LGD workshop

I will be attending a LGD workshop organized by Marfintel in Belgium on 8 October to present practical work and methods for estimating LGD with macroeconomic conditions.

Building models of consumer LGD with macroeconomic data to improve estimation and enable stress testing: (1) How to include macroeconomic variables in models of LGD, (2)Evidence showing improved estimation of LGD at account and aggregate levels, (3) Using the models for stress testing and estimation of downturn LGD.

Other speakers are Prof Bart Baesens and Dr David Martens.

Computational and Financial Econometrics conference 2009

I will be attending CFE 2009 in Cyprus and will present the following paper.

An exercise in stress testing for retail credit cards
Tony Bellotti and Jonathan Crook

Abstract

Stress tests are becoming increasingly important for evaluating consumer credit risk and they are recognised as a key tool in helping financial institutions make business strategy, risk management and capital planning decisions. We present a simulation-based method for stress testing using discrete survival analysis to build a dynamic model of default at the account level which includes macroeconomic conditions as risk factors. We discuss methods to generate plausible but extreme economic simulations within this framework and apply Monte Carlo simulation to compute empirical distributions of estimated default rates. We present experimental results for a large data set of UK credit card accounts that show that bank interest rates, production index, retails sales index and the FTSE 100 index are all statistically significant systemic risk factors for default. Stress tests using this model generate right skewed distributions of default rates and yield plausible extreme values of risk as measured by Value at Risk and expected shortfall. Finally we discuss statistical validation of stress tests through back-test of the loss distribution.

CSCCXI conference

The Credit Scoring and Credit Control XI Conference will take place in Edinburgh on 26-28 August 2009.   This conference bridges the academic-practitioner divide with a range of talks regarding current industry issues (like Basel II) to the latest statistical research findings.   It provides an ongoing forum to debate theory versus practice as well as for meeting up with old – and new – colleagues.

I will be presenting two papers outlined below.

Dynamic Models of Default on UK Credit Cards
Tony Bellotti and Jonathan Crook

Abstract

Typically models of default are built on static data, usually collected at time of application.  We consider alternative models that also include behavioural data about credit card holders and macroeconomic conditions across the credit card lifetime, using a discrete survival analysis framework.  We find that models that include these dynamic variables give statistically significant improvements in model fit which translate into better forecasts of default at both account and portfolio level when applied to an out-of-sample data set.  Additionally, by simulating extreme economic conditions, we show how these models can be used to stress test credit card portfolios.

Macroeconomic conditions in models of Loss Given Default for retail credit

Tony Bellotti & Jonathan Crook

Abstract

Loss Given Default is an important measure of credit loss used by financial institutions to compute risk within credit portfolios, expected loss on individual loans and capital requirements.  We investigate models of Loss Given Default for UK retail credit cards which incorporate macroeconomic conditions.  We find bank interest rates and unemployment rate are important explanatory variables and their inclusion improves forecasts of Loss Given Default for hold-out data sets.  Additionally, the inclusion of macroeconomic conditions is important since it enables stress testing and provides a means to model downturn Loss Given Default as required by the Basel II Accord for the advanced internal ratings based approach to calculating capital requirements.