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.

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.

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.

Sorry for the delay reporting this but better late than never!

On May 7 the Federal Reserve System published its report on stress tests on 19 major US banks.  They used economic scenarios, comparing a “baseline” state assuming the economy follows their predicted trend with a “more adverse” state where the economy performs worse than expected.  The result is that they expect the 19 banks to accrue debts of $600billion during 2009 and 2010 assuming adverse economic conditions.  Of this, $455billion would be for consumer debt.  This underlines the importance of understanding risk within consumer credit portfolios.  Research in this area lags behind corporate credit, but consumer credit accounts for much more of bank risk.  The bottom line is that between them the banks are short of capital requirements by $75billion.

It is also worth reading the BBC report on this stress test.

Several US financial commentators have dismissed these stress tests as too weak and it has been done simply to allay the fears of the public and investors in the US banking system.  The conditions used by the Fed to represent “more adverse” conditions were not sufficiently adverse and some of the indicators, such as unemployment rate, are already close to the value chosen for adveerse conditions (ie adverse is normal!).  For further details, Professor Campbell Harvey at Duke University (NC)  gives a detailed criticism.

I always enjoy reading Bryan Magee’s works on philosophy because he writes so clearly. Here are some remarkable interviews he conducted with some of the great modern philosophers on a variety of subjects such as Wittgenstein, logical positivism and Frege. I haven’t watched all of them yet but I’ve enjoyed what I’ve seen so far. You just don’t get TV like this any more!
http://www.youtube.com/results?search_type=&search_query=bryan+magee&aq=f

I am pleased to announce that my wife Xijuan gave birth to our baby daughter on Wednesday 25 February 2009 at 1745. She weighed 7lb 13oz with a full mop of black hair! Please follow the link to see some early pictures:-

Photos of one day old Alyssa ruò cún (若存) Bellotti

Against the Gods by Peter Bernstein

I thoroughly enjoyed this book. It is a history of risk management from ancient Greek thought, through Bernoulli, to modern quants innovators such as Markowitz and behavioural finance analysts such as Khaneman and Tversky. The style is very easy with no mathematics. Its an outline of the problems in risk management and the people involved.

Bernstein argues that many of the risk management innovations are essential to our development. “The central theme of this whole story is that the quantitative achievements … shaped the trajectory of progress over the past 450 years”. Nevertheless he gives the following warning regarding modern practice (p.336):

Nothing is more soothing or more persuasive than the computer screen, with its imposing arrays of numbers, glowing colors, and elegantly structured graphs. As we stare at the passing show, we become so absorbed that we tend to forget that the computer only answers questions; it does not ask them. Whenever we ignore that truth, the computer supports us in our conceptual errors. Those who live only by the numbers may find that the computer has simply replaced the oracles to whom people resorted in ancient times for guidance in risk management and decision-making.

One common assumption is in using past data to build models:

We cannot enter data about the future into the computer because such data are inaccessible to us. So we pour in data from the past to fuel the decision-making mechanisms created by our models … past data from real life constitute a sequence of events rather than a set of independent observations, which is what the law of probability demands.

This is not to suggest we should not do this, but that we are careful when we do and that we ensure our models are robust. As Bernstein concludes

At the same time, we must avoid rejecting numbers when they show more promise of accuracy than intuition and hunch.

Ultimately, without risk management methods we would not have the tools to manage our increasingly complex financial decisions.

Our paper has been accepted by the Journal of the Operational Research Society (JORS) for publication. It can be viewed online:

Credit scoring with macroeconomic variables using survival analysis

Bellotti T and Crook J

Abstract

Survival analysis can be applied to build models for time to default on debt. In this paper, we report an application of survival analysis to model default on a large data set of credit card accounts. We explore the hypothesis that probability of default (PD) is affected by general conditions in the economy over time. These macroeconomic variables (MVs) cannot readily be included in logistic regression models. However, survival analysis provides a framework for their inclusion as time-varying covariates. Various MVs, such as interest rate and unemployment rate, are included in the analysis. We show that inclusion of these indicators improves model fit and affects PD yielding a modest improvement in predictions of default on an independent test set.

A highly recommended music site… but be warned, only if you’re a fan of ambient!

Our wedding day

XiJuan and I married on Saturday 11th October, a sunny day in Cambridge (UK) surrounded by friends and family. We now have the wedding photos for view. I’ve uploaded them to Flickr so just follow this link to view!

Wedding photos