Duane B. Kennedy

Effectiveness of Neural Network Types for Prediction of Business Failure

The study examines the effectiveness of different neural networks in predicting bankruptcy filing. Two approaches for training neural networks, Back-Propagation and Optimal Estimation Theory, are considered. Within the back-propagation training method, four different models (Back-Propagation, Functional Link Back-Propagation With Sines, Pruned Back-Propagation, and Cumulative Predictive Back-Propagation) are tested. The neural networks are compared against traditional bankruptcy prediction techniques such as discriminant analysis, logit, and probit. The results show that the level of Type I and Type II errors varies greatly across techniques. The Optimal Estimation Theory neural network has the lowest level of Type I error and the highest level of Type II error while the traditional statistical techniques have the reverse relationship (i.e., high Type I error and low Type II error). The back-propagation neural networks have intermediate levels of Type I and Type II error. We demonstrate that the performance of the neural networks tested is sensitive to the choice of variables selected and that the networks cannot be relied upon to “sift through” variables and focus on the most important variables (network performance based on the combined set of Ohlson and Altman data was frequently worse than their performance with one of the subsets). It is also important to note that the results are quite sensitive to sampling error. The significant variations across replications for some of the models indicate the sensitivity of the models to variations in the data.

Accommodating Outliers and Nonlinearity in Decision Models

This paper describes and compares six procedures that can be used in a regression model to adjust for outliers in the data and nonlinearities in the relationship between the dependent and independent variables. The data accommodation procedures are: (1) no-adjustment; (2) winsorizing; (3) trimming; (4) regression on ranks; (5) nonlinear regression; and (6) piecewise linear regression. The results show that the choice of data accommodation procedure has a major impact on the predictive ability and coefficient estimates of the regression model. The winsorizing and ranking procedures produce a regression model that fits the data well and has a low level of prediction error.

Target Ownership Plan and Earnings Management

We examine the determinants of target ownership plans and their association with earnings management. We find that firms adopt target ownership plans not only in response to poor performance as documented by prior research, but also as a means to mitigate moral hazard concerns and in response to peer pressure. Our analyses show that firms engage in less accruals and real earnings management for up to two years following plan adoption. Additionally, we find that the reduction in accruals management is concentrated in adoption firms that have met the target ownership requirements, and in high probability adoption firms whose adoption decisions are driven by performance, moral hazard, governance, capital market, and peer-firm concerns. Our findings support the argument that target ownership plans align CEOs’ incentives with shareholders’ in that these plans result in reduced management short-termism as evidenced by more truthful earnings reporting and less real earnings management through production cost manipulation and discretionary expenditures.