‘Banks must embrace AI, ML, NLP to lend to SMEs that don’t have credit history, are untapped by formal banking’

Credit and Finance for MSMEs: Banks have historically relied on analysing financial statement data and taking collateral while lending to SMEs. This has often led to sanctioning of lower credit limits and an inadequate assessment of the risk.

msme credit
The entry of financial information providers onboard the account aggregator system will provide banks with more financial data that can be used to fine-tune the analytics for SME lending-rating models, early warning triggers etc. (Image: pixabay)

By Dipesh Doshi 

Credit and Finance for MSMEs: Financing to small and medium enterprises (SMEs) has always been challenging for banks with problems ranging from inadequate data to poor financial astuteness of promoters. Yet, SME lending is a lucrative business avenue, especially in the wake of shrinking volumes and margins from large corporates. This has been recognized by some new-age NBFCs and fintechs that have moved into this commoditized lending space, trying to differentiate themselves and pose competition to banks. For banks, true differentiation lies in the deployment of new-age analytics and tools in the entire credit lifecycle-prospecting, appraising, onboarding and servicing of customers.  

Banks have historically relied on analysing financial statement data and taking collateral while lending to SMEs. This has often led to sanctioning of lower credit limits and an inadequate assessment of the risk. SMEs have often complained about being starved of funds and at the same time, bankers have spoken about the high delinquencies in the SME segment. 

Consider for example how banks treat all auto component manufacturers (SMEs as well as corporates) as one segment and peer benchmark them against each other based on historical performance. These clients would be catering to different segments such as commercial or passenger, bicycles or two/three/four-wheeler, original equipment manufacturers (OEMs) or replacement or exports and making different products.  

We know that post-Covid, demand for bicycles went up for two reasons, first, rural consumers downgraded from motorbikes to cycles due to lower incomes, and second, high-income people substituted gyming and other physical activities with bicycle riding. Also, the demand for two-wheelers increased as low-income consumers who used shared mobility solutions and public transport bought their own vehicles.  

However, banks were wary of lending immediately post-Covid, especially for SMEs who had opted for a moratorium and other forbearance schemes. With no additional security to post, SMEs catering to the bicycles and two-wheeler segments could not capitalise on the heavy order inflow. This is a clear case of opportunity loss for banks because they do not incorporate forward-looking information in their assessments based on client-specific needs. 

Leveraging Data 

To meet this challenge, banks can supplement the regular financial data with alternate datasets such as GST returns, online transaction data, and mobile phone data to generate views on the credit quality of SMEs. In the above case, for example, the bank can use data such as monthly GST collections, revenue guidance by bicycles/two-wheeler/four-wheeler auto OEMs, auto loan disbursals for different types of vehicles in the preceding month, crude oil prices, rural incomes, GDP of target export markets, etc.  

These vast and diverse data sets can be processed together and meaningful insights can be drawn by using a combination of artificial intelligence (AI), machine learning (ML) and natural language processing (NLP) techniques. Thus, banks can develop an augmented credit view that can lead to a reduction in the traditional reliance on collaterals to mitigate the risk and decide lending limits, thus enabling a more nuanced and customised lending approach based on cash flows.  

Another significant challenge is that once an SME is onboarded, banks have struggled to get real-time monitoring of the SMEs’ credit health. For example, post the anti-pollution crackdown in China, many SMEs which were importing cheap raw materials from there were forced to find expensive substitutes elsewhere. Similarly, many SMEs engaged in import/export saw their transportation costs go up post-Covid on account of the global container shortages.

Subscribe to Financial Express SME newsletter now: Your weekly dose of news, views, and updates from the world of micro, small, and medium enterprises 

In both these cases, SMEs’ working capital requirements went up as inventory costs and lead times increased while margins declined. Pressed with the need to arrange for funds immediately without sharing the details of their deteriorating financials, many SMEs availed unsecured business loans from NBFCs at more than 1.5-2x of the interest rate charged by banks. Such high costs and short-term debt impacted SMEs’ net profits even more and created liquidity mismatches.  

The banks were caught unaware of such developments until the account was put up for renewal or until the client started delaying the repayments. To avoid such surprises, banks can consider using data analytics-based early warning triggers. These triggers are based on real-time tracking of macroeconomic factors, financial results, bank account entries, statutory compliances, social media behaviour etc. which are specific to the client. The data generated from these sources is then put through AI/ML techniques to compute a credit score and raise red flags in accounts that have the potential to become delinquent in the next few months.  

In the given example, changes in Chinese government regulations (macroeconomic factor), global shipping crisis (macroeconomic factor), disbursal of NBFC loans in the current account (bank statement entries in real-time), etc. would constitute few early warning signals and would have alerted the bank to take prompt corrective action. These data-based tools make credit monitoring for the entire portfolio, proactive rather than reactive and foster better visibility and coordination between credit and business teams through common dashboards. This could potentially replace the current system of mandatory annual reviews of all accounts with a need-based system of reviews at appropriate periodicity leading to better optimisation of a credit analyst’s time. 

The entry of financial information providers onboard the account aggregator system will also provide banks with more financial data that can be used to fine-tune the analytics for SME lending-rating models, early warning triggers etc.

In a nutshell, banking needs to embrace new techniques and tools like AI, ML, and NLP to enhance their credit assessments and credit monitoring process and fulfil the immense need to lend to SMEs that do not have credit history and are still untapped by the formal banking channels. Therefore, it has become imperative that banks intelligently use analytics to derive actionable insights from data to turn the SME problem into an opportunity. 

Dipesh Doshi is the Managing Director, Financial Services – Protiviti Member Firm for India. Views expressed are the author’s own.

Get live Share Market updates and latest India News and business news on Financial Express. Download Financial Express App for latest business news.

Photos