This paper examines prediction errors of various equity valuation models using price multiples. The models examined include price/earnings, price/ forecasting earnings, price/book value, price/sales, price/cash flow, price/intrinsic value, price/EBITDA, EV/sales, EV/EBITDA models. Price multiples concerned were calculated using median, harmonic mean of the multiples of the comparable firms in the same industry group.
Of these models, EV/EBITDA model has been found to outperform other models in almost all cases, followed by price/book value model, and intrinsic value model using residual earnings. The price/earnings model that is widely used in practice does not seem to perform well compared with other models. However, when forecasted earnings are used in place of historical earnings, the performance of the price/earnings model substantially improved. These results do not change for the analysis by year and industry. The result of the paper presents some policy implications. First, for IPO firms, the current regulation uses price multiple methods using only earnings and book value. However, accounting to our results, a more comprehensive set of multiple models needs to be considered. Secondly, forecasting earnings needs to be incorporated in calculating price/earnings ratio.
