Measuring corporate failure risk: Does long short-term memory perform better in all markets?

IF 1.2 4区经济学 Q3 BUSINESS, FINANCE

Investment Analysts Journal Pub Date : 2023-01-02 DOI:10.1080/10293523.2022.2155353

Hyeongjun Kim, Hoon Cho, Doojin Ryu

引用次数: 0

Abstract

ABSTRACT Recently, various corporate failure prediction models that use machine learning techniques have received considerable attention. In particular, using a sequence of a company's historical information, rather than just the most recent information, yields better predictive performance by adopting recurrent neural networks (RNNs) and long short-term memory (LSTM) algorithms in the United States market. Similarly, we evaluate whether these results hold in emerging market contexts using listed companies in Korea. We also compare the logistic regression, random forest, RNN, LSTM, and an ensemble model combining these four techniques. The random forest model with recent information outperforms the other models, indicating that corporate failure prediction models for immature markets, unlike those for developed markets, might have to focus more on recent information rather than on the historical sequence of corporate performance.

查看原文本刊更多论文

衡量企业失败风险：长短期记忆在所有市场中表现更好吗？

摘要近年来，使用机器学习技术的各种企业故障预测模型受到了相当大的关注。特别是，通过在美国市场上采用递归神经网络（RNN）和长短期记忆（LSTM）算法，使用公司的历史信息序列，而不仅仅是最新信息，可以产生更好的预测性能。同样，我们使用韩国上市公司来评估这些结果是否适用于新兴市场。我们还比较了逻辑回归、随机森林、RNN、LSTM和结合这四种技术的集成模型。具有最近信息的随机森林模型优于其他模型，这表明不成熟市场的企业失败预测模型与发达市场的模型不同，可能必须更多地关注最近的信息，而不是企业业绩的历史序列。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Investment Analysts Journal BUSINESS, FINANCE-

CiteScore

1.90

自引率

11.10%

发文量

期刊介绍： The Investment Analysts Journal is an international, peer-reviewed journal, publishing high-quality, original research three times a year. The journal publishes significant new research in finance and investments and seeks to establish a balance between theoretical and empirical studies. Papers written in any areas of finance, investment, accounting and economics will be considered for publication. All contributions are welcome but are subject to an objective selection procedure to ensure that published articles answer the criteria of scientific objectivity, importance and replicability. Readability and good writing style are important. No articles which have been published or are under review elsewhere will be considered. All submitted manuscripts are subject to initial appraisal by the Editor, and, if found suitable for further consideration, to peer review by independent, anonymous expert referees. All peer review is double blind and submission is via email. Accepted papers will then pass through originality checking software. The editors reserve the right to make the final decision with respect to publication.