Disentangling Reliability from Relevance (Job Talk)

Vast accounting literature investigates relevance of financial reporting using the relation between stock returns and earnings. However, these tests are “are joint tests of relevance and reliability” (Barth, Beaver, & Landsman, 2001). This study addresses a recent FASB’s call to disentangle reliability apart from relevance (SFAC 8, BC3.30, 2010). We utilize reliability metric to extend value-relevance framework in two settings. In the first setting, we find that losses are both less reliable and less relevant than profits. Furthermore, we report that reliability and relevance are complements for profits only. Conversely, reliability does not enhance the usefulness of reported losses because relevance is too low. In the second setting, we find that intensive use of accounting estimates harms reliability and enhances the relevance of reported earnings. Our findings also indicate that in estimate intensive reporting, reliability and relevance are complements. However, using a small number of estimates with low reliability does not harm the usefulness of reported earnings. Overall, empirical evidence in both settings suggests that high reliability does not compensate for insufficient relevance. The results expand our understanding how reliability and relevance of accounting information influence the usefulness of reported earnings and offer guidance to standard-setters.