Title: Predicting the likelihood of dividend payment from Indonesian public companies with data mining methods

Authors: Raymond Kosala

Addresses: Binus International, Bina Nusantara University, Jakarta, 10270, Indonesia

Abstract: Making decisions about dividend payment is one of the most important choices for public companies. In finance literature, deciding to pay dividends is a controversial subject. Thus, there are several theories on the reason why companies choose to pay dividends. This paper investigates the likelihood of dividend payment from public companies in Indonesia using some artificial intelligence and statistical techniques. In the process, the possibility of using dividend related data from public companies as a dataset for predictive techniques is analysed. Then domain expert knowledge is used from finance literature to determine the predictor variables. Next predictive models are developed for the likelihood of using dividend payments by applying four different methods: logistic regression, artificial neural networks, decision tree induction, and support vector machines. Afterwards, the predictive accuracy performance of the models generated by these methods is examined. Finally, the resulting decision tree model is analysed and the validity of the resulting predictive model is confirmed with a domain expert. Then the resulting model is converted to create simple and understandable if-then rules, which can be used by company management to make decisions on dividend payments in practice. These if-then rules support the lifecycle theory of dividends as exhibited in previous works.

Keywords: dividend policy; dividend policy forecasting; decision support systems; logistic regression; artificial neural networks; decision trees induction; support vector machines.

DOI: 10.1504/IJBIS.2017.086324

International Journal of Business Information Systems, 2017 Vol.26 No.2, pp.139 - 150

Received: 05 Jan 2016
Accepted: 30 Apr 2016

Published online: 04 Sep 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article