Information disclosure is the institutional basis of the healthy development of the capital market. With the increasing improvement of investors’ ability, information disclosure based solely on historical operations cannot fully meet their information needs. Consequently, CSRC revised the information disclosure principles in 2007, which required listed companies to provide outlooks and forecasts on their future development in annual reports. Forward-looking information can forecast a company’s future development strategies, business plans, earnings risks, and opportunities and challenges, which contains a significant value for investors to estimate future cash flows and risks. Compared with historical information and quantitative information, it is more obscure and complex, so it requires more professional ability to interpret such information. As professional information users and important information intermediaries of the capital market, analysts have been proven to have advantages in information mining and interpretation. Therefore, can analysts effectively mine and interpret forward-looking information?

Taking A-share listed companies from 2007 to 2020 as the sample and using text analysis and machine learning to measure forward-looking information disclosure, this paper investigates the above question. The results show that the higher the frequency of forward-looking information disclosure, the higher the analysts’ earnings forecast accuracy. In firms with high earnings quality and readability, the above positive relationship is more significant. The mechanism test indicates that forward-looking information disclosure can improve the earnings forecast accuracy by alleviating information asymmetry and increasing analysts’ site visits. Further research also reveals that forward-looking information can successfully predict firms’ fundamental changes in the future, which indicates that such disclosure has its information value and can provide incremental information for analysts and other users of the capital market, which in turn can improve the information efficiency of the capital market.

The possible contributions of this paper are as follows: First, based on the perspective of analysts’ earnings forecast, it expands the research on the economic consequences of forward-looking information disclosure. Currently, there is no academic consensus on whether such disclosures are informative, and the findings can promote the understanding of this issue. Second, from the forward-looking disclosure perspective, it enriches the literature on the influencing factors of analysts’ earnings forecasts. Research from abroad suggests that forward-looking disclosure can improve the quality of analysts’ earnings forecasts, but whether the conclusion applies to China’s capital market has not been verified. Therefore, this paper supplements the empirical evidence from China. Third, it has certain enlightenment for listed companies to regulate forward-looking information disclosure, for regulators to deepen the reform of capital market information disclosure, and for analysts to focus on forward-looking information in annual reports when making earnings forecasts.