Abstract The article discusses implications of a study conducted on the idea of classificatory smoothing of income statement statistics other than net income. The study was investigated on a hypothesis using correlational analysis without data on managements' intentions, it sought only to establish that managements behaved as if smoothing were intended. A central issue is the set of hypotheses to be tested in empirical research on income smoothing. Researchers of the study do not address the primary question of whether income smoothing exists. Instead, they attempt to determine if correlation patterns suggest one sort of smoothing rather than another. However, just the possibility of classificatory smoothing makes apparent some fundamental problems in income smoothing research and in the study of as-if smoothing hypothesis. Those problems, which include correlation and data problems and the issue of management intention, are the subject of this article. Suggestions for improvements in income smoothing research also are made.

Abstract The article deals with comments related to two major issues. Firstly, whatever results are obtained, they are consistent with the hypothesis and secondly, the rarity of the phenomenon under investigation. This might suggest insufficient data on which conclusions can be significantly based or that the subject matter is of a minor importance that it does not warrant a study. The first issue suggests that correlation analysis is not appropriate for tests of smoothing behavior if managements' intentions to smooth are not established independently. The commentor suggests that in the case of extraordinary items, any established correlation, whether positive or negative, of such items with either ordinary or net income statistics would point to a possible smoothing behavior, but of different types. Regarding the second issue, authors point out that extraordinary items incorporate three different variables namely nonrecurring expense adjusted for average tax rate, nonrecurring expense net of taxes and nonoperating expense adjusted for the average tax rate derived from basic data items.