中文摘要：

本文从分析师盈余预测修正的角度,考察了管理层业绩预告发布后短窗口内分析师预测行为的变化。无论是基于公司层面还是分析师层面样本的研究均表明,业绩预告发布后,分析师盈余预测修正与业绩预告消息显著正相关;并且,当业绩预告为好消息、点估计形式以及可靠性高时,分析师盈余预测修正幅度与业绩预告消息的正相关关系更显著。拓展性检验还发现,管理层业绩预告有助于降低分析师盈余预测的误差,即分析师预测是朝着更准确的方向进行修正。上述证据表明,管理层业绩预告为市场参与者提供了额外的有关公司预期盈余的信息,分析师可以据此形成更为准确的盈余预测,进而降低资本市场信息不对称。

英文摘要：

In capital market, security analyst plays a dual role of information provider and information user. With their professional knowledge and information gathering advantage, analysts use relevant information of listed companies to make their earnings or prices forecast. Analysts＇ information source includes public and private information. Public information is known to all analysts, and private information is unique to individual analyst. Compared to private information, public information acts as an important information source because of its low cost. Management earnings forecast is one type of public information to the analyst. This article investigates into the effects of management earnings forecasts on analyst behavior. Particularly, we look at the analyst forecast revisions during the 10 days after the management forecast announcement. Both the firm-level and the analyst-level analyses show that analyst forecast revisions increase significantly with the forecast news. Furthermore, we find the positive relationship between analyst forecast revisions and forecast news are more profound for management forecasts with good news, in topic point form, and with greater reliability. Additional tests illustrate that management earnings forecasts are also helpful to decreasing analysts＇ earnings forecast errors. Collectively, our findings suggest that management earnings forecasts provide useful information to market participants, which make analysts are able to generate a more accurate forecast, and hence reduce the degree of information asymmetry in the capital market. Our paper contributes to the literature in the following two aspects： First, we provide more direct evidence for a causal relationship of management earnings forecast and analysts＇ behavior. We use a method such as event study which better controls the cross-section differences of omitted variables and other confounding effects. Second, this paper enriches the existing literature on analyst behavior. We illustrate that how does analyst use firm