Evaluating Methods to Estimate the Implied Cost of Equity Capital: A Simulation Study

We evaluate accounting-based methods to estimate the implied cost of capital using a simulation approach. We simulate a model economy in which the true cost of capital is known and calibrate it to the CRSP-Compustat universe. We then compare the true cost of capital to the implied cost of capital estimates from ten different methods proposed in the literature in terms of bias, accuracy, and their correlation with the true cost of equity capital. Methods based on the residual income model perform better than those based on the abnormal earnings growth model. Methods that estimate the cost of capital and expected growth simultaneously work reasonably well if they rely on analyst forecasts instead of ex post realized values, even if analyst forecasts are biased. We suggest combined methods that are chosen so that the distortions from individual methods compensate each other and show that some simple combinations outperform all individual methods