Using non-traditional data for underwriting loans to thin-file borrowers: Evidence, tips and precautions

Sustainable growth in underserved domestic markets has long been a challenge to lenders. Recent testing with non-traditional data in automated underwriting shows promise as a means to profitably extend credit to the ‘thin-file’ and ‘no-file’ populations without assuming undue risk. This area is in its infancy, and is fraught with risk and challenge. Despite the potential, lenders are advised to proceed with caution and should slowly test their way into this segment with the new methods. As this is a slow process, one of the key challenges is to get the needed commitment from the lending institutions. A prudent credit risk officer can harness the power of non-traditional data by taking a disciplined and methodical approach to testing and implementing. This paper demonstrates the value of non-traditional data as a powerful tool for consumer credit risk assessment while highlighting some of the potential risks and precautions that lenders need to think about before using these tools. Special emphasis is placed on paying attention to the capacity of these customers and creating a life cycle strategy for them that includes credit education. This paper presents some empirical test results, and outlines steps that should be taken by lenders to capture the full value of the data while mitigating risk.