Research On The Multiple Markov Model Of Listed Companies' Financial Distress Based On Partial Order Constraint

For a long time,the prediction of financial distress of listed companies has been animportant issue of concern to academia.A company's financial distress may affect its own operations and cause financial risks to related companies.Effective financial distress prediction can prompt decision makers to take corresponding measures before financial risks occur,and avoid financial distress and reduce losses.As we all know,there is the possibility of risk contagion between companies in the same industry.In order to reflect the mutual influence of financial distress risks between companies in the same industry,this paper builds a model that introduces the influence of partial order between companies.We take the minimum distance between the probability distribution of the steady distribution after the state transition and the original distribution as the objective function,and take the partial order relationship of the influence coefficient between companies as the constraint condition,and then establish a multivariate Markov warning model considering the influence of the partial order of the listed company.The innovation of this study is that in the construction of partial order relationship constraints,we applied the improved G1 group expert decision-making method and online learning group expert decision-making method that reflect the opinions of group experts,and characterized the differentiated risk impact characteristics between different companies.In the empirical part,this paper takes the financial status of listed companies that experienced financial distress in the financial industry from 2001 to 2016 as input parameters,and constructs an existing multivariate Markov model that does not introduce a partial order relationship,and a multivariate Markov model that introduces a partial order relationship.Then we predict the financial situation of listed companies in 2017.The results show that the early warning model of financial distress which introduces the influence of partial order improves the accuracy of prediction and improves the early warning capability of the model.At the same time,compared with the improved G1 method of determining expert weights which only based on the similarity of expert opinions,the online learning method which updates expert weights according to the model prediction effect can better make use of information,construct more accurate partial order influence constraints,and improve the prediction performance of the model.In summary,this paper solves a problem of early warning of financial distress that considers the influence of the company's partial order,and provides a new idea for the construction of a financial distress early warning model for Chinese listed companies.