The purpose of this paper is to examine whether the seasonal anomaly known as the reverse weekend effect detected at index level can also be observed at individual stock level.
This paper's methodology is based on the model first developed by Connolly and then employed by Chang, Pinegar, and Ravichandran in which returns are regressed against the dummy variable for Monday. In addition, the conditional variance is also included into the mean equation following Engle, Lilien, and Robins. Given the increasing evidence that equity returns are conditionally heteroskedastic, the paper includes in the conditional variance the lag of the squared residual from the mean equation (i.e. autoregressive conditional heteroskedasticity term introduced by Engle) and the previous period's forecast variance (i.e. the generalized autoregressive conditional heteroskedasticity term introduced by Bollerslev). Also, the paper controls for the different impact of good and bad news on the conditional variance following Glosten, Jaganathan, and Runkle.
It is found that the anomaly is widely distributed among large firms, not just confined to a few firms. The finding suggests that the anomaly at the index level is not driven by the extreme returns of a few firms. The paper also finds that the anomaly at the firm level is not evenly distributed across the weeks of the month. Furthermore, trading volume and illiquidity of individual firms can only partially explain the seasonal anomaly.
This paper extends the study of the reverse weekend effect in individual firms.
CitationDownload as .RIS
Emerald Group Publishing Limited
Copyright © 2011, Emerald Group Publishing Limited