The use of alternative data sources and machine learning to assess the credit quality of applicants gives US fintechs an advantage over banks when it comes to small business lending, according to a paper from the Bank for International Settlements.
Using data sets from alternative lenders Funding Circle and LendingClub, BIS assessed the geographical footprint of fintech loan origination and the effectiveness of their credit scoring methods in predicting default.
The researchers conclude that fintech lending platforms lent more to small businesses in ZIP codes with higher unemployment rates and higher business bankruptcy filings.
Their internal credit scores were able to predict future delinquencies more accurately than traditional Fico scores or VantageScores, with a large uplift in the area under the receiver operating characteristics curve. Notably, the improvement in predictive performance was highest in areas with higher unemployment.
Concludes the paper: "This indicates that the use of alternative data and machine learning is probably a key factor in fintech lenders' contribution to improving credit access for small businesses."
Read the full paper:
Download the document now 4.5 mb (Chrome HTML Document)