We investigate whether and how the complexity of derivatives influences analysts' earnings forecast properties. Using a difference-in-differences design, we find that, relative to a matched control sample of non-users, analysts' earnings forecasts for new derivatives users are less accurate and...