Experimental design bureau: Scoring of Bureau

Scoring of Bureau - the analytical tool for assessment of reliability of the borrower.

The product works according to the methodology developed by experimental design bureau on the basis of data from credit history of the borrower. The value forms in the form of scoring point.

The value of point reflects the probability of a default of the borrower in the next 12 months on any of its credits. During creation of scoring model the method of logistic regression is used. A set of predictive variables in model has exclusively behavioural character, i.e. contains only information from credit history of the borrower.

The service helps to solve problems effectively:

• Increase in the separating force of the current model of credit institution at adoption of credit solutions;
• Establishment of additional level of cutting off of "bad" borrowers by consideration of the credit request;
• Segmentation of the current customer base for decision making in procedures of cross-sellings, changes of credit limits, collecting of arrears.

Properties of service:

• Use of one scoring value allows to avoid additional costs on parsing and interpretation of volume credit reports;
• The cost of a request of Scoring is lower than the cost of a request of the full credit report;
• It is constructed on the most significant indicators of credit history;
• The model is regularly improved and updated.

With effective result can use service:

• banks;
• IFI;
• collection agencies;
• telecom operators;
• insurance companies;
• employers and recruiting agencies.

Efficiency of Scoring of Bureau

• In November, 2014 upgraded version of scoring is implemented. This model defines "bad" in low intervals and "good" better - in high.

Model of scoring, 2014

• Jeanie's coefficient shows quality of the scoring card and force of discrimination of "bad" and "good" observations in comparison with the accidental choice. On a Lorenz curve (ROC) of Jeanie it is equal to Square S1 relation

to S2. For the updated Scoring of Bureau Jeanie's coefficient makes 72.3 that speaks about high quality model and high discriminatory capability of the card.

Model of scoring, 2014

Features of scoring of Bureau III of generation

• scoring of Bureau of third generation is constructed on a broad spectrum of behavioural characteristics of credit history of the borrower and does not contain sociodemographic data.
• the scoring card is scaled in compliance with the international scale of Experian. It allows to expose the uniform strategy of points of cutting off for banks which is not required to be modified at correction of the scoring card of Bureau.
• the scoring card has an indicator of reliability and the risk index which do scoring point to more informative.
• for the borrowers who are in the current delay of 90 and more days on dny or to several obligations, the separate point returns (-111)
• one of elements of the scoring card is the index of credit loading (CII) which considers only congestion of the borrower credit obligations and does not consider other factors.

2015: The joint Credit Bureau started scoring of the IV generation – Scoring of Reliability

The joint Credit Bureau announced in December, 2015 start of new Scoring of Bureau IV of generation – so-called Scoring of Reliability. Scoring of the IV generation represents the analytical tool for a grade of reliability of the borrower in methodology of Bureau on the basis of data from credit history. The value is issued in the form of scoring point which reflects the probability of a default of the borrower in 12 months from the moment of issuance of credit.

Scoring of the IV generation is intended for banks, IFI, other credit institutions, the collection agencies, telecom operators and insurance companies. Scoring will allow to solve the following problems: increase in the separating force of the current model of credit institution at adoption of credit solutions; establishment of additional level of cutting off of "bad" borrowers by consideration of the credit request; segmentation of the current customer base for decision making in procedures of cross-sellings and changes of credit limits; increase in efficiency of collecting of arrears; economy on procedures of underwriting; decision making about sale of the loan portfolios.

This service differs from Scoring of the III generation in the relevance: during the work on service new population of 2014 and also some data of 2015 was analyzed. New variables were developed for creation of Scoring of the IV generation, and a set of predictive variables contains only information from credit history of the borrower.

The model of new Scoring is effectively complemented with one more analytical tool of experimental design bureau for assessment of reliability of the borrower – the Index of credit load (CII). The index reflects the volume and the nature of debt load and is focused on assessment of borrowers who did not leave in delay yet, but are risky owing to growth of credit obligations.

Scoring of Bureau IV of generation allows to achieve significant increase in efficiency of the working decision making models. Service is constructed on the most significant indicators of credit history. At its calculation data from 191 million credit stories on 61 million borrowers received by bureau from more than 500 banks are used. The share of informative answers (hit reyt) at requests in experimental design bureau makes 96% that guarantees very high degree of quality of the scoring card.

The cost of a request of Scoring is lower than the cost of a request of the full credit report. Use of one scoring value allows to avoid additional costs on parsing and interpretation of volume credit reports.

"The payment discipline of borrowers worsens, and credit loading for the last year grew on average by 3%. Banks should be extremely attentive during the work with the current loan portfolio. In this sense Scoring of Bureau IV of generation answers the most relevant tasks of credit institutions in the current realities of the market, - Akhlomov Alexander, the development director of products of experimental design bureau says. – The advantage of Service is easily checked by retrotesting, its efficiency is obvious from first month of connection. Besides, Scoring will be useful as when using within end-to-end systems on risk management of banks, and as independent model".