BQM&T
3,1
38
Information systems and
benchmarking in the credit
scoring industry
Kevin J. Leonard
School of Business and Economics, Wilfrid Laurier University,
Waterloo, Ontario, Canada
Introduction
In the current study of management practices, it is becoming accepted that
information systems (IS) theory and total quality management (TQM) theory
are very much integrated. On one level, in the development of information
systems, many TQM principles hold. For instance, user participation is critical
in systems development. A comparable component of TQM is the fact that it is
imperative to know and satisfy the demands of the customer. Doing it right the
first time is preached by leaders in both TQM and IS as a method of getting
users/customers over the critical implementation period and in a position to
adopt the new technology willingly. On a second level, TQM principles often
need an information system to become feasible, owing to the overwhelming
need to process a wealth of data almost instantly. A good example is in the area
of continuous improvement and benchmarking, where many inter-firm and
intra-firm comparisons are necessary to update reports and potentially modify
objectives[1].
Banks must evaluate the ability of their customers to repay their financial
obligations according to the agreement established between the respective
parties. This evaluation process can be carried out using credit scoring models.
(Credit evaluation can be done in a number of ways, the emphasis here, however,
is solely on credit scoring). In general, a model used for decision making can be
categorized as either normative or descriptive. A normative model aids the
decision maker in reaching a good decision. A descriptive model, on the other
hand, represents the decision maker’s behaviour. The models discussed here are
normative in nature and are used as part of a larger decision-making process.
This process of mathematically modelling the variables important to the
extension of credit is referred to as “credit scoring”. Very briefly, statistical
analysis is performed on historical data to determine the relationship between
applicant data and payment performance. Based on this analysis, a scorecard of
significant variables is compiled, on which the “weights” of these variables
represent their significance or contribution. The total score is the sum of these
weights, and is then compared to a cut-off score for the final decision. The cut-
off score is based on a break-even calculation where, at the cut-off, the revenue
from the number of goods is precisely equal to the loss from an average bad.
Benchmarking for Quality
Management & Technology,
Vol. 3 No. 1, 1996, pp. 38-44.
© MCB University Press, 1351-3036