Whether the information effect of dividends actually exists has been clarified through empirical analysis. However, when viewed in light of all theories, we become skeptical and negative about the information effect of dividends. The reasons are as follows. First, this is attributable to the low-level stable dividend policy. Because firms have incentives to suppress dividends in order to increase internal reserves, they tend to maintain a stable low dividend payout ratio. When such a low-level stable dividend payout policy becomes dominant, the payout ratio may be raised when earnings are expected to increase, and conversely, it may be lowered when earnings are expected to decrease. Since dividend payout ratios are frequently changed in accordance with fluctuations in earnings, dividends serve an informational role in predicting a firm's future earnings. However, if, for example, the dividend payout ratio is kept constant in order to lower the payout ratio despite deteriorating performance, the payout ratio becomes extremely high, and this is adjusted ex post, reducing the likelihood of providing new information to investors. Second, there is the phenomenon of dividend payout ratios received among firms within the same industry. Even within the same industry, dividend payout ratios may be identical or different, and it is difficult to use dividend payout ratios as signals to distinguish differences in future earnings. Firms with poor performance, or those expected to perform poorly, attempt to maintain the dividend payout ratios of other firms in the same industry, which can be understood as a so-called moral hazard phenomenon. Third, the negative view of dividends as information is due to the existence of diverse information sources other than dividend information. Needless to say, for dividends to serve a role as information or signals, it is premised that appropriate information about a firm's future performance and earnings is not available to investors through other means. As described above, changes in dividend policy tend to be ex post adjustments, but even if changes are made without looking at future performance, the degree to which such dividend policy changes can predict outcomes could also be predicted through various other types of information. Rather, from such information sources, it would be possible to obtain higher-quality predictive information that cannot be obtained from dividend information alone.