Assessing financial accounting measurement alternatives for assets and liabilities

This thesis develops an approach to investigate the accounting for balance sheet line items. The current agenda of the Financial Accounting Standards Board (FASB) indicates that questions of how to measure assets and liabilities for financial reporting are of major interest to the accounting profession. The research provides evidence on criteria established by the FASB for selecting among alternatives. Implications of measurement error for regression results are used to compare alternatives to determine which most closely reflects the measure assessed by investors when valuing the firm. While the approach is applicable to many accounting and reporting issues, it is implemented by investigating a specific, major accounting measurement issue: employers' accounting for pensions. A cross-sectional approach is used.;The thesis makes three important contributions. First, it develops a new perspective on using cross-sectional valuation models in empirical accounting research. A valuation equation and a model of the measurement errors inherent in book values and candidate alternatives vis-a-vis the unobservable market values specified by the valuation equation are used to estimate differences in measurement error variance among accounting alternatives considered. Econometric measurement error is a fundamental feature of the research design. Second, the results may interest those making financial accounting measurement choices. The pension accounting controversy is described and the results interpreted to provide evidence on it. While several measures of pension assets and liabilities are significant in explaining firm market value, significant differences among the alternatives are found. For the entire sample, the fair value of plan assets and accumulated benefit obligation evidence significantly less measurement error variance than other alternatives. The projected benefit obligation evidences less measurement error variance on subsamples where the salary progression rate includes productivity. These results suggest investors include future salary progression in the pension liability but view the projected benefit obligation measure as inherently noisy. Disclosed measures also evidence less error in levels than recognized measures. Third, it documents a significant association between stock prices and information disclosed but not recognized in financial statements.;Characteristics of firms adopting the pension standard early are also explored.