A hybrid decision tree based methodology for event studies and its application to e-commerce initiative announcements

  • Authors:

  • Francis Kofi Andoh-Baidoo;Kweku-Muata Osei-Bryson;Kwasi Amoako-Gyampah

  • Affiliations:

  • University of Texas-Pan American, Edinburg, TX, USA;Virginia Commonwealth University, Richmond, VA, USA;University of North Carolina at Greensboro, Greensboro, NC, USA

  • Venue:
  • ACM SIGMIS Database
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

The event study methodology has been applied in various business disciplines. This methodology typically has two goals: (1) to determine whether an event such as the announcement of an e-commerce initiative in the public media leads to cumulative abnormal returns (CAR); and (2) to examine the factors that influence the observed CAR. Most studies have used parametric statistical analysis in estimating CAR and regression for achieving the second goal. In this paper, we propose a hybrid methodology that involves using nonparametric statistical analysis to obtain the first goal, and the use of Decision Tree (DT) induction as a novel approach to reach the second goal. We apply the hybrid methodology to e-commerce announcements. The use of nonparametric analysis enables us to address some of the prior concerns of event study research in the e-commerce domain with regard to the limitations of short run event windows. The use of the novel DT-based approach leads to additional insights beyond what is reported in the literature through the examination of contingency effects. Specifically, our results indicate that the impact of Governance, Customer Type and Firm Type on CAR is contingent on the innovativeness of the e-commerce initiatives. Our study makes both methodological and theoretical contributions regarding the use of DT induction and nonparametric analysis in event studies especially in situations where prior studies present mixed results and where there are concerns about return variability. We present both research and managerial implications of the findings.