融合临时报告软信息的新三板企业财务困境预测研究

doi:10.16381/j.cnki.issn1003-207x.2021.0773

Abstract

Abstract:

Chinese “New Third Board” market plays an important role in providing financial services for small and medium-sized enterprises. However， with the rapid expansion of this market， the number of firms in financial distress is increasing year by year， which brings severe risk for investments. With the emergence of China as one of the leading markets for international investors， the “New Third Board” market has attracted increasing attention. Hence， predicting financial distress of firms in this market could provide strong support for investors and creditors， especially those in China， to make investment decisions and avoid investment risk. However， compared with listed firms， “New Third Board” firms have longer disclosure cycle of accounting information， and poorer continuity of market trading information， which makes their financial distress prediction faced with severe challenges. In order to alleviate the problem of information asymmetry effectively， a method of financial distress prediction by integrating soft information in current reports is proposed. At first， extract effective features from current reports using a topic model， and combine with accounting features to predict financial distress for Chinese “New Third Board” firms. At the same time， the prediction performances using soft information in current reports or in periodic reports are compared. The results show that soft information in current reports can significantly improve the accuracy of financial distress prediction， while soft information in periodic reports has no significant predictive effect. The study is important in theory meaning and practical value. First， our research broadens the literature on using textual information for financial distress prediction. Second， the proposed method can also be applied to other similar scenarios where the current reports are available， such as credit risk evaluation of listed firms. Third， investors may consider combining soft information in current reports with accounting information to make better investment decisions.

Key words: New Third Board, financial distress, current report, soft information, topic model

CLC Number:

F832.1

Xi-mei LV,Cui-qing JIANG,Yong DING,Zhao WANG. Financial Distress Prediction of “New Third Board” Firms by Integrating Soft Information in Current Reports[J]. Chinese Journal of Management Science, 2023, 31(11): 140-150.

Figures/Tables 14

References 21

1	何牧原，张昀.中国新三板市场的兴起、发展与前景展望［J］.数量经济技术经济研究，2017，34（4）：74-91.
	He Muyuan， Zhang Yun. The emergence， development and prospect of China’s New Third Board market［J］. The Journal of Quantitative and Technical Economics， 2017， 34（4）： 74-91.
2	Geng Ruibin， Bose I， Chen Xi. Prediction of financial distress： an empirical study of listed Chinese companies using data mining［J］. European Journal of Operational Research， 2015， 241（1）： 236-247.
3	Sun Jie， Fujita H， Zheng Yujiao， et al. Multi-class financial distress prediction based on support vector machines integrated with the decomposition and fusion methods［J］. Information Sciences， 2021， 559： 153-170.
4	王昱，杨珊珊. 考虑多维效率的上市公司财务困境预警研究［J］. 中国管理科学，2021，29（2）：32-41.
	Wang Yu， Yang Shanshan. Corporate financial distress prediction based on multi-dimensional efficiency indicators［J］. Chinese Journal of Management Science， 2021， 29（2）： 32-41.
5	Chen Xiaohong， Wang Xiaoding， Wu Desheng. Credit risk measurement and early warning of SMEs： an empirical study of listed SMEs in China［J］. Decision Support Systems， 2010， 49（3）： 301-310.
6	Tsai M， Wang Chuanju. On the risk prediction and analysis of soft information in finance reports［J］. European Journal of Operational Research， 2017， 257（1）： 243-250.
7	陈艺云.基于信息披露文本的上市公司财务困境预测：以中文年报管理层讨论与分析为样本的研究［J］.中国管理科学，2019，27（7）：23-34.
	Chen Yiyun. Forecasting financial distress of listed companies with textual content of the information disclosure： a study based on MD&A in Chinese annual reports［J］. Chinese Journal of Management Science， 2019， 27（7）： 23-34.
8	Zhao Shuping， Xu Kai， Wang Zhao， et al. Financial distress prediction by combining sentiment tone features［J］. Economic Modelling， 2022， 106： 105709.
9	Feng Mai， Tian S， Lee C， et al. Deep learning models for bankruptcy prediction using textual disclosures［J］. European Journal of Operational Research， 2019， 274（2）： 743-758.
10	于忠泊，田高良，曾振.上市公司临时报告对资本市场信息传递的影响［J］.系统工程理论与实践，2012，32（6）：1151-1165.
	Yu Zhongbo， Tian Gaoliang， Zeng Zhen. Impact of listed companies’ interim reports on market information transmission［J］. Systems Engineering-Theory and Practice， 2012， 32（6）： 1151-1165.
11	McMullin J， Miller B， Twedt B. Increased mandated disclosure frequency and price formation： evidence from the 8-K expansion regulation［J］. Review of Accounting Studies， 2019， 24（1）： 1-33.
12	蒋翠侠，黄韵华，许启发.基于Lasso二元选择分位数回归的上市公司信用评估［J］.系统工程，2017，35（2）：16-24.
	Jiang Cuixia， Huang Yunhua， Xu Qifa. Credit evaluation of listed companies via lasso binary quantile regression approach［J］. Systems Engineering， 2017， 35（2）： 16-24.
13	倪志伟，薛永坚，倪丽萍，等.基于流形学习的多核SVM财务预警方法研究［J］.系统工程理论与实践，2014，34（10）：2666-2674.
	Ni Zhiwei， Xue Yongjian， Ni Liping， et al. Research of multiple kernel SVM based on manifold learning in financial distress prediction［J］. Systems Engineering-Theory and Practice， 2014， 34（10）： 2666-2674.
14	张亮，张玲玲，陈懿冰，等.基于信息融合的数据挖掘方法在公司财务预警中的应用［J］.中国管理科学，2015，23（10）：170-176.
	Zhang Liang， Zhang Lingling， Chen Yibing， et al. Based on information fusion technique with data mining in the application of finance early-warning［J］. Chinese Journal of Management Science， 2015， 23（10）： 170-176.
15	Shen Feng， Liu Yongyong， Wang Run， et al. A dynamic financial distress forecast model with multiple forecast results under unbalanced data environment［J］. Knowledge-Based Systems， 2020， 192： 105365.
16	Suk I， Lee S， Kross W. CEO turnover and accounting earnings： the role of earnings persistence［J］. Management Science， 2021， 67（5）： 3195-3218.
17	Liang Deron， Tsai C， Wu Hsinting. The effect of feature selection on financial distress prediction［J］. Knowledge-Based Systems， 2015， 73： 289-297.
18	Yin Chang， Jiang Cuiqing， Jain H， et al. Evaluating the credit risk of SMEs using legal judgments［J］. Decision Support Systems， 2020， 136： 113364.
19	王铁军.媒体对证券市场的影响研究：基于大数据视角［D］.成都：西南财经大学，2016.
	Wang Tiejun. The effect of media on stock markets： from a big data perspective［D］. Chengdu： Southwestern University of Finance and Economics， 2016.
20	Wang Zhao， Jiang Cuiqing， Zhao Huimin， et al. Mining semantic soft factors for credit risk evaluation in peer-to-peer lending［J］. Journal of Management Information Systems， 2020， 37（1）： 282-308.
21	孟庆斌，杨俊华，鲁冰.管理层讨论与分析披露的信息含量与股价崩盘风险——基于文本向量化方法的研究［J］.中国工业经济，2017（12）：132-150.
	Meng Qingbin， Yang Junhua， Lu Bing. The informative content of management discussion and analysis and stock price crash risk： based on text vectorization method［J］. China Industrial Economics， 2017（12）： 132-150.

	LDA主题模型步骤
输入：	文档集D
流程：	1.对主题采样：?_m ~ Dir(β)，m ? [1,M] 2.对D中d文档d ? [1,D] -采样主题概率分布 θ_d ~ Dir(α) -采样文档长度N 3.对文档d中的第n个单词 n ? [1,N] -选择隐含主题Z_d,n ~ Mult(θ_d ) -生成一个词项W_d,n ~ Mult(?(Z_d,n ))
训练：	对θ_d 和?_m 进行参数估计
测试：	主题与对应关键词的后验概率分布
输出：	向量θ_d 和?_m

代码	特征名称	特征描述
X1	基本每股收益	净利润/总股本
X2	每股资本公积	资本公积/总股本
X3	每股未分配利润	未分配利润/总股本
X4	营业收入对数	LN（营业收入）
X5	净利润增长率	（当期净利润/上期净利润）-1
X6	净资产收益率	净利润/净资产
X7	总资产报酬率	息税前利润/平均资产总额
X8	销售净利率	净利润/销售收入
X9	总资产周转率	主营业务收入/总资产
X10	应收账款周转率	主营业务收入/应收账款净额
X11	资产负债率	总负债/总资产
X12	流动比率	流动资产/流动负债

主题	主题下的关键词
权益变动	权益、协议、转让、股权、变更、收购…
资金管理	资金、募集、存放、用途、账户、基金…
关联交易	交易、关系、经营、审议、定价、担保…
对外投资	子公司、投资、控股、全资、资金占用…
违规处理	失信、联合、惩戒、公司法、律师、诉讼…
高管调动	换届、任职、董事、监事、辞职、解除…
经营异常	限售、解除、转让、督导、暂停、恢复…

软信息类型	特征代码	特征名称	特征描述
临时报告软信息	T1	权益变动	权益变动主题的概率值
	T2	资金管理	资金管理主题的概率值
	T3	关联交易	关联交易主题的概率值
	T4	对外投资	对外投资主题的概率值
	T5	违规处理	违规处理主题的概率值
	T6	高管调动	高管调动主题的概率值
	T7	经营异常	经营异常主题的概率值
定期报告软信息	POS	管理层负语调	消极情感词分值/文档总词数
定期报告软信息	TONE	管理层净语调	（正语调-负语调）/（正语调+负语调）

主题特征	M1	M2	M3	M4	M5	M6	M7	M8
常数项	0.688(0.275)	0.783(0.314)	0.675(0.265)	1.968(0.777)	1.125(0.444)	0.365(0.145)	-0.781(-0.306)	0.9582(0.355)
T1	0.075(0.075)
T2		-1.143?(-1.759)						-1.235(-1.413)
T3			-2.814^*(-2.077)					-3.413^*(-2.121)
T4				-2.982^**(-3.022)				-2.505^*(-2.186)
T5					3.352^**(3.172)			2.400^?(1.897)
T6						1.390(1.477)
T7							4.249^***(5.401)	2.957^**(2.951)
AIC	355.14	351.9	350.03	343.87	347.09	353.2	329.15	319.46

Financial Distress Prediction of “New Third Board” Firms by Integrating Soft Information in Current Reports

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 14

References 21

Related Articles 13

Metrics

Comments

Recommended 0

模型	衡量标准	A₁	B₁	C₁	D₁	E₁	F₁
LR	AUC	0.829	0.699	0.623	0.840	0.828	0.842
	KS	0.713	0.494	0.413	0.725	0.717	0.724
	H-measure	0.617	0.382	0.286	0.638	0.625	0.632
	Accuracy	0.785	0.621	0.555	0.797	0.780	0.793
	Type1 Error	0.169	0.331	0.393	0.156	0.170	0.163
	Type2 Error	0.254	0.405	0.425	0.241	0.262	0.245
	F-measure	0.758	0.588	0.533	0.775	0.754	0.774
CART	AUC	0.788	0.663	0.630	0.856	0.795	0.855
	KS	0.629	0.428	0.378	0.714	0.627	0.712
	H-measure	0.493	0.271	0.237	0.607	0.491	0.605
	Accuracy	0.734	0.627	0.592	0.807	0.733	0.804
	Type1 Error	0.249	0.339	0.384	0.168	0.247	0.174
	Type2 Error	0.275	0.393	0.406	0.202	0.278	0.201
	F-measure	0.719	0.601	0.572	0.793	0.716	0.790
RF	AUC	0.858	0.732	0.621	0.939	0.855	0.932
	KS	0.747	0.539	0.420	0.863	0.74	0.855
	H-measure	0.659	0.432	0.286	0.817	0.654	0.807
	Accuracy	0.822	0.643	0.576	0.859	0.810	0.847
	Type1 Error	0.126	0.317	0.344	0.097	0.137	0.104
	Type2 Error	0.218	0.368	0.417	0.148	0.258	0.160
	F-measure	0.794	0.621	0.560	0.844	0.784	0.832

分组	均值排名	调整P值
分组	均值排名	B₁	C₁	D₁	E₁	F₁
A₁	3.76	<0.001	<0.001	<0.001	1.000	<0.001
B₁	2.73	-	<0.001	<0.001	<0.001	<0.001
C₁	2.27	<0.001	-	<0.001	<0.001	<0.001
D₁	4.36	<0.001	<0.001	-		0.169
E₁	3.81	<0.001	<0.001	<0.001	-	<0.001
F₁	4.17	<0.001	<0.001	0.169	<0.001	-
		总观测量=2100 自由度=5 卡方=2358.851

模型	衡量标准	提前半年						提前一年
模型	衡量标准	A₂	B₂	C₂	D₂	E₂	F₂	A₃	B₃	C₃	D₃	E₃	F₃
LR	AUC	0.813	0.684	0.611	0.821	0.816	0.824	0.809	0.676	0.605	0.817	0.805	0.818
	KS	0.687	0.485	0.397	0.692	0.671	0.698	0.678	0.460	0.385	0.690	0.680	0.688
	H-measure	0.583	0.377	0.265	0.599	0.585	0.598	0.574	0.348	0.261	0.595	0.581	0.599
	Accuracy	0.762	0.607	0.527	0.779	0.762	0.788	0.764	0.600	0.529	0.773	0.756	0.776
	Type1 Error	0.198	0.353	0.483	0.186	0.197	0.186	0.253	0.366	0.501	0.235	0.259	0.237
	Type2 Error	0.276	0.456	0.464	0.260	0.273	0.256	0.298	0.463	0.475	0.276	0.290	0.278
	F-measure	0.733	0.562	0.522	0.748	0.735	0.742	0.728	0.558	0.519	0.742	0.730	0.747
CART	AUC	0.763	0.677	0.608	0.832	0.770	0.831	0.731	0.643	0.600	0.786	0.732	0.779
	KS	0.588	0.435	0.345	0.672	0.587	0.670	0.536	0.396	0.341	0.612	0.539	0.607
	H-measure	0.443	0.240	0.203	0.553	0.440	0.552	0.388	0.289	0.195	0.476	0.389	0.475
	Accuracy	0.708	0.604	0.573	0.784	0.707	0.782	0.685	0.635	0.594	0.757	0.701	0.749
	Type1 Error	0.286	0.377	0.389	0.201	0.284	0.206	0.295	0.398	0.425	0.237	0.291	0.245
	Type2 Error	0.313	0.384	0.448	0.238	0.318	0.237	0.355	0.428	0.450	0.267	0.359	0.272
	F-measure	0.688	0.579	0.547	0.766	0.684	0.763	0.655	0.574	0.561	0.739	0.656	0.73
RF	AUC	0.836	0.719	0.600	0.925	0.832	0.917	0.809	0.747	0.613	0.856	0.810	0.856
	KS	0.713	0.524	0.390	0.837	0.705	0.828	0.673	0.566	0.383	0.728	0.677	0.735
	H-measure	0.616	0.395	0.254	0.784	0.611	0.771	0.577	0.482	0.266	0.652	0.579	0.657
	Accuracy	0.799	0.621	0.553	0.840	0.787	0.827	0.754	0.700	0.562	0.774	0.761	0.778
	Type1 Error	0.153	0.351	0.455	0.122	0.147	0.130	0.221	0.390	0.467	0.187	0.220	0.180
	Type2 Error	0.284	0.405	0.476	0.248	0.294	0.192	0.298	0.431	0.493	0.280	0.292	0.275
	F-measure	0.765	0.595	0.530	0.818	0.756	0.805	0.735	0.571	0.533	0.747	0.741	0.751

分组	均值排名	调整P值
分组	均值排名	B₂	C₂	D₂	E₂	F₂
A₂	3.56	<0.001	<0.001	<0.001	1.000	<0.001
B₂	2.48	-	<0.001	<0.001	<0.001	<0.001
C₂	2.03	<0.001	-	<0.001	<0.001	<0.001
D₂	3.90	<0.001	<0.001	-	<0.001	1.000
E₂	3.63	<0.001	<0.001	<0.001	-	<0.001
F₂	3.95	<0.001	<0.001	1.000	<0.001	-
		总观测量=2100 自由度=5 卡方=2217.556

[1]	Dawen Yan,Cun Li,Guotai Chi. Dynamic Financial Distress Prediction for Chinese Listed Companies Based on the Mixed Frequency Data [J]. Chinese Journal of Management Science, 2024, 32(1): 1-12.
[2]	LIANG Mo, LI Hong-xiang, ZHANG Shun-ming. A Measure for Financial Distress based on ST Predictive Model and the Cross-section of Stock Returns [J]. Chinese Journal of Management Science, 2023, 31(2): 138-149.
[3]	Xiang-zhong HUANG,Chang-wu ZOU,Yuan GAO. Research on the Impact of Loans Rollover Policy on Default Risk of Enterprises Based on Structural Model [J]. Chinese Journal of Management Science, 2023, 31(12): 329-339.
[4]	Jing GU,Ya-ting HU. The Transmission Effect of SMEs' Financial Distress on the Credit Risk of Suppliers [J]. Chinese Journal of Management Science, 2023, 31(12): 23-33.
[5]	WANG Yu, YANG Shan-shan. Corporate Financial Distress Prediction Based on Multi-Dimensional Efficiency Indicators [J]. Chinese Journal of Management Science, 2021, 29(2): 32-41.
[6]	CHEN Yi-yun. Forecasting Financial Distress of Listed Companies with Textual Content of the Information Disclosure: A Study based MD&A in Chinese Annual Reports [J]. Chinese Journal of Management Science, 2019, 27(7): 23-34.
[7]	JIANG Cui-qing, WANG Rui-ya, DIGN Yong. The Default Prediction Combined with Soft Informationin Online Peer-to-Peer Lending [J]. Chinese Journal of Management Science, 2017, 25(11): 12-21.
[8]	LI Li-jun, LIU Jie, SUN Shi-min. Study on the Propagation and Control of Financial Distress on the Perception of Supply Chain [J]. Chinese Journal of Management Science, 2016, 24(6): 46-51.
[9]	SUN Yan-hong, CHEN Ming-yan, CHEN Xiao-ling, BIAN Yi-wen. Study on the Decision Behavior of Capital-Constrained Newsvendor under Risks [J]. Chinese Journal of Management Science, 2013, 21(6): 72-79.
[10]	XU Xiao-yan, SUN Yan-hong. Study on the Propagation of Financial Distress along a Supply Chain [J]. Chinese Journal of Management Science, 2008, 16(4): 132-139.
[11]	ZHANG Zhi-wang. An Empirical Study on Cash Flows’ Information Content in Financial Distress Prediction——Evidence from ST firms during the period of 2003-2004 [J]. Chinese Journal of Management Science, 2004, (6): 23-28.
[12]	LI Bing-xiang, . The Research of Predicting Financial Distress in Chinese Listed Firms Based on Expected Default Frequency Model [J]. Chinese Journal of Management Science, 2004, (5): 12-16.
[13]	LI Bing-xiang. The Analyses of the Financial Feature of Investment Behavior in Financial Distress Firm in China [J]. Chinese Journal of Management Science, 2003, (2): 34-39.