The objective of this research was to build artificial neural network models to predict target mergers and post-merger performances of merged companies. In each case, three types of models--horizontal, vertical, and total--were built. A new approach integrating a statistical technique, backward STEPDISC, and feedforward neural networks, was developed. To compare and evaluate the performance of this proposed method, statistical methods were also constructed.; Results obtained by using the combination of backward STEPDISC and feedforward neural network method significantly outperformed other methods in predicting target merger and post-merger performance. |