Relative Valuation – Accuracy of Corporate Valuation Using Multiples

Abstract: This paper investigates the accuracy of three different methods that are used to perform a relative valuation. I also study the valuation accuracy of the three most widely used multiples used within these methods, the enterprise value to EBITDA (EV/EBITDA), the forward-looking price per share to earnings per share (P/EPS1), and the price-book (P/B). In addition, the difference between industries regarding valuation accuracy is investigated. A sample containing 246 companies listed on S&P Europe 350 during 2013-2015 is used. Method 1 uses comparable companies based on industry membership, and harmonic mean as a statistical measure to generate a multiple. Method 2 uses six companies from the same industry, that are similar in a relation to a profitability measure, using harmonic mean as a statistical measure to generate the multiple. Method 3 uses comparable companies based on industry membership and regression as a statistical measure to generate a multiple. I find that Method 2 provides the lowest valuation errors when used with the EV/EBITDA multiple and the P/B multiple. However, for the P/EPS1 multiple, all methods provide similar results. Method 1 provides the second best results, and Method 3 delivers the worst results. Regarding the multiples, I find that P/EPS1 deliver consistently best results, the EV/EBITDA comes second, and the P/B provides the worst results. Finally, the industrials and the consumer services industries have the lowest valuation error, and the financials industry highest valuation error.