Consumer finance: challenges for operational research

Abstract

: Consumer finance has become one of the most important areas of banking both because of the amount of money being lent and the impact of such credit on the global economy and the realisation that the credit crunch of 2008 was partly due to incorrect modelling of the risks in such lending. This paper reviews the development of credit scoring,-the way of assessing risk in consumer finance-and what is meant by a credit score. It then outlines ten challenges for Operational Research to support modelling in consumer finance. Some of these involve developing more robust risk assessment systems while others are to expand the use of such modelling to deal with the current objectives of lenders and the new decisions they have to make in consumer finance.

Using semi-supervised classifiers for credit scoring

“…Consider payday loans which are typically small, short duration (less than one month) with extremely high interest rates. It is necessary to construct scorecards that can respond in a timely fashion to shifts in economic and market behaviour, as well as to sudden changes in the borrower's circumstances and behaviour (Thomas, 2009). Clearly one-class classification is suited to such tasks.…”

“…Another problem is authors' expertise in their own method and failure to undertake a corresponding effort with existing methods (Michie et al, 1994). Indeed, Thomas (2009) highlights that studies which have endeavoured to avoid the aforementioned problems (Baesens et al, 2003;Xiao et al, 2006) have reported that the differences between the performance of classification techniques were small and regularly not statistically significant. Great care and consideration was taken to avoid these issues in this work, details of which are given in Section 4.…”

“…The sample sizes vary considerably from 125 to 5,397 instances. Although commercially used scorecards are usually constructed from an initial sample size of between 10,000 and 50,000 (Thomas, 2009), the above data sets are all easily available in public literature whereas many other data sets used in credit scoring studies are privately held and cannot be shared amongst researchers. As per Keogh (2007), we believe that the irreproducibility of results caused by, amongst other things, the refusal to share data or to give parameter settings hinders the research process.…”

“…The upheaval in the financial markets that accompanied the sub-prime mortgage crisis has emphasised the large proportion of the banking industry based on consumer lending (Thomas, 2009). Credit scoring is an important part of the consumer lending process that attempts to predict the repayment behaviour of borrowers.…”

“…It is influenced, amongst other factors, by the expertise of the user with each of the methods used and the effort invested in refining and optimising each method (Hand and Zhou, 2009;Thomas, 2009). We attempt to overcome this problem by restricting our study to a single application area (low-default portfolios); by selecting appropriate performance measures (H measure and harmonic mean); and finally by using nine different data sets of varying size and dimension to capture as many as possible of the particular aspects of the LDP problem.…”