注册 投稿
经济金融网 中国经济学教育科研网 中国经济学年会 EFN通讯社

【12月6】上海:Corporate Failure Prediction (CFP)

题 目:  Corporate Failure Prediction (CFP) Modelling with Special Reference to China

演讲人:  Vic Edwards, University of New South Wales

主持人: 周春阳 博士

时间:  2010年12月6日 (周一) 3:30-5:00pm

地点:  上海交通大学北楼102

Abstract

This study is in two parts: the first dealing with the Australian economy and Australian data provided by the Corporate Scorecard Group (CSG). My research team used the opportunity to work on the CSG database to find the best corporate failure modelling technique from a number of non-portfolio approaches. The non-portfolio approach was taken because the second part of our work was to use a modern Chinese corporate database (the RESSET database). It is generally known in China, that portfolio techniques do not work because the equities market is both inefficient and opaque. The first part of our work concentrates primarily upon the testing and deriving of a “best” model from a number of techniques for predicting corporate failure among listed companies in Australia. The results achieved were excellent as we obtained over 90 percent prediction of corporate failure from one year beforehand with three of the models and matched the best of any previously published Australian work on the topic. Each model was capable of displaying very good explanatory significance and causality which is normally a serious weakness of heuristic models.

We used the Australian experience to develop our approach in the second part of the study dealing with corporate failure prediction in China and draws strongly from Erick Luwia’s honours thesis, “Bankruptcy Risk Prediction Using Neural Network and Logistic Regression Analysis in China”. Using the Australian work as a starting point, we found that neural networks gave the best results in China though we also used a highly performing logistic regression model as a benchmark. Our results were highly encouraging as they are among the best results obtained on Chinese CFP studies to date (approximately 90% for t-1 and 85% for t-2.

Should you wish to quote from this working paper, please seek permission from the corresponding author, Vic Edwards first.

文章评论
关注我们

快速入口
回到顶部
深圳网站建设