Developing Multicriteria Decision Aid Models for the Prediction of Share Repurchases

This study presents the first attempt to develop classification models for the prediction of share repurchases using multicriteria decision aid (MCDA) methods. The MCDA models are developed using two methods namely UTilites Additives DIScriminantes (UTADIS) and ELimination and Choice Expressing REality (ELECTRE) TRI, through a ten-fold cross-validation approach. The sample consists of 1060 firms from France, Germany and the UK. We find that both MCDA models achieve quite satisfactory classification accuracies in the validation sample and they outperform both logistic regression and chance predictions.