"Relevance of Earnings Levels and Surprises: Evidence from Manufacturing Firms in Japan" (in Japanese)

The purpose of this paper is to investigate the value relevance of earnings by two steps. First, we identify the factors that introduce transitory earnings into the reported earnings. Next, we reexamine the value relevance of earnings by controlling those factors. In the first step, we investigate the earnings changes model, testing whether earnings changes are associated with stock price changes. Especially we focus on whether the relation between earnings changes and stock price changes is affected by the sign and size of earnings surprises. This research provides major four results as follows. First, since negative earnings changes contain more transitory earnings, the coefficient on the reported earnings for firms decreasing earnings is lower than for firms increasing earnings. Second, even after controlling the negative earnings changes, the coefficient on losses is very small, or not significantly different from zero. Losses are not value relevant, excluding a few exceptional years. Third, our empirical results reject the hypothesis that large earnings changes contain more transitory earnings (S-shaped relation). On the contrary, the large positive earnings changes seem to be more persistent, due to the competitive advantage, than the small changes. The coefficient on earnings levels for firms experiencing the big earnings surprises is higher than for others (J-shaped relation). Fourth, to control three factors ] losses, negative changes, large positive changes ] improves greatly the relevance of the reported net income. This shows that it is also effective, for estimating the permanent earnings, to control the generation factor of transitory earnings as well as classifying the components of earnings.