Comparison of accounting-based financial distress prediction models of companies listed on the JSE

dc.contributor.authorKhan, Sheriff
dc.date.accessioned2023-02-16T06:25:02Z
dc.date.available2023-02-16T06:25:02Z
dc.date.issued2022
dc.descriptionA research report submitted in partial fulfilment of the Degree of Master of Commerce to the Faculty of Commerce, Law and Management, University of the Witwatersrand, Johannesburg, 2022
dc.description.abstractThe study compared the forecasting accuracy of binary state corporate failure prediction models (multiple discriminant analysis and logit), and multistate models (multinomial and mixed logit models), to assess which models were more reliable in predicting financial distress and corporate failure from a South African context. The study used a sample of 108 firms listed on the Johannesburg Stock Exchange (JSE) for the period between 2010 to 2019. The sample was sub-divided into a testing and a validation sample and the accuracy of the models was tested 5 years prior to failure, 3 years prior to failure and 1 year prior to failure. The empirical results indicate that the binary models performed relatively well up to 3 years prior to failure; but their performance dropped considerably beyond that. The multistate models produced better results overall and their performance did not materially drop as the lead time from failure increased. This study provides evidence that multistate corporate failure prediction models can be used to predict corporate failure.
dc.description.librarianPC(2023)
dc.facultyFaculty Commerce, Law and Management
dc.identifier.urihttps://hdl.handle.net/10539/34529
dc.language.isoen
dc.titleComparison of accounting-based financial distress prediction models of companies listed on the JSE
dc.typeDissertation
Files
Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
1. Research Report (Final).pdf
Size:
583.39 KB
Format:
Adobe Portable Document Format
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.43 KB
Format:
Item-specific license agreed upon to submission
Description:
Collections