Deriving value from data: How AI can power smarter credit decisions

Banks have put AI on the innovations agenda for years now, but have yet to execute. Meanwhile, the underserved SME financing market, a $2.6tn opportunity, could be snatched up by e-commerce platforms, payment processors, and even telecommunication companies.

The time for banks to put AI into action is now before they become displaced by disruptors. Former Cisco chairman and CEO John Chambers famously said that more than 40% of businesses would disappear over the next decade if they fail to execute on an AI strategy. That was back in 2015, so time is of the essence.

A leading global bank has teamed up with a fintech startup, Flowcast, to accelerate its AI execution. Together, Flowcast and the bank have made huge advances in credit decisioning for the supply chain ecosystem. This is a great example of how traditional lenders can leapfrog their peers and disruptors, leveraging their respective core competencies in lending and machine learning.

The rise of disruptors in lending

$2.6tn is the addressable market size of SME financing according to calculations done by the World Bank. This has prompted an influx of players into a space that banks once dominated. Banks are losing market share and rapidly so to more agile players.

The likes of Alibaba, Amazon, and eBay are all capitalizing on small business lending. Amazon alone has made more than $3bn in business loans ranging from $1,000 to $750,000 in less than a decade’s time to help small and medium-sized businesses grow. Payment processors like Square, iZettle and Paypal are also encroaching on the banks’ turf. PayPal Working Capital, which debuted in 2013, has since provided more than 115,000 businesses around the world with over $3bn in financing. They are achieving such scale with minimal manual underwriting.

Traditional lenders’ weaknesses are being exploited by disruptors. Given that AI is instrumental in leveling the playing field, its adoption should be a business imperative.

Sitting on a treasure trove of data

Consider one of our bank partners, a large tier-1 global bank, has over 10 petabytes of data held. This is equivalent to 500,000 days worth of real-time transaction data. Banks need to realize that although disruptors are making bold moves, they are at a huge advantage. We all have heard it before, “Data is the new asset class”. In those terms, banks are sitting on a wealth of assets – but data has no value if it does not lead to insights and actions. That is where AI and machine learning come into play.

Although B2B networks (e-commerce, supply chain platforms) are taking the advantage of additional sources of credit-related data, they are, for the most part, limited to data at the point-of-sale. Banks, on the other hand, have a tremendous amount of historical data, and the rate at which more data is being created is exponential. However, data alone is weak and ineffective.

The ‘data graveyard’

As UBS’ white paper on artificial intelligence suggests, banks face the huge challenge of making sense of legacy data and overcoming data silos and data graveyards. McKinsey further highlights the growing importance of strong data management, and how banks need to digitize their credit processes to remain competitive. While data management is a step, the biggest stride towards unlocking the value of data lies in its transformation.

The application of AI is especially relevant in terms of credit decisioning where transforming data is essential. Flowcast, for example, has helped banks in harnessing their underutilized data and combines it with proprietary alternative data to provide powerful and actionable insights for credit decisioning.

How machine learning can empower banks to serve the underserved market

One of a bank’s most challenging task is accurately assessing the credit risk of SMEs because they have such thin files. SMEs often cannot present the depth of data needed for banks to make well-informed credit decisions. This is compounded by the fact that traditional credit scoring is linear, static and one-dimensional. Rejection rates reaching 74% is the result of the banks’ systemic shortcomings.

AI can help solve this problem. Machine learning utilizes a wide scope set of available and emerging data sources and along with alternative data sources, it can achieve ‘deeper data’ analysis. In this way, machine learning helps to accurately predict the creditworthiness of SMEs that were previously uncaptured by traditional lenders.

This method of credit scoring captures a far greater number of SMEs than traditional scorecard methods, because it allows for thousands of data points to be analyzed and at a far faster rate, almost instantly, than a human underwriter is capable of manually. It looks at alternative risk indicators based on transaction behavior in the ecosystem. Through the dynamic transaction-based risk scoring, the bank is able to actively predict risk and optimize credit appetite proactively.

AI is no longer in PoC stage

According to a recent study by Accenture, AI could boost average profitability rates by 38% and lead to an economic increase of $14tn by 2035. The proven benefits of AI are immense, and banks are well-positioned to adopt AI technology given the breadth and depth of data they collect on a daily basis. However, to gain any value from this data requires a tremendous amount of data transformation – and that in itself is no easy feat. Banks should consider teaming up with fintech firms to accelerate their AI execution, especially because speed to market is critical.

To learn more about how a leading global bank has partnered with Flowcast to achieve powerful data transformation for credit decisioning, download Flowcast’s white paper: ‘Big Data, Smart Credit: Closing the SME Finance Gap through Artificial Intelligence and Machine Learning’.