Testing the generalizability of the bankruptcy prediction models of Altman, Ohlson and Zmijewski for Dutch listed and large non-listed firms

Abstract

Bankruptcy and bankruptcy prediction is a very actual subject in the news and academic literature. The problem of the bankruptcy prediction models is the generalizability of the models because they there are developed with a specific sample. In the original studies, the sample included firms in a specific industry and a specific time period. The goal of this study is to test the generalizability of bankruptcy prediction models to industries and periods outside of those in the original samples. In the literature of bankruptcy prediction the models of Altman (1968), Ohlson (1980), and Zmijewski (1984) are the most cited ones that are based on accounting variables. These bankruptcy prediction models use different explanatory variables and statistical techniques. Therefore, the predictive power of these bankruptcy prediction models differ. I re-estimate these bankruptcy prediction models using an estimation sample which covers the period 2008-2010 and validate the models with another sample which covers the period 2011-2012. There are 15 bankrupt and 476 non-bankrupt firm included in the estimation sample. For the validation sample there are 14 bankrupt and 326 non-bankrupt firms included. All these firms are Dutch listed and large non-listed firms. Firstly, I test the bankruptcy prediction models with their original statistical technique. Secondly, to examine the role of the accounting ratios, I test all the bankruptcy prediction models with the logit regression. When the original statistical techniques are used, the accuracy rates for the models of Altman (1968), Ohlson (1980), and Zmijewski (1984) are respectively 80.6%, 93.8%, and 95.3%. At first sight it looks like the model of Zmijewski (1984) has the highest predictive power. But the model of Zmijewski (1984) predicted 0% of the bankrupt firms correctly and 99.4% of the non-bankrupt correctly. The accuracy rates for the models of Altman (1968), Ohlson (1980), and Zmijewski (1984) models are respectively 49.1%, 93.8%, and 87.7% when the logit regression is used. At first sight it looks like the model of Ohlson (1980) has the highest predictive power. But the same applies for the model of Ohlson (1980) as for the results of the model of Zmijewski (1984). The model of Ohlson (1980) is the most accurate when all the models use the same statistical technique. This implies that the explanatory variables of this model are the best predictors of the likelihood of bankruptcy. In conclusion, practitioners should use the bankruptcy prediction models of Altman (1968), Ohlson (1980), and Zmijewski (1984) cautiously because the frequency of Type I errors is high (Ohlson [1980] and Zmijewski [1984]) or the accuracy rate is low (Altman [1986]). To use these models in practice, I recommend to re-estimate the coefficients of the bankruptcy prediction models with a specific and bigger sample to improve the predictive power.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 33 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.