Bankruptcy prediction in construction companies via Fisher's Linear Discriminant Analysis

Abstract

Based on Fisher's Linear Discriminant Analysis, the paper proposed an early warning model to predict the bankruptcy possibility for construction companies in China. Sample and variable set of the study came from 56 privately owned Chinese construction companies and their financial ratios, and 27 companies' crisis occurred during the period 2008–2009. Kolmogorov-Smirnov test was used to examine the significance of the basic financial characteristics. The correlative analysis was used to exclude the influence of multicollinearity. Six financial ratios were determined to be significant. Based on these financial ratios, Fisher-LDA was applied to construct the bankruptcy prediction model. Availability of the bankruptcy prediction model was evaluated with other statistic models and intelligent technique (Bayes, PCA, Logistic regression and Neural Network) with the same sample data. The accuracy of the model reached 88.46% and 92.31% on the validation sample with two years and one year in advance respectively, which was better than other models. Results of the study show that our model has good accuracy in early prediction.