In this study benchmark error is tested for as a source of the small firm effect by comparing the results from ordinary least squares and instrumental variable methods. Although the instrument is not perfect, results show that benchmark error could be a cause of the overall (all months) small firm effect. Results from the instrumental variable method indicate that large January abnormal returns are still present, but that they are offset by negative non‐January abnormal returns. As a result, the instrumental variable results show that there is no longer a significant overall “effect,” merely a seasonal effect. It is also found that the results are not sensitive to the choice of the market index.