For decades, lending institutions like banks and NBFCs(Non-Banking Financial Companies) have been judging the borrower’s credit worthiness using the past financial data. For example: ratio of the income for last year to the second last year, change in profits over the past years, and, one of the most favorite for the lending institutions, average balance analysis of the past months.
Analyzing the current credit capability of a borrower solely based on his past poses two major problems:
- The businesses who have been doing really good in the very recent past are assigned a good credit limit. This has proved dangerous because on solely analyzing the past, the future downfall trends of the businesses is ignored thus leading to increasing NPAs(Non-Performing Assets) for the lending institutions. We have seen this in multiple economies esp. in India, during the bad-loan crisis around 2017, where companies with good financials defaulted on big loans due to business downfall.
- The businesses who have not been doing really good in the recent tend to get either rejected for credit or get a much lower credit sanctioned even if they have a high future growth prospect. This hurts the business severely as these businesses are in dire need of credit and a proper credit stream is usually the criteria to make or break such businesses. In either case, the economy on the whole suffers when businesses do not flourish. This point, in fact, has a correlation with point-(1). As lending institutions suffer large size NPAs from established big players as mentioned in (1), they cut short the funding to the financially weak businesses (by solely judging from their past).
The aim for this paper is to use a combination of modern mathematics, statistics and machine learning algorithms to predict the future financials of a business based on its past data and subsequently increasing the accuracy of the predictions.
Please fill out the form to access the content
Resource Sponsored By
