← Search

Do Black Fintechs Matter? The Long and Winding Road to Develop Inclusive Algorithms for Social Justice

Eduardo Henrique Diniz1; Bruno Henrique Sanches2; Marlei Pozzebon3; Simone Luvizan2

1 Department of Technology & Data Science, FGV EAESP, São Paulo, Brazil & Coppead UFRJ, Rio de Janeiro, Brazil · 2 Department of Technology & Data Science, FGV EAESP, São Paulo, Brazil · 3 Department of International Business, HEC Montreal, Montreal, CANADA, and Department of Technology & Data Science, FGV EAESP, São Paulo, Brazil

MIS Quarterly 2024

Racism in the financial sector is a complex global phenomenon involving intertwined social, economic, and digital systems. Credit denial rates for Black applicants suggest that financial systems have assimilated racial discrimination into their algorithms and credit scoring tools. Considering that digital tools incorporate their developers’ life experiences and perspectives, and in view of the low representation of the Black community in the digital startup universe, it is evident that even well-meaning fintechs are reproducing racial bias in their technology-intensive business models. In this paper, we investigate how three Black-owned Brazilian fintechs design and use inclusive algorithms to decrease persistent social injustice involving racial financial inclusion. We combine the concept of sociotechnical reconfiguration, taken from the South American tradition of “tecnologia social” (social technology), with the three core concepts of Nancy Fraser’s theory of social justice—representation, recognition, and redistribution—to analyze the process by which Black-owned fintechs develop their particular solutions for combating racial bias. We found that pressured by the need to be profitable, the lack of capital to develop their own credit scoring tools, and the low profile maintained by the Black community ecosystem, Black-owned fintechs focus on a sort of affirmative experimentation to gradually create original digital solutions aimed at achieving social justice in the financial system.

DOI
10.25300/misq/2024/18288
Volume
48 (4)
Pages
1721-1744
Language
en
Export
BibTeX
Sources
crossref