Authors: Mihail Diakomihalis, Maria Kyriakou

Title: Port Categories and Bankruptcy Prediction

Abstract

This paper examines the bankruptcy prediction of three different port categories (Cargo, Ferry and Cruise), for ten public Greek ports, and, for the period 2002-2017. The research question is whether the port categories are safe enough or not from bankruptcy in the near future. Our methodology, first of all, it takes into consideration the process of ANOVA and through this analyzes the mean differences between the three port categories based on Altman Z-score. The results indicate that the means of Safe, Grey and Distress zones do not differ for the Cargo port category. For the other two port categories the means differ. This indicates the significance of zones in the port categories. In addition, we observe that the predicted ability of ANOVA is more accurate for the Ferry port category, meaning that the bankruptcy prediction in this port category is more possible to be observed. Furthermore, the Confidence Intervals of ANOVA indicate that the two of the three zones are good indicators of bankruptcy prediction. These are the Grey and the Distress zones.

The three graphs show that there is a clear difference between the Safe and the other two zones (Grey and Distress). In particular, only in Cargo port category the difference is smaller among the three port zones. The graphs indicate the importance of ANOVA results and confirm these findings, overall.

Continuing the approach with our own Z score calculation, we observe that when the X1, X2, X3 and X4 variables are taken into consideration as Altman suggests with a binary logit model, the results state higher possibility of bankruptcy in the near future for the three port categories. Finally, we do a discriminant analysis in order to observe at which degree the percentage of dataset is accurate, as it has been separated, and is based on Altman Z score.

The results indicate that nearly 49.3% has been categorized correctly. The dataset for the whole period of investigation (2002-2017) shows that for all the port categories the above percentage is true. The squared distance shows the distance between the three port categories and it is the smallest for Ferry and Cargo ports, meaning that for these two ports the best classification of dataset occurs. In addition, this classification is verified by the largest regression coefficients of X1, X2, X3 and X4 variables. The largest regression coefficients differ among the three port categories given significance on the whole process of investigation. Last but not least, there are some misclassified observations (around 51.7%) which have not been classified correctly. Through the discriminant analysis, we provide the correct classification group on which each misclassified observation belongs.