Risk Management in the Personal Financial Services Sector

Lead Research Organisation: Imperial College London
Department Name: Dept of Mathematics


The personal financial services sector is concerned with mortgages, bank loans, credit cards, bank accounts, and related products. It is an area of increasing concern to the Government, because personal debt in the UK has reached unprecedented levels, putting many people in a position which makes them vulnerable to changes in personal circumstances and changes in the economy (the concern about a possible collapse in house prices is just one example of this). It is also, obviously, of central importance to individuals, and to the financial bodies concerned. Over the last 30 years there have been dramatic advances in the sorts of tools used to model risk in the financial markets, but quite different kinds of tools are needed in the personal financial services sector. The aim of this application is to stimulate the development of such tools by: (i) carrying out three core projects which develop novel kinds of tools; (ii) carrying out additional concurrent projects, funded from other sources; (iii) organising a workshop programme which will enable others, not involved in the three core projects, to take part; (iv) organising a visitor programme, allowing other researchers to visit the core sites and other relevant institutionsThe three core projects are:1) Current approaches to risk assessment and management in the retail sector use models which seek to predict likely outcome based on observed characteristics and previous behaviour. In fact, however, the aim is to use the model to guide appropriate choice of action (e.g. extend a loan repayment period; send a warning letter, etc.). Actions, of course, mean that the predictive data are no longer appropriate. We aim to build models, based on how customers behave, to predict the effect of such interventions.2) Current models for credit risk or portfolios of personal loans are based on variants of the models used for corporate loans. However, such models are inappropriate for personal loan portfolios (where, for example, cash flow is more important than det/asset ratio). Three approaches to building appropriate models will be explored, based on modifying the assumptions of structural models used in the corporate sector, building equivalents of the reduced form models used in the corporate sector, and using competing risk models to predict likely consumer default.3) The literature on how changes in the state of the economy influence consumer credit behaviour is limited. This project will explore various aspects of this, including incorporating time-varying economic factors into survival models, adjusting the default probability cutoff as economic conditions change, and including economic factors in early repayment models.


10 25 50
publication icon
Andreeva G (2007) Modelling profitability using survival combination scores in European Journal of Operational Research

publication icon
Bellotti T (2017) Credit scoring with macroeconomic variables using survival analysis in Journal of the Operational Research Society

publication icon
Brentnall A (2008) A statistical model for the temporal pattern of individual automated teller machine withdrawals in Journal of the Royal Statistical Society: Series C (Applied Statistics)

publication icon
Brentnall A (2010) Predicting the amount individuals withdraw at cash machines using a random effects multinomial model in Statistical Modelling: An International Journal

publication icon
Malik M (2017) Modelling credit risk of portfolio of consumer loans in Journal of the Operational Research Society