VALIDITY OF Z-SCORE

Validity of Z-score



Validity of Z-score

Introduction

Discriminant analysis (DA) is a classification method that can predict group membership of a newly sampled observation. In a use of DA, a group of observations whose memberships are already identified are used for the estimation of weights (or parameters) of a discriminant function by some criteria such as the minimization of misclassifications or the maximization of correct classifications. A new sample is classified into one of several groups by DA results.

In reviewing the previous works, the first research effort (1999) described a goal programming formulation of DEA-DA and then compared it with the additive model of DEA. Sueyoshi and Kirihara (1998)1 documented how to incorporate prior information into the evaluation process of DEA-DA. The original DEA-DA (1999) had a major drawback because it could not deal with a negative value in data. To overcome the shortcoming, Sueyoshi (2001) extended DEA-DA in such a manner that it could deal with the negative value in a data set. Sueyoshi (2004) extended DEA-DA further. The extended version (2004) incorporated the total number of misclassifications (not its total distances) into the objective function of DEA-DA. The objective function minimized the number of misclassifications. In the stage, the extended version (2004) reformulated DEA-DA by mixed integer programming (MIP), because it needed to incorporate a binary variable to count the number of misclassified observations. Sueyoshi (2006) compared the performance of DEA-DA (2004) with those of other DA methods originated from other disciplines such as statistics, econometrics and computer science.

To attain the research objective, this study compares the analytical structure of DEA-DA with Altman's financial ratio analysis and then applies it to the performance evaluation of Japanese manufacturing industry. Altman's study (1968) provided a pioneering work regarding bankruptcy prediction and corporate rating. Many researchers in finance have been long paying attention to his work. His research is still insightful and valid. Hence, this study compares DEA-DA with Altman's original work in terms of research strategy. The comparison identifies the position of DEA-DA in the research development of financial ratio analysis.

Discussion

To examine the performance of a business enterprise, we must specify how to assess its corporate achievement. Then, we select an analytical method(s) used for the corporate assessment. This study evaluates the corporate performance by bankruptcy -based financial assessment because many shareholders seriously care about an extreme loss that may drive their invested firms into bankruptcy. The bankruptcy evaluation has become a standard of corporate assessment in establishing a long-term financial strength.

The previous research on bankruptcy -based performance assessment was broadly grouped into three categories. First, there was research that centered on a particular model. These studies tested how a particular model performed in predicting corporate failure vis-à-vis another model (e.g., Lo, 1986). Second, there was research that focused on the selection of an appropriate set of variables to implement a particular model. While this may seem trivial, sample and variable selection is an important part of the bankruptcy assessment (e.g., [Keasey and Watson, 1991] and [Platt and Platt, ...