Credibility and cheap talk of securities analysts:theory and evidence

This paper studies how investors react to public messages that may be optimistically biased. We first construct a communication game between an investor and a (possibly) biased securities analyst. We find an equilibrium characterised by the following properties: first, the investor reacts more to bad news than to good news, and second, the difference in this reaction is higher when the investor has a greater prior suspicion that the analyst is a biased type. We then use nonparametric techniques and a large database of earnings forecasts to test these predictions, and find that the evidence supports them. Lastly, we use our empirical strategy to discriminate between the causes for analysts’ bias.