Predicting corporate bankruptcy from financials and payment default entries
No Thumbnail Available
URL
Journal Title
Journal ISSN
Volume Title
School of Business |
Master's thesis
Authors
Date
2023
Department
Major/Subject
Mcode
Degree programme
Information and Service Management (ISM)
Language
en
Pages
61+8
Series
Abstract
This thesis studies the effect of including payment default entry data into a bankruptcy prediction model. Because in previous research the effect has not been studied extensively and still payment default entries are widely used in practice in determining the probability of a company’s bankruptcy, it is interesting to see if this piece of data actually is important in the prediction process. Additionally, the goal of this thesis is to find out whether or not different financial variables are important when predicting bankruptcy for a defaulted company versus a non-defaulted company. A total of four models are created to answer the research questions, and variables are constructed from financial statements, bankruptcy information and payment default entry data provided by Suomen Asiakastieto Oy. Logistic regression is chosen as the main modelling method and together with a downsampling technique, an elastic net regularization is used for variable selection. The models’ coefficient estimates are interpreted when necessary, and their predictive performance is compared in order to find similarities and differences in the models. The study suggests that payment default entry data is important when predicting a corporate bankruptcy as it enhances the predictive performance of the model. Moreover, there are clear differences in the variables that get selected into the models predicting for defaulted versus non-defaulted companies, and it is found that altogether predicting for defaulted companies is difficult perhaps due to lack of useful information in their financial statements.Description
Thesis advisor
Malo, PekkaKeywords
konkurssin ennustaminen, tilinpäätökset, maksuhäiriöt, tilastollinen mallinnus