Financial technology firm Plaid has set forth Plaid LendScore, a credit risk score that harnesses real-time cash flow data and insights to offer lenders an up-to-date view of borrower risk.
In addition to cash flow information, Plaid LendScore leverages unique account connection data from the Plaid Network, which delivers differentiated signals about a borrower’s financial health that can enable advanced predictive models. Made available through Plaid’s consumer reporting agency, the solution comes as a more optimal way to evaluate risk and equips consumers with the ability to share a more complete view of their financial lives, enabling them to benefit from expanded access to credit.
Taking credit risk scoring to the next level
Even if traditional credit scores still play their part, they utilise historical data and might miss what the borrower’s financial life is currently impacted by, negatively or positively. By launching LendScore, Plaid aims to take a different direction to assessing credit risk, with the solution using cash flow insights, income patterns, and financial account connections to reveal a borrower’s real-time financial situation. Bringing these capabilities allows LendScore to function alongside traditional scores and improve credit decisioning.
Plaid’s first model, LS1, is trained on a dataset of approximately one billion transactions to predict the probability of default in 12 months for unsecured loans. The company also inked a deal with FairPlay, an advanced AI fairness techniques provider used by banks and fintech firms, to complete an independent evaluation of its model. Additionally, after a borrower consents to sharing their bank account data through Plaid Link, lenders can call Plaid’s API to get a score from 1 to 99, together with adverse action reason codes to assist in FCRA and ECOA compliance.
Early results and offering
Plaid developed LendScore in collaboration with unsecured lenders, which supported the company in creating the attributes it puts forward, the structure for its reason codes, and how it measured model performance. According to company data, in the testing phase, it generated millions of scores and provided a 25% increase in predictive performance for lenders, compared to traditional credit information alone. Additionally, Plaid noted that it facilitated a 20% relative risk reduction for some subprime and near-prime borrowers without decreasing originations.
Furthermore, Plaid LendScore aims to bring the tools to assess creditworthiness with more precision, efficiency, and context to lenders. This is achieved through insights into financial behaviour that optimise risk prediction, Open Banking data for holistic financial visibility, and simplified user experiences that support conversion.
With Open Banking continuing to develop and expand across the US, which will lead to credit scoring getting smarter, Plaid intends to continue to improve its model to reflect how borrower behaviour changes and to further support lenders.
For more information about Plaid, please check out their detailed profile in our dedicated, industry-specific Company Database.
