ABSTRACT We examine how shareholders' trust in managers is affected by (i) the outcome of earnings management (inconsistent vs. consistent with shareholders' interests) and (ii) the method of earnings management (accruals vs. real methods). Using a controlled experiment, we predict and find that trust is impaired when the outcome of earnings management suggests that managers have put their interests above shareholders' interests and/or when the method of earnings management suggests that managers misreported the firm's economic performance. We argue that shareholders assess managers putting their interests above shareholders' interests as a signal of untrustworthiness because it involves a transfer of the firm's resources away from shareholders to managers. We argue that shareholders also assess managers' use of accruals to manage earnings as a signal of untrustworthiness because, in this instance, managers misreport the firm's economic performance. Finally, we show that trust mediates the combined effects of the outcome of earnings management and the method of earnings management on investment decisions. Our study incrementally contributes to the literature by highlighting the adverse implications of managers' use of accruals to manage earnings even when its outcome serves shareholders' interests.

Abstract Earnings items are typically classified in financial reports based on their persistence and measurement subjectivity. Archival research examines investors' use of persistence and measurement subjectivity classifications for forecasting and valuation. However, this research typically examines only one of these classifications at a time and ignores the potential interactive implications of an earnings item's persistence and measurement subjectivity classifications. We recruited experienced financial analysts to participate in two experiments that examined the effect of measurement subjectivity classifications on analysts' use of persistence classifications when forecasting earnings items. We find that analysts rely less on an earnings item's persistence classification when measurement subjectivity is high relative to when measurement subjectivity is low. We also find that presentation format affects analysts' use of these two classifications. Specifically, we find that the matrix format (i.e., rows display persistence classifications and columns display measurement subjectivity classifications) facilitates analysts' combined use of persistence and measurement subjectivity classifications relative to the sequential format (i.e., the classifications are displayed separately). These findings suggest that archival research could improve its examination of market participants' use of earnings classifications for forecasting and valuation by recognizing that the implications of an earnings item's persistence classification can vary according to the item's measurement subjectivity classification. By also demonstrating how presentation format affects analysts' use of earnings classifications, our study provides further insights into this fundamental issue in accounting research and standard setting.