The most common approach that financial examiners and researchers use for understanding bank earnings is comparing allowances for uncollectible loans (loan loss provision) with changes in loans that are of bad quality (non-performing loans).

The common assumption in this comparison is that there is a linear relationship between loan loss provision and changes in non-performing loans. However, Wei Wang, Sudipta Basu and Justin Vitanza discovered that the connection between loan loss provisions and non-performing loans is not linear, but V-shaped.