The turn of the year effect is an empirical regularity which refers to the observation that equity returns on days around the turn of the calendar year are significantly high. A common explanation for the effect is the tax-loss selling hypothesis. Three studies have tested this hypothesis by examining equity returns before and after the introduction of the War Revenue Act of 1917. However, these studies have reached inconsistent conclusions. Following these inconsistent findings, this paper examines whether the previous findings are robust to the extension of the pre-tax period, or whether the previous findings are robust to the use of daily data. In contrast to the evidence from existing research which has used daily data, a significant turn of the year effect is found prior to 1917. Furthermore, this effect is shown to be independent of other empirical regularities such as the turn of the month effect and the holiday effect, and seasonality in dividend payments cannot explain the finding. Thus, we conclude that the tax-loss selling hypothesis cannot offer a complete explanation of the turn of the year effect.