linkedin icon twitter icon facebook icon

Private Area

headline

Do you need some more informations?

Fill our contact form >

Credit risk analysis

Risk assessment and credit portfolio quality analytics play an important role in defining the business strategies of financial institutions. On one hand, it is important to assess the risk exposure relative to the market to better define the supply and management of financial products, on the other, it is important to analyze the quality of the portfolio and measure its performance so as to correctly size provisions against cost of funds and law requirements.

Credit Risk Analysis conducted by Nextbit at a major European banks covers the entire process of credit evaluation. Following is a list of the key steps undertaken; issues and expected results are highlighted for each.

Gathering requirements and defining indicators>

The list below depicts some of the most important indicators:

  • the rate of approval, acceptance and delivery of new contracts, the rate of positive and negative override the decision of approval;
  • the bad rate observed in the portfolio of contracts within a vintage framework;
  • the delinquent cycle rate of contracts that worsen or improve their performance over the month;
  • evaluation of contracts with the worst performance;
  • efficiency measurement for the process of debt collection:
  • recovery monitoring for debts collection actions undertaken from the first missed payment.

Define data modeling, and Enterprise Analytic Environment>

Based on the objectives defined in the requirements phase and the outcome of a downstream analysis of the data sources, the analytical team identifies the appropriate data model to best meet the criteria.

During our most recent project on Credit Risk, the design of the data model required two levels of data warehousing. The first level containing data with a logical structure which partly reflects that of the source data, the "snow flake" schema used contained:

  • dimensional tables with product and socio-demographic data;
  • fact tables with historical data on the performance and interactions of the subjects of analysis, e.g. behavioral data of customers.

The second level was a datamart (in a denormalized form) that allow users to analyze entities and phenomena, more accessible and easier to analyze than relational databases. It used for:

  • Analysis on credit approval process (application);
  • Live portfolio analysis (behavioral);
  • Analysis of the recovery process (collection).
  • Forecasting

Extract Transform Load & Data Quality>

The phase was needed in the first level of data warehouse to transform the operational data into useful and meaningful information.

Significant attention goes to the quality of the data within each process; requiring an evaluation of both syntactic and semantic correctness of input data and calculated indicators. Additional mechanisms of signaling abnormalities (through diagnostic rules) allow for the effective management and monitoring of data quality standards.

Another important feature introduced in this phase is the control and load functionality for a fully automated update process.

Advanced Analysis and Reporting>

Project objectives are supported by a set of advanced reports on indicators used to evaluate and monitor the entire portfolio and specific sub-segments. Among the most significant indicators are:

  • Credit Application analysis (credit approval)
    • Pipeline of credit application data, from automatic approval to final outcome.
    • Analysis of the policy rules applied when granting contracts/credit
    • Rating of the historical Probability of Default on the contracts issued
  • Vintage Analysis
    • Evaluation of the bad rate according to issuing time frame of contracts in the portfolio. The purpose of the analysis is to compare or monitor trends, compare the quality of families of contracts delivered at different time frames.
  • Portfolio Reporting
    • Historical time-series of the outstanding portfolio classified by state of insolvency
  • Characteristic Analysis
    • Evaluation on the explanatory power/capacity of selected variables with respect to the target, such as insolvency (defined as a number of unpaid installments).

Data Mining and Statistical Analysis >

The evolution of this “Enterprise Credit Analytic Environment" described by the case study, includes the implementation of statistical models that describe specific phenomena and trends within the portfolio churn, survival analysis, performance evaluation models of PD and behavioral acceptance, time-series analysis and forecasting for future budget and capital reserves.

The creation of these models relies on the use of advanced data mining and statistical techniques implemented within the “Enterprise Credit Analytic Environment" by Nextbit .

Additionally, the statistical models generate new predictive indicators that can be integrated within the "Enterprise Analytic Environment" or may involve other corporate functions as the key decision metrics.