Empirical Analysis of Accounting Estimates When Predictions are Poor

Abstract

This study empirically analyzes the reliability of a significant class of accounting estimates by focusing on forecast errors in estimated pension returns for the period 2001-2010. These estimates have a major impact on the reporting of pension costs and income for even the largest and most solvent of plan sponsors, and hence they are a good example of the effect of accounting estimates on financial reporting. We find substantial evidence that these estimates are frequently not a faithful representation of actual returns in a given year, management’s reasonable ranges concerning them are poorly calibrated, the resulting accounting risk from forecast errors is unacceptably high in many instances, and, consequently, auditors cannot provide reasonable assurance on them. This has major implications for the use of these and similar estimates in financial reporting and accounting standards, and their verification by auditors.