The gendered algorithm: navigating financial inclusion & equity in access to credit

Thesis

Artificial intelligence (AI) has the potential to help solve global problems and be employed “for good”. One area of immense recent investment and interest is the financial technology (“fintech”) sector. Boasting its ability to provide financial services for the underbanked, various startups are developing apps that collect mobile phone data and use machine learning (ML) to provide credit scores – and subsequently, opportunities to access loans – to groups often left out of traditional banking in low- and middle-income countries (LMICs).

This thesis explores whether ML-based credit assessment tools by fintech companies reinforce or mitigate gender inequitable access to finance in LMICs. I ask: (1) In what ways do the underlying logics, design choices, and management decisions of ML-based credit assessment tools by fintechs embed or challenge gender biases, and how do these choices influence gender equity in access to finance? (2) What benefits and challenges do users experience, and how do these compare between women and men? Feminist and postcolonial theory, as well as Science and Technology Studies (STS), focus on the role of power and maintain that it matters how – and by whom – technologies are designed and managed. These theories prompt my research questions, while also informing my hypotheses and parallel mixed methods approach.

My findings reveal that while fintech innovations hold promise, they fall short in addressing gender inequitable access to finance. Algorithmic lending tools are shaped by underlying logics of their developers. Developers and managers do not consider or adequately address how gender shapes access to and use of the technology, nor how gender “blind” algorithms and profit priorities can inadvertently privilege male-coded financial and digital behaviors. Perceptions of fairness by fintechs fail to challenge – and even legitimize – gender inequities in financial access. Both male and female users report positive benefits that the technology facilitates, yet gender differences persist in app access and use. In addition to the empirical contributions, my findings expand existing theories of algorithmic bias and feminist STS.

Authors: Smith, G

• DOI: 10.5287/ora-v0zqn04ap
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: fairness; machine learning; Kenya; gender equity; gender; financial inclusion; artificial intelligence
• Subjects: Equality; Machine learning; Gender; Finance; Fintech