Credit Risk Analytics

Given the on-going turmoil on credit markets, a critical re-assessment of current capital and credit risk modelling approaches is more than ever needed. Professor Baesens’ research group endeavours to come up with new approaches for better credit risk modelling.

Our research team currently focuses on the following topics in credit risk analytics:

  • Novel techniques for credit risk model calibration, validation, monitoring
  • Developing credit risk models for microfinance
  • Using survival analysis techniques for credit risk modeling
  • Studying the impact of data quality on credit risk modeling

Developing optimal credit risk model calibration procedures (PD / LGD / EAD)

Credit risk models aim to discriminate obligors and/or exposures in terms of default (PD), loss (LGD) and exposure (EAD) risk. However, losses need to be absorbed by capital in an absolute way! Hence, we aim to develop calibration procedures that come up with optimal cardinal measures of risk taking into account both past experience and future expectations. Therefore we need to study which statistical techniques work well on a time series of historical credit risk data and how survival analysis can be used to work with different time horizons (see Survival Analysis below). Furthermore, we will investigate the impact of both point-in-time (PIT) and through-the-cycle (TTC) calibration.

Using survival analysis techniques for credit risk modeling

Survival analysis is often associated with the medical field, particularly in treatment trials where researchers are predicting how long a patient will survive or continue in remission, considering a certain treatment plan. More generally, survival analysis models predict how long before a particular event occurs. In many domains, where simple classification has been used to predict the occurrence of an event, survival analysis can be used to add the valuable dimension of time until that occurrence.

Credit scoring generally falls into two categories: application and behavior scoring. The first uses the customer data from the time of application as well as information about the loan to predict whether the customer will default on the loan. After a loan is granted, the creditor begins gathering information about the customer’s behavior, such as payments or spending. Survival analysis models incorporate this behavior data into a model to predict not only if a particular account will default, but how long before that happens. This time component allows financial institutions to prepare for upcoming conditions, predicting potential losses more accurately and allowing better use of capital. In some cases, a default after a certain period is still a profitable account, so the traditional good/bad classification is insufficient in terms of financial planning. Therefore we need to study which statistical techniques work well on a time series of historical credit risk data and how survival analysis can be used to work with different time horizons.

Devising new frameworks for monitoring and validating credit risk parameters (PD / LGD / EAD)

New approaches for monitoring and validating the economic capital model parameters have to be devised. Theoretical contributions will necessitate new approaches for model backtesting, model benchmarking and qualitative validation of the various economic capital parameters. Besides, we must find out how to correctly measure data quality and we must set up data quality programs (see Data Quality below). In addition, it is necessary to investigate how to quantify model performance, model comprehensibility and model readability at the model level.

Introducing innovative economic capital stress testing procedures

New approaches for stress testing the economic capital model are necessary. Using stress testing, economic capital is affected by extreme, adverse economic circumstances, similar to the recent events in the credit markets. The research will focus on defining stress scenarios, integrating various types of risk, understanding the behaviour in the tail of the loss distribution, and gauging the impact on credit risk and capital positions at the firm level. It will be required to define solid, financial theories in case of limited historical data.

Developing new ways to quantify and deal with model risk

Risk quantification models are not perfect and as such also have risk associated with them. It is important to be aware of this and to appropriately act upon it, using e.g. conservative parameter calibration and/or other novel approaches. Therefore, we endeavour to define model risk based on data quality and model limitations and to relate economic downturn calibration to stress testing and study the implications.

Developing credit risk models for microfinance

Due to growing competition, over-indebtedness, and economic crises, microfinance institutions have to pursue their social and financial objectives in an increasingly constrained environment. Using powerful risk management tools, therefore, becomes more than ever a key competence to survive. It is in this context that established techniques from traditional financial organizations are introduced to the microfinance industry, with the aim to improve both social impact and financial efficiency.

Studying the impact of data quality on credit risk modeling

Recent studies have indicated that companies are increasingly experiencing data quality related problems as more and more complex data are being collected. In order to address such problems, the literature suggests the implementation of a Total Data Quality (TQM) management program that should consist of the following phases: data quality definition, measurement, analysis and improvement. Data quality is often defined as “fitness for use”. Although “fitness for use” captures the essence of quality, it is difficult to measure data quality using this broad definition. Thus, it has long been acknowledged that the quality of data is best described or analyzed via multiple attributes or dimensions.

Despite broad discussion in the data quality literature, there is no one definite set and exact definition of data quality dimensions because data quality is context dependent. Therefore, data quality dimensions should be identified and defined in relation to tasks to achieve a suitable level of data quality.

Our research identifies important data quality dimensions for evaluating the quality of the data for credit risk assessment. We also explore the key data quality challenges and causes of data quality problems in financial institutions, based on statistical analysis.

Further Reading

PhD Theses

Selected Publications

  • Dirick L, Claeskens G., Baesens B., An Akaike information criterion for multiple event mixture cure models, European Journal of Operational Research, Volume 24, pp. 449-457, 2015.
  • Tobback E., Martens D., Van Gestel T., Baesens B., Forecasting loss given default models: impact of account characteristics and the macroeconomic state, Journal of the Operational Research Society, Volume 65, Number 3, pp. 376-392, 2014.
  • Louis P., Baesens B., Do for-profit microfinance institutions achieve better financial efficiency and social impact?, Journal of Development Effectiveness, Volume 5, Number 3, pp. 359-380, 2013.
  • Louis P., Van Laere E., Baesens B., Understanding and predicting bank rating transitions using optimal survival analysis models, Economics Letters, Volume 119, Number 3, pp. 280-283, 2013.
  • Louis P., Seret A., Baesens B., Financial Efficiency and Social Impact of Microfinance Institutions using Self-Organizing Maps, World Development, Volume 45, pp. 197-210, 2013.
  • Berteloot K., Verbeke W., Castermans G., Van Gestel T., Martens D., Baesens B., A Novel Credit Rating Migration Modeling Approach using Macroeconomic
    Indicators, Journal of Forecasting, Volume 32, Issue 7, pp. 654–672, 2013.
  • Van Gool, J., Verbeke, W., Sercu, P., Baesens, B. (2012). Credit Scoring For Microfinance – Is It Worth It?. International Journal Of Finance & Economics, 17 (2), 102-123.
  • Martens D., Van Gestel T., De Backer M., Haesen R., Vanthienen J., Baesens B., Credit Rating Prediction Using Ant Colony Optimization, Journal Of The Operational Research Society, Vol. 61, Pp. 561-573, 2010
  • Setiono R. , Baesens B., Mues C., A Note On Knowledge Discovery Using Neural Networks And Its Application To Credit Card Screening, European Journal Of Operational Research, Vol. 192, Number 1, Pp. 326-332, 2009
  • Van Gestel T., Martens D., Baesens B., Feremans D., Huysmans J., Vanthienen J., Forecasting And Analyzing Insurance Companies’ Ratings, International Journal Of Forecasting, Vol. 23, Number 3, Pp. 513-529, 2007
  • Louis, P., Van Laere, E., Baesens, B. (2011). Motivating And Predicting Bank Rating Transitions Using Optimal Survival Analysis Models. Proceedings Of The 24th Australasian Finance & Banking Conference: Vol. Accepted. Australasian Finance & Banking Conference. Sydney (australia), 14-16 December 2011.
  • Louis, P., Van Laere, E., Baesens, B. (2011). Predicting Bank Rating Transitions Using Optimal Competing Risks Survival Analysis Models. Proceedings Of The Credit Scoring And Credit Control Xii Conference: Vol. Accepted. Credit Scoring And Credit Control Conference. Edinburgh (uk), 24-26 August 2011.
  • Baesens, B., Van Gestel, T., Stepanova, M., Vanthienen, J. (2003). Neural Network Survival Analysis For Personal Loan Data. Proceedings Of The Eighth Conference On Credit Scoring And Credit Control (csccvii’2003). Conference On Credit Scoring And Credit Control (csccvii’2003). Edinburgh (scottland).
  • Baesens, B., Van Gestel, T., Stepanova, M., Vanthienen, J., Van Den Poel, D. (2005). Neural Network Survival Analysis For Personal Loan Data. Journal Of The Operational Research Society, 56 (9 (sept.)), 1089-1098.
  • Moges, H., Dejaeger, K., Lemahieu, W., Baesens, B. (2012). A Multidimensional Analysis Of Data Quality For Credit Risk Management: New Insights And Challenges. Information & Management.
  • Moges, H., Dejaeger, K., Lemahieu, W., Baesens, B. (2012). A Total Data Quality Management For Credit Risk: New Insights And Challenges. International Journal Of Information Quality, 3 (1), 1-27.
  • Van Laere E., Baesens B., The Development Of A Simple And Intuitive Rating System Under Solvency Insurance: Mathematics And Economics, Forthcoming 2010
  • Martens D., Baesens B., Van Gestel T., Vanthienen J., Comprehensible Credit Scoring Models Using Rule Extraction From Support Vector Machines, European Journal Of Operational Research, Vol. 183, Pp. 1466-1476, 2007
  • Hoffmann F., Baesens B., Mues C., Van Gestel T., Vanthienen J., Inferring Descriptive And Approximate Fuzzy Rules For Credit Scoring Using Evolutionary Algorithms, European Journal Of Operational Research, Vol. 177, Number 1, Pp. 540-555, 2006
  • Van Gestel T., Baesens B., Van Dijcke P., Suykens J., Garcia J., Alderweireld T., Linear And Nonlinear Credit Scoring By Combining Logistic Regression And Support Vector Machines, Journal Of Credit Risk, Vol. 1, Number 4, 2005
  • Van Gestel T., Baesens B., Van Dijcke P., Garcia J., Suykens J.a.k., Vanthienen J., A Process Model To Develop An Internal Rating System: Sovereign Credit Ratings, Decision Support Systems, Vol. 42, Number 2, Pp. 1131-1151, 2006
  • Huysmans J., Baesens B., Van Gestel T., Vanthienen J., Failure Prediction With Self Organizing Maps ,expert Systems With Applications, Vol. 30, Number 3, Pp. 479-487, 2006
  • Somol P., Baesens B., Pudil P., Vanthienen J., Filter-versus Wrapper-based Feature Selection For Credit Scoring, International Journal Of Intelligent Systems, Vol. 20, Number 10, Pp. 985-999, 2005
  • Baesens B., Van Gestel T., Mues C., Vanthienen J., Intelligent Information Systems For Financial Engineering, Expert Systems With Applications, Vol. 30, Number 3, Pp. 413-414, 2006
  • Van Gestel T., Baesens B., Suykens J.a.k., Van Den Poel D., Baestaens D.-e., Willekens M., Kernel Based Classification For Financial Distress Detection, European Journal Of Operational Research, Vol. 172, Number 3, Pp. 979-1003, 2006
  • Baesens B., Van Gestel T., Stepanova M., Van Den Poel D., Vanthienen J., Neural Network Survival Analysis For Personal Loan Data, Journal Of The Operational Research Society, Vol. 59, Number 9, Pp. 1089-1098, 2005
  • Mues C., Baesens B., Files C.m., Vanthienen J., Decision Diagrams In Machine Learning: An Empirical Study On Real-life Credit-risk Data, Expert Systems With Applications, Vol. 27, Number 2, Pp. 257-264, 2004
  • Baesens B., Setiono R., Mues C., Vanthienen J., Using Neural Network Rule Extraction And Decision Tables For Credit-risk Evaluation, Management Science, Vol. 49, Number 3, Pp. 312-329, 2003
  • Van Gestel T., Baesens B., Garcia J., Van Dijcke P., A Support Vector Machine Approach To Credit Scoring, Bank En Financiewezen, Vol. 2, Pp. 73-82, 2003
  • Lima E., Mues C., Baesens B., Monitoring And Backtesting Churn Models, Expert Systems With Applications, 2010
  • Van Gestel T., Martens D., Baesens B., From Linear To Non-linear Kernel Based Classifiers For Bankruptcy Prediction, Neurocomputing, 2010