We explore whether income tax accounting (ITA) for pensions provides measurement benefits incremental to U.S. GAAP. We use publicly reported Form 5500 defined-benefit pension expense and funding measures, which are calculated using ITA rules, and contrast them with parallel GAAP amounts. We document that ITA measures are more strongly associated with market value, future cash contributions, the cost of equity capital, and credit ratings than comparable GAAP measures, particularly when ITA better maps to economic fundamentals and when GAAP offers greater managerial discretion. Using intraday returns and disclosure times obtained through a Freedom of Information Act request, we also find that markets quickly react to the release of ITA measures in Form 5500. Our study provides novel empirical tests of theory about the benefits of ITA using actual, not imputed, ITA measures and provides evidence that investors can and do use the unique information in a publicly disclosed tax form.

Investors demand conditional conservatism to restrict managers' ability to opportunistically exploit unverifiable accounting estimates. The fair value estimation process is subject to verifiability concerns when market prices are unavailable and, thus, susceptible to managerial discretion. We explore whether banks' exposure to less-verifiable fair value estimates is associated with conditional conservatism. General and bank-specific conservatism measures indicate that banks with greater proportions of less-verifiable fair value assets exhibit more conditional conservatism. Cross-sectional analyses provide evidence that this relation varies predictably with investor-demand and manager-supply proxies. Further analyses indicate that monitoring institutional investors drive the demand for conservatism. We identify high-quality auditors and board independence as two mechanisms used to invoke conservatism. Findings are robust to the exclusion of fair value earnings components, suggesting that the effect is not confined to fair value accounts. Together, our results indicate that less-verifiable fair value estimates generate demand for conditional conservatism in the financial industry.