and X5 (period of lenders’ recruitment: days between posted time and raised time). The multinomial logit links category probabilities to predictors relative to a baseline category; model parameters are estimated via maximum likelihood, with model fit assessed using likelihood ratio tests and goodness-of-fit statistics (Pearson χ², G²). Robustness checks include Box-Tidwell for linearity in the logit, Cook’s distance for influential points, Weighted MLE versus classical MLE (yielding similar results), and VIF for multicollinearity. Data come from Kiva: N = 1,048,575 loans disbursed between July 25, 2007 and June 30, 2020. A simple random sample of n = 385 loans was drawn using a standard sample size formula (z = 1.96, p = 0.5, e = 0.05). The 385 loans span 56 countries and 14 sectors. Gender composition in the sample: 71 individual men, 263 individual women, and 51 groups (2 men-only, 32 women-only, 17 mixed). Variables were assembled from Kiva’s public snapshots; interest rate data were unavailable (Kiva loans are interest-free to lenders), and information on payment delays was not provided.

• Model fit: The full model log-likelihood LL1 = -222.8479 versus constants-only LL0 = -250.1453; likelihood ratio χ²(10) = 54.5949, p = 3.759e-8, indicating significant improvement over the null model.
• Significant predictors (as reported): X1 (loan amount), X4 (repayment system), and X5 (recruitment period) are overall significant; X2 (term) and X3 (number of lenders) are not.
• Category-specific estimates: • Category 2 (female) vs 0 (male): X4 coefficient = -0.9943 (p ≈ 5.9e-5; exp(b)=0.3700) and X5 = -0.02976 (p ≈ 0.000524; exp(b)=0.9707) are significant; X1, X2, X3 are not significant at 5% in this contrast. • Category 1 (mixed) vs 0 (male): X1 coefficient = 0.7873 (p ≈ 0.0167; exp(b)=2.1974) is significant; X2, X3, X4, X5 are not significant at 5% in this contrast.
• Interpretation highlights (from the reported odds effects): • For female vs male borrowers, more reliable repayment systems (monthly vs bullet/irregular) and shorter recruitment periods are associated with higher odds of being female. • For mixed vs male groups, higher loan amounts are associated with higher odds of being mixed.
• Correlations among predictors indicate high correlation between loan amount and number of lenders (r ≈ 0.8066), suggesting overlapping information content.
• Descriptive scope: Sample includes 385 loans across 56 countries and 14 sectors; the Philippines and Tajikistan are prominent in the sample’s country counts.

Findings address the research question by evidencing gender-related differences in microloan features within a crowdfunded setting. Women emerge as more reliable borrowers based on repayment system and recruitment period, aligning with mainstream microfinance literature that documents better repayment outcomes among women. These results have policy and practical implications for MFIs and crowdfunding platforms seeking prosocial impact and ESG alignment: targeting women can enhance financial inclusion and outreach, supporting SDG 5 (gender equality) and broader sustainable development. Crowdfunding may reduce capital access barriers faced by female entrepreneurs, reinforcing links between crowdfunding and microfinance. Networked digital platforms and MFIs act as key nodes enabling real-time monitoring, data collection, and scalability, potentially improving the quality and sustainability of lending operations. While debates persist about broader welfare implications and possible unintended consequences for women, the study contributes empirical evidence that can inform lender strategies, platform design, and supportive policies to promote equitable, sustainable finance.

The study contributes evidence that gender matters in crowdfunded microloans: women, compared to men, tend to be associated with more reliable repayment systems and shorter recruitment periods, and overall results support the view that women are strong micro-borrowers. This aligns with SDG-related goals on gender equality and supports the case for targeting women to foster sustainable financial inclusion. The paper bridges practice and academia, suggesting that “green and pink” microfinance—where female-led initiatives intersect with environmental goals—could attract ESG-compliant crowdfunding resources. Future research directions include: refining credit-scoring models for female-led projects; exploring gender homophily between crowd-investors and micro-borrowers; analyzing group composition effects (proportions of men/women) on lending dynamics; and expanding data on repayment quality (e.g., delays) to strengthen inference. Such work can guide practitioners and regulators in optimizing inclusive, ESG-aligned microfinance.

• Data constraints: Interest rate information was unavailable (Kiva loans to lenders are interest-free), and data on delays in payments by micro-borrowers were not provided. These omissions limit the analysis of repayment quality dimensions.
• Platform scope: Results are based on Kiva data and may not generalize across all microfinance or P2P platforms with different models or data availability.
• Sample: Although randomly drawn, the analytical sample is 385 loans from a very large population, which may introduce sampling variability.
• Correlation among predictors: High correlation between loan amount and number of lenders may affect individual coefficient interpretability despite multicollinearity checks.
• Model coverage: The study focuses on association, not causality; other relevant predictors (e.g., interest rates, borrower credit history, local economic conditions) were unavailable.
• External validity: Context-specific dynamics (countries, sectors, time window 2007–2020) may affect generalizability.