Resumen
This paper aims to investigate the effect of financial analysts? recommendations on the overconfidence and over or under-reaction to previous years? earnings, as well as their impact on investment decisions in the Tunisian stock market. Literature mostly turned out that a positive bias in analysts? forecasts overreacted to prior earnings changes. Our study is based on the assumption that overconfidence among analysts can be understood through the accuracy of their forecasts, but also it is detected by the way that analyst provides a clear recommendation or not. The analysis employs a panel regression models using annual and bi-annual data over the period 2010-2015. Empirical results show that analysts on the Tunisian stock market are too confident in their forecasts on average, and there is clearly an overall over-reaction to past earnings changes. However, self-confidence is greater for those forecasts that are equipped with a recommendation, when the over-reaction is greater for the not equipped forecasts.Keywords: Overconfidence; Overreaction; Analysts? recommendations; Earnings forecasts.JEL Classifications: G02,G11,G17,G21