CMStatistics 2017: Start Registration
View Submission - CFE
A1160
Title: Confirming signals are hard to resist: Blessing and curse of information under confirmation bias Authors:  Stefanie Schraeder - UNSW (Australia) [presenting]
Abstract: According to empirical evidence, individuals pay more attention to confirming than to contradicting information. In the context of this confirmation bias, we study the effects of additional information on perception correctness -- contrasting the competing effects of total signal precision and the possibility to search for the most suitable signal. We provide the testable hypothesis that managers report bad news in a more diffuse signal compared to good news. This, in turn, provides a rationale for the dispersion anomaly: dispersed analysts' earnings forecasts are followed by stock under-performance. Then, we include the confirmation bias in an overlapping generations model with a continuous signal distribution. Several results of more simplified models do not hold any longer. For instance, confirmation bias leads to underreaction instead of overreaction. A momentum effect is accompanied by various time-varying market participation, volatility, trading volume, and market depth effects.