Research of data assets disclosure and analyst forecasts: Evidence from text analysis

In the era of big data, data has gradually become a new production factor and plays a key role in the market competition. Therefore, lots of data resource in firms has been qualified to be recognized as assets. With the rise of digital economy, such “data assets” have been widely existed in different industries. Moreover, the national big data strategy was released at the Fifth Plenary Session of the 18th Central Committee of the Communist Party of China in October 2015, which further highlights the growing importance of big data as a strategic asset on China′s development agenda. Therefore, data assets become increasingly important to firms in the digital economy. Firms with more information on data assets will help better assess the firm value, such as improving the accuracy of analysts′ forecasts. According to the relevant laws and regulations of information disclosure in China, data asset information disclosure is usually a voluntary off-balance-sheet information disclosure provided mostly in the form of text. In addition, with the implementation of the registration system of stock issuance in China′s A-share market, the importance of information intermediaries has also increased. Above all, this paper finds that it is necessary to investigate that whether more textual disclosure on data asset by firms can contribute to the forecast accuracy of analysts under the current situation. To be specific, this paper uses the China′s A-stock listed firms followed by analysts from 2010-2018 as the sample and builds the text dictionary by Word2Vec neural network algorithm for the text information mining to empirically study the impact of firms′ data assets information disclosure frequency on the coverage extent and accuracy with which analysts track and forecast their earnings. To further explain the above viewpoints, this paper investigates two potential channels for analysts applying data asset information in information processing. Finally, this paper further tests the influence of textual readability of firm information disclosure and the state of the market to the relationship between higher frequency of data asset disclosure and analysts′ forecast accuracy, respectively. In the first part, this paper begins with the investigation of whether higher frequency of data asset disclosure can attract more analyst coverage, and then tests the relationship between data asset disclosure frequency and earning forecast error of analysts. We find that the higher disclosure frequency of data assets information in the annual report of individual stocks, the more forecasting reports about next year′s earnings of the stocks from analysts, and the lower forecast error of the earnings level. These results indicate that analysts pay attention to data assets information, and data assets information disclosure can significantly improve the accuracy of analyst forecasts. In the second part, this research empirically finds that if firms lack voluntary earnings preannouncement behavior or poor lag earning information quality, the textual data asset disclosure of firms can be more helpful to improve analysts′ forecast accuracy. According to the results, data assets information disclosure can not only help provide forward-looking information, but also help increase the information transparency of firms. This paper provides evidence for the data asset disclosure to improve analysts′ forecast accuracy. In the third part, this paper attempts to further analyze some situations which may possibly influence the effects of data asset disclosure frequency on analysts′ forecast accuracy, such as textual readability in firms′ information disclosure and the state of the market (i. e. bull or bear market). This research finds that when the individual stock information provision shows higher readability, or the market is a bull market, the higher frequency of data assets information disclosure, the better analysts′ forecasts accuracy. In conclusion, with the rise of digital economy, the efforts of many firms aiming to break “isolated data island”, fully realize the data mining and facilitate data sharing make more and more data resource qualified to be recognized as assets. Hence, sufficient provision on data assets will contribute to completing the information environment of individual stocks, and further help promote analysts to better play the role of information intermediaries. This paper finds that more textual voluntary disclosure about data assets by firms can reduce forecast errors of analysts from multiple channels, which can support that analysts′ forecast has gradually paid attention to data asset information and efficient disclosure about data asset is obviously valuable to these information intermediaries. Besides, according to results shown in the study, textual information readability should be concerned for firms while providing data asset information, and analysts should appropriately apply data asset information considering the situation of bull and bear market. Above all, this paper holds that for individual stocks, the effort of providing and perfecting corporate textual information disclosure behavior on data asset is undoubtedly necessary, while for information intermediaries, such as analysts, should pay more attention to collecting and efficient processing of data assets information when making earnings forecasts in era of big data.