SPSS - Customers

CMSS

Situation

CMSS is the largest buildings savings bank in the Czech Republic, with more than three million customers. It offers a range of loans for building- and housing-related finance such as mortgages, redevelopment, and construction.

Challenge

Like most European banks, CMSS is preparing to meet the requirements of Basel II regulations, which define new principles for bank risk management and minimal capital requirements. For a loan provider such as CMSS, the qualification of credit risk—particularly related to the risk of default on a loan—is one of the key areas of focus for meeting Basel II requirements.

Solution

CMSS searched for a data mining tool with a wide variety of modeling capabilities, as well as the ability to work with large amounts of data and hundreds of different variables. Following a competitive tender and a comparative evaluation of the solutions available, the bank chose Clementine®, SPSS’ data mining workbench, as the best solution to meet its needs.

Improved credit risk analysis

The first step in the bank’s project was to build a database for credit risk analysis. This involved creating a large data warehouse that was able to record not only the complete history of each client on the lending side of the business, but also information on the millions of customers with savings accounts. The data collected included basic socio-demographic information, as well as detailed transactional data. The data warehouse, based on the Microsoft® SQL Server™ platform, was built in cooperation with analysts from SPSS, who had rich experience in this area.

The main purpose of creating the data warehouse was to ensure access to high-quality data. This data could then be used to model and predict behavior of CMSS’ retail clients, such as the probability of defaulting based on previous savings and borrowing behavior.

SPSS’ solution was invaluable in this stage of development, as it enabled the team to ensure the high quality of the data held in the warehouse by easily identifying duplicate records and thoroughly cleansing it.

Once the data warehouse was set up, the next phase of the project involved using Clementine to develop models for assessing credit risk and scoring customers.

Reduced computing time by 75 percent

CMSS is now using SPSS to perform a whole portfolio of credit risk analyses, including behavioral scoring, risk assessment, fraud detection, and application scoring. Behavioral scores are calculated within the credit risk database on a monthly basis, and the appropriate probability of default is defined for each loan. CMSS can then determine the expected loss and capital requirements for its entire portfolio. This analysis has enabled CMSS to establish a risk management policy based on the above reports.

Using SPSS, it became easier for CMSS to develop scoring models that are necessary to comply with Basel II. The design of Clementine has helped the bank to reduce computing time by 75 percent for much of its analyses, as calculations can now be done automatically. In addition, the ability to assign certain operations to a separate server has reduced the amount of disk space required, despite the large amount of data being processed.

“The entire project took three months, from the initial preparation of data to testing and ‘go live.’ In all, over 80 percent of the analysis was completed using SPSS’ software. Without it, we wouldn’t have been able to set up the high-quality credit risk analysis system that we have today,” said Jiri Lakosil, risk manager at CMSS.