PointPredictive Inc., the San Diego-based company that builds Artificial & Natural Intelligence [Ai+Ni] models using machine learning, has identified that 1-in-5 applications for loans have incomes that are materially inflated.
In a study of over three million applications, PointPredictive compared the stated income from the application to the borrower’s actual income and determined that, on average, 20% of stated incomes were materially inflated by 15% or more. In the same analysis, the income that applicants report to one lender may differ significantly from what they report to another in short timeframes. In fact, of those applicants that applied to multiple lenders, 1-in-10 changed their income by over 50% compared to what they reported to the original lender.
“While the risk of income misrepresentation on loans is high, the approach that lenders are using to identify and prevent it from happening is often counterproductive,” indicates Frank McKenna, Chief Fraud Strategist of PointPredictive. “Some lenders are requesting paystubs on up to 100% of approved applications, but what they receive includes countless forged or internet-generated fake documents. Our research indicates that one in every 12 paystubs a lender receives in response to a request for proof of income is fake or falsified in some manner. Requesting paystubs as ‘proof’ to validate income is a broken paradigm – every applicant or fraudster with a need to do this has already figured out the workaround. As a result, lenders’ risk management departments are overwhelmed with so many paycheck stubs that the quick scans necessary to keep up with the volume often miss the forged or fake documents. In addition, consumers will sometimes abandon loan processes where lenders ask for this documentation because it puts an undue burden on the consumer to prove their information.”
As part of the analysis, PointPredictive enlisted their Fraud Analyst team to research the problem of fake paystubs. The analysts identified more than 300 websites offering to generate fake paystubs at a cost generally between $5 and $15. The average time it took the fraud analysts to create a realistic fake paystub at one of these sites was four minutes. The prevalence of fake paystub sites means that requesting paystubs as proof of income has become notoriously unreliable.
“Changing the way we think about validating income is critical to the future of lending,” says Tim Grace, CEO of PointPredictive. “The days of requiring applicants to provide paystubs to get a loan are over. Sixty percent of employees have direct deposit and never even get a paper paystub; requesting it from the applicant really disrupts the lending process. Asking consumers to produce a paycheck stub often requires them to go to their employer and figure out how to obtain such a document. When asked, many employees have never even seen what their paycheck stub looks like or know how to get a copy of it. That’s the type of barrier that will prompt many of today’s borrowers to look elsewhere for loans or credit cards. We believe a better way is to leverage technology to accurately judge an income’s reasonability BEFORE asking for onerous documentation. We launched IncomePass™ last year to fundamentally change the way incomes are validated.”
IncomePASS analyzes a borrower’s stated income against millions of reported incomes and salaries from seven diverse sources. Then, using the applicant’s employer, occupation, job title, residence, and estimated years of experience, a sophisticated machine-learning model predicts if the borrowers stated income is within range.
Real-world lender validation of the solution reveals that the approach is working. Results from lender tests show that, on average, stated income on 75% of applications can be trusted and cleared at an accuracy rate above 90%. For those lenders that request paystubs on 100% of their loans, they can now choose to selectively target the subset of loans where the income is unreasonable or likely to be inflated to confirm if the income and documentation is valid. Alternatively, for those lenders that want to be able to validate 100% of their stated incomes, they can use IncomePass to validate all of them prior to funding.
IncomePASS is available to banks, lenders and card issuers to validate stated income on applications. The service is available today for real-time integration into loan origination and underwriting workflows.
For more information on IncomePass from PointPredictive, please contact email@example.com.
About PointPredictive Inc.
PointPredictive Inc. enables lenders to fund more loans simply with a unique combination of Artificial Intelligence and Natural Intelligence [Ai+Ni] to power machine learning technology solutions. PointPredictive helps automotive, mortgage, retail and personal loan finance companies identify the consumer applications with truthful and reliable information without the intense interrogation and verification of data caused by lower tech solutions currently in use. With the legacy of being the most trusted fraud and misrepresentation analytic solution providers, PointPredictive has transformed that trust to enable lenders to fund more loans to more consumers more easily. PointPredictive uses big data powerfully orchestrated from millions of examples of true and falsified loan applications, billions of derived proprietary data elements, and scientifically selected third-party data sources to build powerful machine learning models with the added natural intelligence of human experience. Located in San Diego, California, more information about PointPredictive can be found at www.pointpredictive.com.