After the financial meltdown of the last decade, many lenders have found themselves in a tough situation where they just cannot use the conventional underwriting guidelines to identify enough “good borrowers” to whom they can lend money.
For example, subprime auto lender Prestige Financial Services found itself rejecting 25% more potential borrowers, and its managers realized that they HAD to find a way to increase their volumes, without simultaneously substantially increasing their credit risk by reducing their approval standards.
Wrestling with how to boost business while maintaining underwriting standards, Prestige partnered with an artificial-intelligence software developer (founded by Google’s former CIO) in 2017, and started drilling down into some 2,700 borrower characteristics, instead of the 20-30 that they typically relied on in their risk-assessment scorecard, by using advanced artificial intelligence (AI) techniques including machine learning (ML).
Instead of looking simply at whether a potential borrower has ever filed for bankruptcy, for example, the machine-learning system helped Prestige consider such factors as when the bankruptcy happened, and analyze that data with other variables, including previous car-payment records and time spent living in his or her current residence. In 2018 Prestige said its monthly lending volume doubled, to $55 million from a low of $25 million during the tightest credit conditions. They also reported that the new loans perform as well as those made under the old system.
In another scenario Synchrony financial – credit card business that GE spun off in 2015, moved towards using AI based techniques to ease the access to credit which has been challenging due to the quadrupled growth of data and diverse data type in 20 years, and is expected to grow further exponentially. The new technique is helping to look at hundreds and thousands of “signals” including rent payments, phone bills or months since the recent delinquency to process credits to customers.
AI & ML are changing the way credit has been assessed for the last 3 decades using only credit bureau and credit bureau data. The new approach allows lenders to deep dive into their existing customer base to look for new credit approval metrics based on their purchasing history, bank data or social media habits to better forecast defaults even for people with low credit scores or thin credit file.
What can you do?
Here are ways to use AI to harness historical data in taking smarter business decisions allowing better prediction of results of loans;
- You must have a clear goal to generate results that can be used in the real world. You must aim to solve a critical problem that matters to the business leaders.
- You MUST have an internal champion who can move the project along, and get risk, legal, compliance and business teams engaged in the process.
- Computing power is easily available, thanks to the hosted “cloud” environment. You HAVE to get as much data into the model as possible to maximize return on investment.
- Lastly, lending and credit companies should ensure the Machine learning models’ regulatory compliances.
