Abstract: Online media releases a large amount of financial news every day, attracting people to read and share it. With its advantages of timeliness and wide information coverage, online media has surpassed traditional financial paper media and gradually become an important source for people to obtain information about listed firms. As key information intermediaries in the financial market, professional financial analysts, who play a pivotal role in mitigating information asymmetry between listed companies and market investors, must utilize all available information to improve their forecasting accuracy. However extant literature rarely discusses how public market information affects the earnings forecasting behavior of professional financial analysts from a psychological perspective. Compared to traditional financial paper media, online media is deficient in stringent and effective regulatory measures. Firstly, to ensure the timeliness of information, online media may rely on multiple information sources of varying quality, resulting in a mix of information in financial news. Secondly, under the increasingly fierce competition, online media will do whatever to attract public attention for the sake of profits or to keep up with rivals, even disregarding conflicting information in financial news. The deluge of online media news poses challenges to analysts' ability to capture and process information, with such information processing consuming a significant amount of their time and energy. Affected by the psychological constraint of limited attention, analysts tend to rely on quick and simple intuitive cognition in earnings forecasting analysis, resorting to heuristic decision-making, which reduces the accuracy of their earnings forecasts. Online media coverage data used in this paper are sourced from the CNDRS database, which includes 20 mainstream online financial media and more than 400 other major websites, industry websites, or local websites such as Jinrongjie, Hexun, Huaxun, Sina Finance, Phoenix Finance, Sohu Finance, etc., including stock news, macroeconomic reports, industry reports and more. Based on the perspective of limited attention, this paper empirically examines the impact of online media news on the accuracy of analysts' earnings forecasts by using individual analysts' research reports and listed company data during 2007-2020. The empirical results show that online media news can significantly reduce the accuracy of analysts' earnings forecasts, which still holds in the endogeneity tests of IV methods and exogenous shocks. The mechanism tests confirm that online media news undermines the accuracy of analysts' earnings forecasts due to the influence of limited attention, thereby ruling out the information mechanism and the noise mechanism. The heterogeneity tests indicate that the reduction effects of online media news on analysts' forecasts accuracy are more significant for non-state-owned enterprises, listed companies with low information transparency, those with high book-to-market ratio, and those covered by non-star analysts. Our study not only theoretically explore analysts' forecasting behavior from a psychological perspective, hence, enriching the literature on its influencing factors, but also provides empirical evidence in practice for regulatory authorities to strengthen supervision over online media, improve news information quality of online media, and urge analysts to enhance their information discernment capabilities and improve forecasting accuracy. The online media coverage data utilized in this paper is derived from the CNDRS database, as it is not possible to directly observe how analysts access online media news data during their earnings forecasting process. The black box of analysts' information acquisition thus remains a topic requiring further in-depth investigation in future research.

Key words: Analyst Earnings Forecast, Online Media Coverage, Limited Attention, Heuristic forecast deviation