Kevin M Lowery

Kevin is a Director in KPMG’s Modeling, Credit, and Liquidity team. He has over a decade of modeling experience, with a primary focus on model risk management. He assists his clients in validating models across the typical banking model inventory using both on-shore and off-shore staffing structures and enhancing their model risk management programs. In recent years, he has helped his banking clients deal with emerging challenges related to the risk management of Machine Learning and Generative AI models.

Professional and industry experience

• The use of machine learning (ML) and artificial intelligence (AI) has increased exponentially at financial institutions over the last five years. As model inventories have been filled with AI/ML algorithms, model risk and audit departments have struggled to deal with the new risks and challenges they pose. I have led multiple AI/ML reviews for my clients, identifying where model validation practices may be lacking relative to what my client’s peers are doing and where regulators have expressed concern so far.
• Stress testing has become a central driver of evolving standards for advanced risk management. I have led model validations and audits for all of the major drivers of CCAR/DFAST results, such as Retail credit, Wholesale credit, and Pre-Provision Net Revenue (PPNR). These validations have, at times, revealed critical errors resulting in >$100 million in capital impact and in others have revealed significant gaps to common industry practices that expose my clients to regulatory risk.
• The Current Expected Credit Loss (CECL) exercise has posed one of the most significant modeling challenges financial institutions have experienced in the last decade. As a member of a Big 4 audit firm, I have been able to keep my clients abreast of industry practices and accounting expectations related to CECL modeling. In addition to standard model testing, the CECL validations I lead utilize benchmarks obtained from our audit work which can be used to sense-check my client’s model output.
• Treasury departments rely on asset liability management (ALM) models to drive strategic decision making and assess their risks. These models are typically some of the most complex in a Bank’s model inventory – including the ALM system itself, interest rate models, behavioral models, and credit models at the very least. I have led multiple validations for both of the major ALM systems (QRM and PolyPaths) at my clients, including the key sub-models.