Contrast effects: The phantom of an analyst's latest earnings forecasts

Psychology research on contrast effects suggests that information from a previous decision may be compared with the information of the current task. We document a new stylized fact that an analyst's current annual earnings forecast error of one firm is negatively correlated with her latest forecast revision of another firm. We tease out contrast effects as the main driving force of this negative relationship by excluding other potential mechanisms such as firms' information environment and analyst constraints. We show that variables which may influence analysts' retrieval of past experiences have significant impacts on the magnitude of contrast effects, as predicted by the memory-based valuation model.