Source: businessworld.in
Traditional financial institutions have thus far treated rural customers as a monolithic whole. This is one reason rural customers often prefer informal providers or remain excluded from formal financial services. Significant advances in technological innovations coupled with the availability of granular data at individual levels, imply that the needs of a large rural population can now be analysed better to provide customized solutions, factoring-in and respecting rural customers’ heterogeneity.
Despite misgivings of usury, for generations the ubiquitous moneylenders or sahukars have continued to be patronized by communities and are hard to dislodge, for rural credit across India. Moneylenders play an important role in rural communities mainly because of the flexibility and choice they offer customers: a) doorstep service b) little or easy documentation c) instant approval d) guarantee that one’s application for a loan will not be rejected e) often no guarantor requirement f) loan amount and repayment schedule tailored to customer cash flows g) low prejudice around past borrowing history. This does not diminish the sometimes-catastrophic downsides customers incur, since the informal provider will not hesitate to use muscle or act outside the law.
In contrast, for a formal financial institution with structural issues such as centralisation, inadequate local autonomy, rigid policies and products, resource constraints and other factors, there is very little flexibility in decision making. This leads to significant formal financial exclusion for rural customers.
AI can help to change this for formal providers.
To compete in the rural context, financial institutions need to offer customized solutions to individuals addressing the heterogeneity of the rural population, while minimizing risks and keeping costs low – imperatives to remain profitable. This is where emerging technologies such as Artificial Intelligence (AI) can help. AI and Machine Learning (ML), along with other techniques such as Natural Language Processing (NLP), robotics, and predictive analytics can be used to design customized, flexible, cost effective solutions for rural populations that are financially excluded and are often clubbed with urban customers.
For example, a fintech providing last mile distribution services across remote rural geographies was plagued by low first attempt resolution of routine queries from field banking agents. Simple queries, such as a password reset, when not addressed instantaneously lead to lost sales and commissions. To address this, the company developed a menu-based ‘smart-bot’ app designed specifically for low- literacy settings to automatically handle first pass queries. On deployment, first attempt resolution went up by a whopping 60%. Call volumes for quick resolution queries at the call centre also reduced and they were able to dramatically shift agents from a costly human call centre to a digital channel to self-service their needs.
Similarly, people with low-incomes are far more likely to be credit invisible and generate data-points that cannot be scored through normal appraisal processes. In India, that’s more than 700 million people. A leading Indian fintech is offering a simple and intuitive solution to help micro-borrowers establish financial identities so that they can qualify for loans from formal financial institutions. The app, called Pehchaan will help people without a formal credit history and in resource poor settings to build their financial identity. The app can also help lending institutions identify creditworthy borrowers, customize their product suite to meet the varying needs of the micro borrowers and mitigate credit risk due to un-scorable or invisible borrowers.
As illustrated in the example cases above, AI can deliver business insights against large diverse data sets in a real-time and cost-effective manner. It can offer personalised experiences and build agent loyalty. It can help increase customer conversion and open new markets for financial institutions. Use of AI based chatbots with the ability of Natural Language Processing can encourage greater engagement of rural customers.
Building an AI/ML model requires a clear definition of the problem statement, availability of sufficient quantity of contextually specific training data, decisions on whether the model needs to perform real time and contingent data storage implications. For example, the chatbot in the example described previously only required one week’s worth of calls from agents to be trained to handle the quick resolution queries at 60% higher efficiency. However, an AI fraud engine started out at ten thousand records and went up to one hundred thousand records before meeting reliable performance goals. These factors influence the levels of investment and the organization’s business case for AI.
Initiatives by the government of India towards financial inclusion, coupled with disruptive innovations in the fintech space have started showing positive results. Almost all the income groups in rural India have benefited from new government schemes that address issues related to income inequality and growing disparity between urban and rural India. However, financial institutions still have not been able to effectively navigate the heterogeneity of rural markets due to technological and infrastructural bottlenecks. We are witnessing trends towards increased heterogeneity and deeper basic financial inclusion. The role of authentic and un-siloed rural data, and disruptive next- gen technologies such as artificial intelligence can’t be stressed enough as the enablers for a massive transformation toward a more inclusive India. By using data, AI, ML and NLP judiciously, financial institutions can create personalized offerings for rural individuals and have a viable business model at the same time.