Retail product purchase model and behavioural scorecard development with macro-economical boundary conditions

Abstract

Due to the more and more close competition in bank sector it becomes momentous to predict accurately and to exploit the growning possibilities. The risk of the decisions has to be revealed and the incoming risk coming with the change of the portfolio have to be predicted.These aims are served by product affinity models on CRM departments and by credit score cards and basel parameters (PD, LGD) on risk departments of a retail bank. In most cases supervised learning techniques are used to model retail portfolio, with the condition that behaviour in the past is similar to the behaviour in the future. In this way the changing of the customers relation to the economic environment, difficulty of borrowing process and the marketing activity of the banks remain out of the scope. A future can be generated from the macroeconomical processes and from the data mining based models can lead to an inconsistent picture.In this work we ask wheter there is a relation between the consumer loans outstanding, the consumer loan portfolio quality and the consumer cinfidence index. How the cunsomer confidence index can be used in product affinity model and creadit score card development? And what other parameters are worth to involve into consistent model development processes to support the daily operativ decision making and the strategy planning as well.