Abstract ABSTRACT: Prior research has shown that time-series models were able to make better predictions of prospective earnings by utilizing segmented data. Those results suggest that humans may be able to utilize segmented data to improve their forecasts. This study used analysts' forecasts to address this latter issue. Analyst forecast accuracy was evaluated before and after implementation of the SEC's line-of-business disclosure requirements that became effective in 1971. Three sample groups of companies were studied: multisegment firms that first reported segment earnings in 1971, multisegment firms that had voluntarily disclosed segment earnings data prior to 1971, and a group of single-segment firms that continued to report only on a consolidated basis. Analyst forecasts were evaluated for accuracy before and after mandatory segmented earnings reporting with a multivariate repeated measures ANOVA model. While it was found that both the mean and variance of forecast error decreased for all three groups, the most significant change was for multisegment firms without prior segment disclosures.
Bruce A. Baldwin (Sun,) studied this question.