Short-Horizon Event Study Estimation with a Star Model and Real Contaminated Events

Andreou, Panayiotis C.; Louca, Christodoulos; Savva, Christos S.

doi:10.2139/ssrn.2516068

Cited by 1 publication

(1 citation statement)

References 60 publications

(82 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because business cycles include an upward and downward trend consecutively, parameters should correspond to this accordingly. Examples from the event study literature include Andreou et al (2016) that use STAR model to estimate abnormal return and Atkas et al ( 2007) that utilize Markov switching model to estimate abnormal returns and show that their specification is superior in detecting abnormal returns.…”

Section: Introductionmentioning

confidence: 99%

Optimal forecast error as an unbiased estimator of abnormal return: A proposition

Enginar

Atıcı

2021

Journal of Forecasting

View full text Add to dashboard Cite

In the event studies, the accuracy of the abnormal returns assessment is highly dependent on the accuracy of the preceding expected return model. If the expected return model is inadequate, there is a possibility that a part of returns is labeled as abnormal returns even though they are not. Currently, we have a variety of options to set up an expected return model. To obtain unbiased abnormal returns, one should pay attention to the performance of the expected return model. In this research, we propose that the optimal forecast lemma can be consulted beforehand so that minimizing the optimal forecast error in the expected return model will yield unbiased abnormal returns. We introduce and prove a proposition that the optimal forecast error is an unbiased estimator for abnormal return. The proposition induces assessing the performance of abnormal return estimation to preemptively evaluate the out-sample forecast accuracy of the model employed. In an illustrative dataset, we examine various models. The approach requires preliminary computational effort; however, it is useful for accurately obtaining the abnormal return predictions.

show abstract

Section: Introductionmentioning

confidence: 99%