For this research study, CRA consultants Marsha Courchane and Arthur Baines analyzed the use of various types of cash flow data in highly automated underwriting systems utilized by six financial services institutions which offer and originate consumer and small business loans across a broad set of geographies in the US. The use of the various types of cash flow data, in conjunction with, or in lieu of, more traditional credit bureau derived data has been used to underwrite credit for both consumer loans and small business loans.
We find compelling evidence that indicates that among the sample populations and products CRA analyzed, the cash flow data are predictive of credit risk and loan performance across the highly heterogeneous set of participants. In our separate analyses of each participant, the results appear to be robust across both consumer and small business populations as well as across the credit spectrum, including among borrowers with no, or very low, traditional credit scores, some of which may reflect ‘no-file’ or ‘thin-file’ borrowers. Among the sample populations and products, the cash flow data and traditional credit data, when analyzed, displayed some degree of asymmetric information, and the cash flow data frequently improved the sorting of risk among borrowers posing similar credit risks, as measured by the traditional credit data.
FinRegLab engaged Charles River Associates to conduct analyses of the use of cash flow data by participating financial services institutions in highly automated underwriting models of credit applications and loan originations. The intent is to undertake a quantitative analysis of important questions raised by the increased use of cash flow data in the market for consumer and small business loans.
COVID-19 crisis – Mortgage forbearance considerations
During the current COVID-19 crisis, many financial institutions have responded by tightening their lending standards while increasing allowable forbearance...