Accounting for the Measurement Bias: A Study of Market Efficiency in the United States and the Relevance of Extensive Fundamental Analysis in Equity Valuation

Sammanfattning: This thesis investigates abnormal returns over the period 1983-2021 from an investment strategy that is based on public accounting information. Investment positions are taken in US manufacturing firms and are held for 36 months using a self-financing (hedged) portfolio. The strategy is comprised of (1) an accounting-based prediction model estimating the probability of an increase in future return on owners' equity (ROE), and (2) a consideration of the market's expectation of the development in ROE using the residual income valuation (RIV) model. Investment positions are taken when (1) and (2) differ, enabling an assessment of whether market prices incorporate the statistical prediction. Horizon values in the RIV model are estimated using both a 'sophisticated method' based on fundamental analysis of the accounting valuation bias and a 'parsimonious method' based on reverse-engineering horizon values through stock prices. By deploying these two methods, the utility of a theoretically sound approach of estimating horizon values can be assessed through a comparative evaluation with an approach that holds practical relevance due to its simplicity. The investment strategy utilizing fundamental analysis of horizon values generated a 25.8% return above the S&P 500 index and an equivalent abnormal CAPM return of 11.4%. The results point to a discontinuance of market mispricing over time, and the mispricing in earlier periods is found to be sensitive to the choice of abnormal return metric. Regardless, estimating horizon values through fundamental analysis resulted in more significant and prolonged returns compared to when using reverse-engineering through stock prices. This indicates that utilizing fundamental analysis when estimating horizon values should not be neglected in future research of market efficiency, nor amongst practitioners.