University of Twente Student Theses

Login

Prediction power of accounting-based bankruptcy prediction models : evidence from Dutch and Belgian public and large private firms

Bouwmeester, A. (2020) Prediction power of accounting-based bankruptcy prediction models : evidence from Dutch and Belgian public and large private firms.

[img] PDF
1MB
Abstract:This study examined the prediction power of the accounting-based bankruptcy prediction models of Altman (1983), Ohlson (1980), and Zmijewski (1984) for Dutch and Belgian public and large private firms. These bankruptcy prediction models include different financial ratios as independent variables and are developed using different econometric methods: multiple discriminant analysis, logistic regression, and probit regression, respectively. It is tested if one of the prediction models outperforms the others, which econometric method is best for developing the models, if the coefficients of the models are non-stationary, and what the optimal time horizon for predicting bankruptcy is. The performance of the three models is assessed using two different estimation samples with firm observations from 2007 – 2010, and 2012 – 2015, and a hold-out sample with firm observations from 2016 – 2019. These different time periods for the estimation samples are chosen because of the different economic environments. During the first time period (2007 – 2010), several important events occurred such as the financial crisis in 2007 and the European debt crisis in 2010. Firms from all industries, except the financial and insurance industry, were included in the samples. The results show that the models of Altman (1983), Ohlson (1980), and Zmijewski (1984) predicted respectively 32.39%, 47.89%, and 38.03% of the bankrupt firms and 99.58%, 99.72%, and 100.00% of the non-bankrupt firms correctly. No statistical significant difference was found between the three prediction models of Altman (1983), Ohlson (1980), and Zmijewski (1984), and between the three econometric methods multiple discriminant analysis, logit regression, and probit regression. Additionally, no evidence was found for the non-stationarity of the coefficients. Finally, this study concludes that the optimal time horizon for predicting bankruptcy is one fiscal year before the event.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:85 business administration, organizational science
Programme:Business Administration MSc (60644)
Link to this item:http://purl.utwente.nl/essays/82525
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page