Graduate School of Business Seminer / UGUR KOYLUOGLU

Ugur Koyluoglu,  Oliver Wyman

Title: 50 years of corporate credit risk analytics From rule of thumb to statistical learning

Date: June 27, Wednesday
Time:  10:00
Location: Altunizade Yerleşkesi, 1 nolu Toplantı Salonu

Abstract:

The first model on predicting corporate bankruptcy was published in 1968. It was quite simple - with just five financial ratios. A number of structured models based on financial and qualitative models were developed by practitioners in the following years, and most banks started to use these models to assess riskiness of corporate clients, estimate capital adequacy, establish reserves and reflect model output onto risk-based costing of loans. About 30 years into banking industry-wide use of probability of default (PD), loss given default (LGD), exposure at default (EAD) models, we are now on the cusp of a new breed of credit risk models that incorporate signals from unstructured data such as announcements, news, social media, satellite, balloon, drone data, and create better and scalable risk measures to support commercial credit decisions. This is prompted by continued advancements in computing power, statistical learning methodologies and natural language processing, all of which have enabled the mining of rich alternative content sets and accelerated large-scale data analysis.In this seminar, we cover applications of statistical learning in commercial credit, covering wholesale lending, corporate, middle market and small and medium enterprise segments. We will start with the basics for credit risk analytics and quickly move to discuss leading practices where the new breed of credit risk models beat traditional approaches in multiple ways:

• They extract more out of standard data sources such as firm financials, for instance, by uncovering nonlinear relationships effectively·        
• They make use of more and varied data, such as unstructured text data found in news and social media·        
• They can incorporate data that is updated in real time, allowing analysis to be much more timely, and avoiding stale signals.

The initial benefits of using advanced statistical learning models include increased Gini coefficients, richer credit risk dashboards, faster and more comprehensive warning signals; and  promise a profound impact for credit underwriting, leading to fewer unprofitable customers, better differentiation between marginal credit cases, as well as for portfolio monitoring, resulting in more timely detection of credit deterioration and potentially more cost effective hedging. However, they also come with certain challenges and potential pitfalls, and testing cultural readiness.

The presentation covers concepts and real examples. No pre-requisite is needed.

Bio:

Ugur Koyluoglu is a Partner with Oliver Wyman, based in New York. He is currently Vice Chair for Financial Services, leading content development.  Prior to this role, Ugur led Americas Finance & Risk, Public Policy, and Corporate & Institutional Banking practices, and continues to steer the firm’s high-impact, corporate strategy, finance and risk management projects for financial institutions across the globe.In his recent client work, Ugur focuses on the best use of bank’s resources, aligned with strategy, within the risk appetite and regulatory constraints for capital, liquidity, leverage and collateral; and development of innovative models for corporate credit risk analytics, operational deposits, farm lending, and impact of environmental change on lending portfolios.In addition to project work, Ugur has contributed significantly to industry debates in risk management and published several articles. His most cited article, “Reconcilable Differences” with Andrew Hickman, demonstrates that mathematically different credit risk portfolio modeling approaches can actually be mapped onto a common framework, and ingenuity therefore lies in the best calibration of credit risk parameters rather than choice of modeling approach. This important finding was incorporated into Basel II, the capital requirements framework used by banking regulators worldwide, and is viewed as a breakthrough concept in credit risk management practices. The paper is also published as independent chapter in four books.Ugur has a PhD in Civil Engineering and Operations Research from Princeton University, and taught applied mathematics and engineering before coming to Oliver Wyman in 1997.