"Value Relevance of Earnings Levels and Changes: Evidence from Manufacturing Firms in Japan---Part II Industry-Period Analysis---" (in Japanese)

The value relevance of earnings is generally affected by macro-economic environments and micro-economic competitive conditions. Therefore, the relevance differs by industries and by periods. In order to examine the variation in the relevance directly, simply and clearly, it is necessary to estimate the relevance of earnings in each industry and each period. In that way, this paper investigate the value relevance of earnings levels after the preparatory study on the relevance of the changes in earnings. We also try to find the regularity in the variation through analyzing the estimated relevance by non-parametric and semi-parametric methods. The four major findings in this research are as follows. First, the changes in earnings are consistently value-relevant over the sample periods (through March 1979 to March 2000) across almost all industries. We cannot find the evidence that the bubble economy nor the following recession diminishes the relevance of earnings. Second, there is an asymmetric relation between the relevance of increase and decrease in earnings. However, the latter is not always transitory and less value-relevant than the former. Contrarily, the decreases in earnings seem to be more persistent and persuasive during the recent depression in Japan. Third, while the relevance of ordinary income levels is consistently related to the relevance of changes in ordinary income, we cannot find the systematic relation between the relevance of levels and changes as to operating profits and net income. As for operating profits and net income, it seems that the unknown different factors affect on the relevance of levels and changes respectively. Fourth, the results cannot support the S-shaped relation hypothesis that the relevance of earnings experiencing more large surprises is relatively lower, which is often detected in prior studies. We find the J-shaped relation that earnings experiencing large positive shocks is more value-relevant in the quiet period (through March 1979 to March 1985).