Financial Crisis Warning Based On MD&A Language Emotion

Uncertain factors such as complex and ever-changing macro-micro environment and fierce competition and management risk within the company make avoiding financial risks a problem that every company must solve,it must not only prevent problems before they happen,but also consider how to solve them in a timely and effective manner if financial risks arise.This requires the company to respond in a timely manner based on the financial situation that may be revealed by current financial data.Therefore,a good financial crisis warning model can meet this realistic demand.In the limited sample information,the best compromise between the complexity and learning ability of the model is sought,in order to obtain the best promotion ability.It is a computer learning method with a strict theoretical foundation.Support vector machines have many unique advantages in solving small sample,nonlinear and high dimensional pattern recognition.Management Discussion&Analysis is an important part of the annual financial report of listed companies,and its language is flexible and changeable.Based on the understanding of the company's financial performance,management will use different sentimental language to convey forward-looking information to the market and investors.Therefore,this paper quantifies some emotional descriptions of management discussion and analysis,and then combines financial indicators to construct a support vector machine financial crisis early warning model.lt can be verified that the management discussion and analysis section does contain useful forward-looking information and can improve the prediction of the model.The research data of this paper is selected from the financial report data of T-2 years before the listed company in the industrial manufacturing industry in 2014-2017,and the sample is randomly selected according to the ratio of 1:2 of ST company and normal company.This paper selects the financial indicators from the Guotaian database from the three principles of comprehensiveness,representativeness and true reliability,and finally adds the MD&A language sentiment indicators as the indicator system of the whole early warning model.The content of this paper is divided into six parts:The first chapter is the introduction,introducing the research background and significance of this paper,the research methods,the structure of the paper and possible innovations.The second chapter is literature review,reviewing the research status of MD&A usefulness and the selection of early warning model indicators.At the same time,based on the domestic and foreign literatures,the development of the financial crisis early warning model is clarified.The third chapter is financial crisis and early warning model related theory.Firstly,it elaborates the concept of financial crisis in detail,and then introduces relevant theories,including information asymmetry theory,speech effective theory and impression management theory,and finally the theoretical support of optimal classification hyperplane,linear support vector machine,nonlinear support vector machine and kernel function is introduced in detail.The fourth chapter selects samples and indicators.First,it introduces the selection of samples and data sources.Then it introduces the selection principles of financial indicators and the quantification of MD&A language sentiment indicators.Finally,it introduces the screening steps of indicators.The fifth chapter is an empirical test based on the MD&A language sentiment indicators.It conducts normal distribution test,T test and non-parametric test on financial indicators and non-financial indicators.Then use matlab simulation to construct the financial crisis early warning model for prediction,compare and analyze the prediction results of the SVM model before and after adding MD&A positive emotional index,and finally conduct the robustness test.The research finds:on the basis of financial indicators,introducing MD&A language sentiment indicators,the classification accuracy of the model was found to be improved.lt shows that the management discussion and analysis part of the listed company's annual report does contain valid information and can reflect the development of some companies' future perfonnance.The sixth chapter is the conclusion and prospect of this paper.According to the above theoretical analysis and empirical research,the conclusions of this paper are summarized,the shortcomings in this paper are analyzed,and the future research directions are prospected.