Paul André

Abstract

Special items, earnings announcements and divergence of analysts' beliefs

This thesis examines the consequences of a recent change in Canadian accounting recommendations with respect to the classification of special items (e.g., restructuring charges, asset sales/writedowns). Since January 1990, section 3480 of the CICA Handbook limits below-the-line extraordinary items to items that do not depend primarily on decisions by management or owners. Prior to the change, managers had flexibility in presenting special items above-the-line (as unusual items) or below-the-line (as extraordinary items). Following the change, most special items are presented above-the-line as atypical and infrequent items. In effect, a dimensionality reduction from two classification choices (extraordinary, unusual) to one (atypical and infrequent) occurred. Whether this resulted in an information loss to financial analysts is the thesis' focus of enquiry.

Specifically, the thesis examines the impact of current earnings announcements under the new classification regime on divergence of financial analysts' beliefs about upcoming earnings. This thesis contrasts two views as to the impact of the change. The information signalling view suggests an increase in divergence of beliefs for firms reporting special items in the post-change period, if managers' classification choices with respect to special items in the pre-change period are viewed as informative. The no-effect view suggests no significant change in divergence from one period to the other, if analysts perceive that managers are using the classification choices inconsistently or for opportunistic earnings management purposes in the pre-change period.

Empirical results are consistent with the no-effect hypothesis. The change in classification rules for special items under section 3480 did not result in greater divergence of analysts' beliefs as implied by the signalling hypothesis. However, a potential explanation for the no-effect result is suggested by the earnings announcement environment. Findings indicate that special items are widely discussed in the preliminary earnings announcements both in the pre-change and post-change periods. Furthermore, results are consistent with a decrease in divergence of beliefs when special items are discussed in the press releases period. Hence, it is possible that managers may have been able to undo the potential effects of the reporting change (the information loss) by detailed discussions of their special items in the post-change preliminary earnings announcements. Consequently, the exact nature of accounting recommendations with respect to the classification of special items may not be an issue for sophisticated users such as financial analysts as long as special items are discussed in the earnings announcements.