Key to Snaring US$5 Trillion in Unmet Credit Demand? Artificial Intelligence

Rich Data Corp

US$5 trillion in global credit demand is unmet. How to tap it? With artificial intelligence.

How do you lend to people and companies with no credit history, or even a bank account? By using artificial intelligence and non-traditional data sets.

THEY ARE a huge part of the world economy, and often the strongest drivers of economic development and employment in developing nations. But micro, small and medium businesses also the least able to grow: US$5 trillion of credit demand from these businesses is unmet annually, according to the G20’s Global Partnership for Financial Inclusion.

And that’s a massive market waiting to be tapped.

But how do you lend to people who don’t have little or no credit history – or even a bank account? Using artificial intelligence (AI) is the key, coupled with vast amounts of data and very nuanced algorithms. That’s what a partnership between financial AI company Rich Data Corporation and Inspur Group Co Ltd, China’s biggest cloud computing solutions provider, is delivering.

“We’ve established that you can get a really accurate understanding of a person’s credit risk by cross-referencing a whole lot of non-standard data sources,” said Ada Guan, chief executive of Rich Data Corp. “Customers don’t need to fill in reams of forms – you rely on data that already exists, like payment transactions and telecommunications usage, then use artificial intelligence to analyse that data from a lending perspective. And then you can really predict likely behaviour.”

These insights are also applicable to mature markets in Europe, North America and Australasia, allowing credit providers to better understand the needs and risk profile of existing or potential customers, according to Rich Data, a global player with offices in Singapore and China.

The Sydney-based company has established a partnership with Inspur and proven that their proprietary algorithms – trained over the past three years on other large data sets from around the world – are exceptionally accurate at assessing credit risk.

Inspur is the world’s third largest server manufacturer and a leading e-government solution provider in China. The Shandong-based company manages the IT governance of 110 local government bodies in China and – with over 30 years of such experience – has deep knowledge and understanding of the Chinese small business sector.

That partnership is set to expand, with Inspur engaging Rich Data to assist banks in using government data to better assess small business credit risk. This will set the groundwork for extending credit to more small businesses, a stated mission of the Chinese government.

“We’ve helped several banks improve credit assessment so they can confidently lend to small business clients,” added Guan. “That’s not only validating our approach, but improving it, as the prediction models become more sophisticated the more real-world examples they’re exposed to.”

Traditionally, banks and credit providers rely on credit scoring based on simple regression or expert scorecards which assesses a potential client’s creditworthiness. But due to the often low predictive capacity of these scoring approaches for those with a limited credit history, lenders tend to either charge a higher fee, or simply don’t offer credit at all. This leaves large swathes of un-banked or under-banked people and businesses without credit.

The G20’s Global Partnership for Financial Inclusion estimates that 131 million – or 41% – of micro, small and medium enterprises (MSME) in developing countries have unmet financing needs, a gap totalling US$5 trillion. Women-owned businesses are particularly hard-hit, comprising 23% of businesses but 32% of those with unmet financing.

In addition, the World Bank estimates that 1.7 billion individuals do not have a bank account and are therefore excluded from the formal financial system, a statistic it and other international agencies are working to change. In China alone, 200 million rural adults remain outside the formal financial system. Insights gained by Rich Data in its Inpsur partnership will also help clients working with consumers to better assess credit risk for businesses.

Rich Data uses an AI technique known as ‘transfer learning’ to extract insights from lending areas with sufficient data, such as consumer lending, and apply these to domains with less data, such as unsecured lending to small business.

Using state-of-art machine learning approaches, Rich Data is able to add non-standard data sources for individuals and businesses – such as census data, bankruptcy rates by geographic area, and utility bill payments – to build up an understanding of credit risk and likely customer behaviour. It then takes what has been learned from one industry, product or domain and transfers that knowledge into other domains. This can then be run alongside existing credit scoring models and tested on a client’s real data.

Importantly, how Rich Data’s software arrives at a credit score is explainable to human decision-makers, removing the ‘black box’ component of many machine learning algorithms, which deliver results but not how they were arrived at. What’s more, that ‘explainability’ puts that data into a credit risk context for lenders, helping to address concerns around fair and responsible lending.

“That Inspur, one of the biggest technology companies in China, has selected us to develop credit risk analysis with them, speaks volumes about the clout of our software,” said Gordon Campbell, Rich Data’s chief of product and strategy. “It also legitimises our credentials as a safe and trustworthy partner when it comes to privacy and security.”

/Public Release.