Predicting takeover targets on Oslo stock exchange : an extension to the prediction literature on the Norwegian M&A market
MetadataShow full item record
- Master Thesis 
The purpose of this research is to extend the previous literature on prediction of takeover targets on Oslo Stock Exchange. We aim to improve the prediction model developed by Khan and Myrholt (2018) by introducing new variables related to intellectual property and target-firm announcements, in addition to applying a gradient boosting algorithm (XGBoost). Furthermore, we seek to obtain positive abnormal returns from the predicted target portfolios. We examine the marginal contribution of predictive power from the new variables with logistic regression, using the predictors from Khan and Myrholt (2018) as control variables. Using the area under the Reciever Operating Characteristics curves (AUC-ROC) and Precision-Recall curves (AUC-PR) as performance metrics, we examine the difference in prediction skill between logstic regression and XGBoost. Furthermore, we evaluate the systematic risk adjusted returns of the target portfolios using the Fama-French 3-factor model. We find that companies without a patent portfolio are significantly more likely to receive a takeover bid. Further, we observe that companies with a more negative tone in their announcements are more likely to be the target of a takeover bid. However, we only find significance at a 10% level. By applying the XGBoost algorithm, we observe better AUC-ROC and AUC-PR values. However, we cannot conclude that the improvements are significant. We are not able to achieve positive abnormal returns for the target portfolios individually. Moreover, in line with the performance metrics, we find no significant abnormal returns when comparing the portfolios from XGBoost to logistic regression.