The ability to Identifying a borrower’s ability to re-pay a loan is a mixture of Art and Science. Generally, an underwriter will manually reviews the borrower’s credit history, collateral value, credit score, grade and recommended system decision to approve or reject a loan. I believe including Artificial Intelligence in the Underwriting process will greatly enhance and simplify the process. I think time has come to include AI to contribute to the Underwriting process.
Making swift and accurate credit decisioning is the key for winning in the digital market-place. Modern Origination Systems provide automatic technical analysis and decision capabilities to assist in the underwriting process. This allows more loan applications to be processed in less time and fewer human resources are required. Scoring and auto-decision engines use following parameters, calculation and empirical scoring to recommend a decision based on pre-configured rules.
o Customer, Collateral Age, Base income
· Financial Ratios Calculations
o Debt to Income
o Payment to Income
o Applicant Stated Vs Actual Verified
o Before and After New Loan
· Income, Liabilities, Assets and Net worth Calculations
· Credit Score Cards
· Credit History Evaluation
o Past Bankruptcy, Foreclosures, Liens and Judgments
o Near Past Credit delinquencies, repossessions, collections, or charge-offs
o Different types of credit account, age, limits, usage and status
o Borrower's request for new credit in recent past
This automatic analysis is adequate but not enough. It looks as at very small set of data. Based on my personal experience looking at multiple score cards instead of relying on a single credit score card increase the probability of identifying potential borrowers Instead of rejecting a credit applicant it’s better to provide counteroffers based on the credit potential Borrower may be eligible for the more credit or may require putting more down payment to qualify for an offer.
Artificial Intelligence looks at thousands of variables to analyze patterns and identify the risk profile of a potential borrower. AI combines numerous variables with iterative processing and intelligent algorithms, allowing the software to learn automatically from the patterns in the data. These data meta variables are fed into multiple models, each with a different perspective. These models vote to create the final score and the decision.
In addition to the origination and credit data it is very important for the self-learning capabilities of the models to feed current credit information and portfolio performance data to the AI Models. These modern technologies are becoming fundamental while evaluating your next Origination System (on-premise or on-cloud) for the success and growth of your business. Feel free to contact me for more information and questions.
About the Author:
Vinay Bhatia is an experienced product leader, passionate about building financial products that delight customers. He has an extensive background in technology, cloud and agile software development.