Scholarlink

<p>Access to working capital is critically limited in the informal retail sector due to the absence of appropriate financial products. Improved credit scoring can thus facilitate increased access to working capital, supporting the growth and sustainability of small businesses in the informal sector. This study addresses this challenge by developing tailored credit scoring models for informal merchants using supervised learning techniques. Leveraging data from a financial technology company in Lesotho (South Africa), we applied logistic regression and support vector machines to predict the likelihood of loan defaults among merchants. Our methodology involved the evaluation of six logistic regression models and twelve support vector machine models, assessing their effectiveness in default prediction. The results provide a robust tool for more accurate assessment of creditworthiness, reducing the risk of lending to potential defaulters. The study underscores the potential of supervised learning methods to create impactful financial solutions and suggests a pathway towards narrowing the financial inclusion gap in the informal African economy. This approach not only aids in risk reduction for lenders but also empowers a critical segment of the economy by enabling better financial support and growth opportunities for informal sector merchants.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Empowering Africa’s Disfranchised SMEs: Machine Learning-Based Credit Scoring for Informal African Merchants

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月2日

查看全文
  • Derick Kazimoto,
  • Said Baadel

摘要

Access to working capital is critically limited in the informal retail sector due to the absence of appropriate financial products. Improved credit scoring can thus facilitate increased access to working capital, supporting the growth and sustainability of small businesses in the informal sector. This study addresses this challenge by developing tailored credit scoring models for informal merchants using supervised learning techniques. Leveraging data from a financial technology company in Lesotho (South Africa), we applied logistic regression and support vector machines to predict the likelihood of loan defaults among merchants. Our methodology involved the evaluation of six logistic regression models and twelve support vector machine models, assessing their effectiveness in default prediction. The results provide a robust tool for more accurate assessment of creditworthiness, reducing the risk of lending to potential defaulters. The study underscores the potential of supervised learning methods to create impactful financial solutions and suggests a pathway towards narrowing the financial inclusion gap in the informal African economy. This approach not only aids in risk reduction for lenders but also empowers a critical segment of the economy by enabling better financial support and growth opportunities for informal sector merchants.