New Databank system predicts corporate bankruptcies
Share this content

Japan's largest corporate credit research company, Teikoku Databank (TDB), has developed a system to predict the likelihood of a corporate borrower going bankrupt within a year.

The system, powered by SAS' analytical software, will also automate and speed up the credit approval process - always a problem for organisations that deal with a large number of small accounts.

"The system can predict whether or not a company is likely to go bankrupt with an accuracy of 70%," said Katsushige Iwabuchi, a member of the TDB team that built it. "We verified its performance using bankruptcy cases that occurred five months after the system was completed. The system certainly enables bad debt to be avoided quite efficiently."

TDB will calculate one-year bankruptcy prediction values, which it will offer on a trial basis to major organisations in Tokyo. The predictions will be revised monthly. The model also provides a set of standards that lenders can use to evaluate their customers.

Constructed with TDB-owned data from corporate surveys of approximately 820,000 companies, the system uses 10 variables based on information such as corporate management teams and a company's solvency.

The prediction system is based upon three data sources - company credit reports, internal TDB credit data, and data on businesses that went bankrupt during 2000.

"The software provided by SAS is superior in its ability to process and analyse large volumes of data at high speed," Iwabuchi said. "SAS has become the de facto standard for analytics within TDB."

About Teikoku Databank
Founded in 1900, Teikoku Databank has become the largest corporate credit research company in Japan. Its staff of 1,500 field reporters engages in research activities via a network of 82 offices nationwide.

Teikoku Databank



Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.