Artificial intelligence (AI) is transforming many spheres of human activity and financial services is no exception.
In one of the lesser developed regions of China, Xin’An Bank, UNCDF’s partner organization under the i3 Program funded by MetLife Foundation, is testing one such AI technology — natural language processing (NLP) to uplift the lives of women entrepreneurs and improve their financial security.
Ms. Zhang, a vegetable vendor busy at work on a Tuesday morning in Anhui.
Photo credit: Mr. Yunbo Yan, Digital Finance Expert, UNCDF China
Investing in female-run enterprises: The “Goddess AI” loan
On a busy Tuesday morning, Ms. Zhang sells vegetables to her customers. Most women of Anhui have businesses like Ms. Zhang’s. Due to a lack of employment opportunities in Anhui province, many men emigrate to other regions of the country in search of work and send back remittances. This has meant that a large number of women stay back in Anhui to take care of their households and their enterprises. As a supplement to remittances, the women of Anhui run micro and small businesses such as vegetable stalls to keep their households in Anhui going. These women, like Ms. Zhang have financial aspirations for themselves, their families and businesses but often lack the economic wherewithal to fulfill them.
Women entrepreneurs have long been unseen in Anhui. In comparison to other parts of China, Anhui is a more traditional society, where women’s roles are confined to the household. Yet, even as women choose livelihoods that they care about and that can support their families, they struggle. For one, they manage household responsibilities including caretaking, allowing them little time to devote to their entrepreneurial ventures. Second, if they muster the resources to set up a venture, they face bottlenecks along the way to fully realize their entrepreneurial dreams. One of the biggest challenges they face is lack of access to affordable and instantaneous business credit.
Enter Xin’An Bank. Recognizing the entrepreneurial potential of the women residents of Anhui as well as the distinct challenges they face, Xin’An Bank teamed up with UNCDF earlier this year to develop financial services to enhance the financial health of these women. UNCDF has been supporting this engagement by applying a financial health approach, where aspects like customer centricity and an understanding of financial behaviours are at the heart of developing a financial product or service.
With its celebratory product, aptly titled “Goddess AI”, Xin’An Bank processes small amounts of loans to women entrepreneurs, over a mobile app, in under 24 hours. This is especially important for women short on time with limited mobility due to their many responsibilities and roles. The collection of data from potential applicants and its eventual processing is all done online. Women entrepreneurs in need of a loan merely engage with the mobile app, record voice responses to questions and upload requisite paperwork such as identity and business documents.
Upon receipt of the application, Xin’An Bank uses optical character recognition (OCR) technology to verify the authenticity of submitted documents using searchable government databases. In addition, it uses natural language processing (NLP), a form of artificial intelligence that reads, understands and derives emotive and cognitive meaning from human languages to interpret the various responses of potential credit customers. For example, the response to the question, “purpose of loan” will be scoured by the NLP for evidence of hesitation, out of turn pauses and repeated corrections.
Facial expressions are also studied by the NLP to supplement voice responses, and traits such as familiarity with business and self-efficacy are tested. Responses fed into the NLP are analyzed real-time and a credit decision is instantaneously made. Both the OCR and NLP checks work in tandem, helping Xin’An Bank make a fully informed credit decision. In the event of a negative credit decision, Xin’An Bank uses its discretion to examine factors underlying a rejection and re-engages with potential borrowers on a case-by-case basis. Once a loan is approved, loan documentation including registration of mortgage and the transfer of funds are done online, and within a few hours.
Regional Technical Specialist — Asia
Data & Insights Consultant