Explaining cross -sectional differences in unrecognized net assets

This thesis examines sources of cross-sectional differences between market value and book value of equity, i.e., unrecognized net assets (UNA ). Two sources are identified: (1) the existence of positive net present value projects and the ability of a firm to appropriate and sustain the abnormal profits from these projects, and (2) accounting recognition rules that conservatively measure income and assets. The sources of UNA are denoted as competitive structure and accounting recognition respectively. The analysis consists of forecasting future book return on equity (ROE) and explaining UNA.;I conduct two empirical studies. The first study uses a broad sample of Compustat firms. The empirical results show that industry concentration and capital intensity are not strong predictors of future ROE, and often negatively affect UNA. Market share is a strong predictor of future ROE and has a positive relation with UNA for low growth firms. Firms with slower accounting recognition are found to have more persistent ROE series, and have higher valuation multiples on current ROE, irrespective of their growth rate.;The second study focuses on pharmaceutical firms. The theoretical model relates UNA to ROE and R&D. The empirical evidence supports the four hypotheses. First, pioneering drug manufacturers have a higher coefficient on ROE and lower coefficient on R&D than generic drug firms. Second, firms with larger therapeutic market shares have a higher coefficient on ROE and a lower coefficient on R&D. Third, firms with more patents per dollar R&D investment have a higher valuation multiple on scaled R&D. Finally, higher growth in R&D positively affects the ROE valuation coefficient, and does not affect the R&D multiple.;The dissertation contributes to the literature on UNA in three ways. First, I simultaneously address competitive structure and accounting recognition. The finance and industrial organization literature largely ignored the accounting recognition dimension of UNA, while the accounting literature only recently started to consider the economic context in which accounting data are generated. Second, I provide new evidence on the empirical validity of a residual income model similar to that proposed by Ohlson (1995). Third, while prior literature primarily focused on the sign of the valuation coefficients on both accounting and nonaccounting information, I examine the magnitude of these coefficients.