Accounting for defined-benefit pension plans is complex, and reported financial statement amounts are significantly impacted by a myriad of assumptions. In its interpretative release FR-72 (SEC 2003), the SEC called for more informative and transparent Management Discussion and Analysis (MD&A) disclosure of critical accounting estimates (CAE), including those regarding pension plans. This paper uses a random sample of 147 firms with relatively large defined-benefit pension plans to analyze firms' MD&A pension-related critical accounting estimate disclosures.
We find that 60 (61) percent of our sample firms quantify the effect on pension measurements—primarily pension expense—of a given change in discount rates (expected asset returns). The median effect on earnings per share of these CAE-disclosing firms is between two and three cents. Firms rarely disclose the effects of potential changes in future salary assumptions or other estimates on pension measurements. While FR-72 encourages MD&A disclosure of information to assess the past accuracy of or predict future changes in critical accounting estimates, few firms provide such information with respect to their pension plans, suggesting avenues for improvement in disclosures. Finally, we use logistic regression to analyze the determinants of firms' decisions to disclose the sensitivity of pension expense to pension discount rate or expected asset return assumptions. Results indicate that the likelihood of providing a discount rate or expected asset return CAE is positively related to firm size, having a Big 4 auditor, and the variability of pension plan funded status, and is lower for firms operating in regulated industries and for firms with better-funded pension plans.