Even though rare, bank fraud still happens. Among the 4 bank failures happened in 2019, one of them, the Enloe State Bank, was caused by fraud.
Commissioner Cooper of Texas stated “we are forced to close this bank which was chartered in 1928 due to insider abuse and fraud by former officers.” https://www.dob.texas.gov/public/uploads/files/news/press-releases/2019/05-31-19pr.pdf
Back to December 2017, after a quiet year of 2018 with 0 bank closure, the Washington Federal Bank for Savings in Chicago, Illinois was also closed due to massive fraud.
According to the reported issued by the Treasury Department, employee and the bank's then-president had been regularly falsifying loan payments for at least 29 loans totaling approximately $68 million in aggregate assets, and had falsified Washington Federal's loan trial balance before providing it to examiners.” https://www.treasury.gov/about/organizational-structure/ig/Audit%20Reports%20and%20Testimonies/OIG-19-009.pdf
“As of September 30, 2017, Washington Federal Bank for Savings had total assets of $166.3 million and total deposits of $144 million, of which there were approximately $11.6 million that exceeded FDIC insurance limits.”
Comparing the bank’s total assets of $166 million with the aggregated fraud amount of $68 million, it can be seen that the effect of the fraud is pervasive.
For bank fraud cases, the statistical model is less effective. Because statistical models are data driven, when cooked data is used as input, the model output will inevitably be contaminated.