Pub. 6 2016 Issue 1

Challenges Chief executive officers (CEOs) of na- tional banks received The Supervisory Guidance on Model Risk Management document in 2011 from the Federal Reserve and OCC (Supervisory, 2011). Banks created or enhanced their model risk management groups, but model validations still failed examinations. Model validation challenges for banks include 1) knowledge gaps, 2) limited resources, 3) inconsistent line-of-busi- ness development processes, and 4) proper documentation. Identifying and obtaining resources that contain the experience and knowledge to connect the dots between business units, compli- ance, legal, technology, and modeling groups is difficult. Inconsistent docu- mentation across an organization's functional units requires additional resources and time to validate models and resolve audit issues. Model validation documentation presents other problems. First, inter- preting guidelines requires resources that can decipher and communicate the guidelines to leadership, compliance, modeling, and line of business teams. Second, modelers are normally not the optimal choice to write validation documents. Modelers have a tendency to write documentation for statisticians. Finally, leaders may prefer a particular documentation framework that does not pass a regulatory exam. Conclusion Model validations can appear to be a daunting, time-consuming task. However, the necessary information to create the validation documents should be readily available. First, technology should have documenta- tion that includes the data sources, controls, processes, and service level agreements. Next, internal model development should follow documented model design, development, testing, and deployment processes. Finally, results from judgmental and model driven decisions should be available to under- stand the business impact. If the above information is available, the validation time decreases. If the information is not available, the model validation docu- ments can be used to reduce the bank’s information gaps. The good news is that after the first model validation is complete, leveraging the validation process and documenta- tion framework will reduce validation time and resources. The reduced time will allow banks to expedite model validations and use the valuable results and recommendations as inputs into profitable business strategies. Dr. Eric Golla is an Innovation and Data Science advisor for TechMile- age. TechMileage is a Phoenix Metro Area, AZ-based software develop- ment, advanced analytics and model validation company with extensive experience in developing enterprise solutions and products. TechMileage specializes in providing end-to-end software solutions, modeling, model validations, and business intelligence. For more information, contact, Rajesh Kumar at (602) 334-9964 or rajesh. kumar@techmileage.com . Burns, R. L. (2005). Supervisory Insights. Retrieved from https://www. fdic.gov/regulations/examinations/ supervisory/insights/siwin05/article01_ model_governance.html External Audit (June 24, 1996). Statement of Policy Regarding Inde- pendent External Auditing Programs of State NonMember Banks. Re- trieved from http://www.azdfi.gov/ LawsRulesPolicy/SPS%20Links/ DFI-AD-PO-BA-SUB_POL_STATE- FDIC_Attachment_BA-1-062496ver1. pdf Examination, (n.d.). Bank Examina- tions. Retrieved from http://azdfi.gov/ Licensing/Licensing-FinInst/Banks/ BanksExamination.html. Supervisory. (April 4, 2011). Su- pervisory Guidance on Model Risk Management. Retrieved from http:// www.federalreserve.gov/bankinforeg/ srletters/sr1107a1.pdf. w MODEL VALIDATIONS SHOULD IDENTIFY ISSUES AND PROVIDE RECOMMENDATIONS THAT CAN RESOLVE COMPLIANCE ISSUES, ENHANCE MODELS, AND CREATE BEST PRACTICES. 19 ISSUE 1 . 2016