Dr Cindy Yoshiko Shirata

Predictors of Corporate Bankruptcy in Japan: An Empirical Analysis of Annual Reports by Text-Mining

When:  Wednesday 1 February 2012, 16.00 - 17.30,
Where:  Seminar Room SRA06, School of East Asian Studies, 6/8 Shearwood Road, University of Sheffield.
http://www.shef.ac.uk/seas/contact
Event type:  NIJS Seminar


Bankruptcy predictions are one of the most interesting research topics in accounting. The number of bankruptcy prediction models boomed after Altman (1968) presented the Z-model, with most subsequent models being based on financial ratios. However, frequent updating of accounting standards in recent years has made it very difficult to compare such financial numbers between fiscal years and companies, creating confusion and difficulties for information users.

 Although Japanese Accounting Standards have not yet been completely adapted to International Financial Reporting Standards (IFRS), they are moving closer to these standards year by year. Against the backdrop of this development, the analysis of qualitative non-financial information has become an important topic in recent years. Since 31 March 2004, annual reports in Japan have to include information on uncertainty risk, corporate governance and a discussion and analysis section by senior management. This new information has trigged the development of new research methods for evaluating a company’s financial position.

 Signs of the changing financial position of companies may appear in non-financial information earlier than in financial numbers. Non-financial variables such as the intellectual capital index can be included in bankruptcy prediction models, but such non-financial variables do not allow sufficient comparability.

 In this research, I demonstrate a way to differentiate between companies which became bankrupt and those that did not by paying attention to such non-financial information. However, rather than analyzing non-financial variables as such, I concentrate on textual data in financial reports. Annual reports contain substantial supplementary information in narrative form, and by analyzing the texts in this supplementary information, we were able to extract key phrases/descriptions to predict bankruptcy.

 Dr. Cindy Yoshiko Shirata, BA, MBA, Ph.D.

After 23 years industry experience, Professor Shirata shifted her activities to research and education 16 years ago. Since April 2007, she teaches at the Graduate School of Business Science, University of Tsukuba. She is currently the Secretariat General of the Science Council of Asia, Council Member of the Science Council of Japan and Committee Member of the Ministry of Land, Infrastructure, and Transport. Being one of the most well-known bankruptcy experts in Japan, her bankruptcy prediction model SAF2002 is used by major Japanese banks and rating agencies. Professor Shirata can be reached by email at: shirata.cindy.fe@u.tsukuba.ac.jp


Comments