Every time an institution extends a loan, it faces credit risk. It is the risk of economic loss when an obligor does not fulfill the terms and conditions of his contracts. Measuring and managing credit risk is imperative to financial organisations as this information exposes the credit worthiness of the borrowers and help banks lower the risk of default.
Over the last decade, a number of the world’s largest banks have developed sophisticated systems in an attempt to model the credit risk arising from important aspects of their business lines. Financial institutions make use of vast amounts of data on borrowers and loans and apply these predictive and analytical models. Such models are intended to aid banks in quantifying, aggregating and managing risk across geographical and product lines.
The outputs of these models also play increasingly important roles in banks’ risk management and performance measurement processes, including performance-based compensation, customer profitability analysis, risk-based pricing and, active portfolio management and capital structure decisions.
In this course, our objective is to learn how to build these credit risk models. While credit risk arises in almost all business lines for a bank, our focus will be on the credit risk involved in the personal and corporate loans which is of major importance to banks.
We will learn credit risk modeling using case studies. Specifically, we will use two case studies starting with a simpler one using which we will learn the methodology and important concepts and techniques.
Case Study 1: German Credit
In the first case study, we will use a popular dataset called German Credit. Our objective in this case study is to determine the Probability of Default (PD). We will build a predictive model that takes as input the various aspects of the loan applicant and outputs the probability of default of the loan applicant. PD is one of the most highly used measures for calculating the credit score of the borrowers. PD is also the primary parameter used in calculating credit risk as per the internal ratings-based approach used by banks.
The German Credit dataset contains observations on 21 attributes for 1000 past applicants for credit. Each applicant was rated as “good credit” (700 cases) or “bad credit” (300 cases).
In this case study, we will perform all the steps involved in model building and along the way, we will also understand the entire spectrum of the predictive modeling landscape.
Case Study 2: LendingClub
In the second case study, we will build upon the knowledge we have gained in the first case study and apply it to a new data set which is more realistic in nature. We will use the loan data available from LendingClub's website. LendingClub is a US peer-to-peer lending company which matches borrowers with investors willing to fund their loans. The loan dataset contains actual data of the loans extended by them in their business. The dataset is much larger in size compared to the German Credit data and also contains a lot more variables that we need to work on. This case study will give us a more real-life experience of what we can expect when we build a model in our role as a data scientist in a bank.