Equity valuation using accounting numbers : a cross-industry analysis of equity valuation : theory and practice

Abstract

This study presents an industry-level comparison of multiples-based and flows-based equity valuation models for a dataset of 1,251 unique firms from the United States, in the period 2006-2011. The purpose is to investigate which model performs best, and if there are significant differences across industries. Models are compared in terms of signed and absolute prediction errors, as well as the explanatory power of their estimates, vis-à-vis contemporaneous stock prices. Across industries, forward price-toearnings and an abnormal earnings growth model with no terminal value are rated as the models with best fit in predicting market prices, on a risk-adjusted basis. This analysis is complemented with a robustness assessment of peer selection criteria, finding that Coarsened Exact Matching (CEM) – a systematic and more refined approach to identify comparable firms – yields superior results for multiples-based models than traditional peer selection methods. Further, it allows for measuring of the level of firm proximity. The study concludes by comparing theoretical results with models used in practice by financial analysts in the United Kingdom, documenting inconclusive results for the two sampled industries – pharmaceutical and utilities.