Lenders have an age-old problem when underwriting a loan or determining a credit limit. Once creditworthiness is established, the ability of a borrower to repay a loan must be assessed under any terms. While this is often done with income and employment verification, those methods can be expensive, point-in-time, and susceptible to fraud (https://www.businesswire.com/news/home/20200122005272/en/). Moreover, many lending processes employ income estimates that are notoriously error-prone. In effect, the ability to pay can be as difficult to assess as the willingness or intent to pay.
In our first blog post, we referred to our RevealedAffordabilityTM data technology, which enables lenders to understand whether a prospective loan is likely to be repaid. Our data elements are derived from granular bank transactions data, which are the byproduct of everyday banking activity. The essence of RevealedAffordability is summarizing everyday activity to draw deep inference about the maximum feasible amount (and the right frequency) of a prospective withdrawal. Actual transaction data from an applicant’s bank account is certain to be the most useful data in assessing affordability. In the parlance of traditional credit analysis, RevealedAffordability is a more sophisticated and reliable method of assessing capacity – one of the famous Five C’s of credit recently referenced in comments by LendingClub (https://www.pymnts.com/consumer-finance/2020/lendingclub-president-fico-model-wont-save-credit-scoring-dinosaur/).
“Sometimes even the wisest of men and machines can be in error.” -Optimus Prime
There is much more to RevealedAffordability than merely observing payroll deposits or inflows and weighing them against withdrawals. In most cases, calculating excess periodic balance (disposable income) doesn’t suffice for assessing the likelihood that a loan of any term will be repaid and will in practice result in declining good customers.
Only 47% of consumers and 52% of small businesses carry a positive daily running balance in the last 30 days with no overdrafts
While many bank transaction providers provide running balance and overdraft snapshot statistics, using them to approve or decline a business could have two negative impacts to profitability. Not only would a bank decline almost half of its applicant pool, but it wouldn’t see materially different credit performance in the bargain. The small businesses we evaluated have similar write off rates (11.1% vs 9.9%), which is only ~10% lower risk and consumers have similar FPD rates (16.1% vs 14.2%), which is only ~12% lower risk.
“There’s more to them than meets the eye.” – Optimus Prime
On the other hand, many people and businesses never miss a payment, but do not appear to carry much if any excess balance in their bank accounts. Others have widely varying bank balances, with transaction patterns that are difficult to discern manually. Still others may have routine transactions which may not appear to yield insight into the ability to service a new recurring transaction. We frequently observe that consumers and small businesses that have current experience with other financial products in their bank account at the same time they are applying for new credit are typically lower risk of default than those with no experience.
Default rates for consumers and small businesses with existing credit obligations are 31% and 34% lower than those without current credit obligations
While it may seem counter-intuitive that existing credit obligations result in lower loss rates RevealedAffordability picks up on this trend and others that reveal ability to prioritize payments.
Transaction Science’s patent-pending technology, combined with historical loan performance from a lender, is based on a simple premise: if a loan is repaid as scheduled, then it is said to have been revealed affordable. Multiple samples of bank data, combined with loan performance, allow us to derive useful estimates of the largest incremental payment a given bank account can bear. The logic of Revealed Affordability implies that better terms cause affordability estimates to change though not always as much as might be assumed. The alignment of the loan period and terms against historical bank transactions provides insight into the prioritization of payments alongside others. A large set of examples of such loan performance has allowed us to create our data technology.
To learn more about our patent pending data technology for Revealed Affordability, send us an email, or call us at 858-935-4477!