Valuation Using Industry Multiples: How to Choose the Most Relevant Multiples
Our research focuses specifically on the methodology to be applied to improve the relevance of the multiples-based valuation method regarding the identification of the most relevant multiples (i.e., that reduce the relative absolute valuation error within any industry-based peer group). In line with prior empirical studies, our results confirm that Enterprise Value multiples based on EBIT and EBITDA perform better, compared to Sales and Capital Employed, and that multiples based on forward-looking EBIT and EBITDA are more relevant compared to corresponding actual earnings. In the absence of forward-looking earnings, available at the date of valuation, our study shows that EBITDA multiples provide better estimates than EBIT multiples do. Beyond these general results, the approach implemented in our research can be easily reproduced by practitioners (e.g., financial analysts, M&A advisors, independent appraisers) to identify case-by-case the multiples that are the most relevant within any industry-based peer group.
The multiples method, also referred as the guideline public company or transaction method, is becoming more widely used to value any company in most valuation contexts (such as financial analysis, mergers and acquisitions, or fairness opinions provided by independent appraisers). This “popularity” can be explained by its relative “simplicity” compared to the discounted cash-flows method and by its “objectivity” because it is based on the value of comparable companies than can be directly observed on the market. However, researchers and practitioners agree that it is little studied, often misunderstood, and sometimes misapplied (Crow, Gibbs, and Harms 2001).
This article presents the results of a study that compares the relevance of different valuation Enterprise Value multiples (EV multiples), also referred to as Market Value of Invested Capital multiples (MVIC multiples). Beyond the results of general application, we show how the recommended approach is able to improve the efficiency of the evaluation process and therefore the relevance of the values obtained by the multiples method.
Theory and Practice
It is customary to consider only three approaches in valuing a business: (1) income based, (2) market based, and (3) asset or cost based. These approaches are available in practice in a set of three of the most common valuation methods: (1) the discounted cash-flows method (DCF), (2) the multiples method, and (3) the net revaluated asset method (suitable for real estate companies or industrial companies externalizing low profitability).
Most surveys show that the DCF method and the multiples method derived from comparable listed companies are predominantly used by practitioners (Chastenet and Jeannin 2007; Harbula 2009). Confirmed by current practice and increasingly addressed in valuation textbooks (Hitchner 2006; Koller, Goedhart, and Wessels 2007), the multiples method is also considered by many valuation and accounting standards (FASB 2001; ISVC 2001). Its ease of use and its direct reference to market valuation of comparable listed companies tends to obscure the fact that the multiples method is based on solid theoretical foundations.
Theoretical background
The DCF method and the multiples method rely on common theoretical foundations, in that they can be directly related to the theoretical model of discounted cash flows, whose authorship can be attributed to Williams (1938) and Modigliani and Miller (1958, 1961). The discounted dividend method (DDM) was initially derived by the former from the theory of interest sets by Fisher (1907) and applied to the valuation of equity shares. The DCF method owes much to the latter authors, the founders of modern finance, Nobel Laureates in Economics in 1990 and 1985 respectively, in that it leads the evaluator to distinguish the stream of cash flows generated by the assets of the company and the stream of cash flows generated by its shares.
To apply the DCF method one should determine the future free cash flows that may be generated by these assets (regardless of its financing structure). The evaluators then use the weighted average cost of capital and the capitalization formula of Gordon and Shapiro (1956) to discount these cash flows to perpetuity. The value obtained by applying this method is the so-called debt-free/cash-free value of the company, also called the Enterprise Value.
The multiples method is notably based on the principle of market efficiency developed by Fama (1970), which states that stock prices reflect more or less all the information available to investors: All things being equal, the multiples method assumes that the market is efficient enough to value similar assets at similar price (Esty 2000). In other words, when the multiples method is applied for valuing the shares of a privately held company whose shares are not listed, the method is assumed to provide the value that the business would have if it was itself listed.
According to the theoretical model of discounted free cash flows, the multiples method assumes that it is possible to identify publicly traded companies, said to be comparable, in that they have a profile of free cash flows, growth prospects, and a level of risk similar to those of the company that is being valued. Considering the single-period model derived from the Gordon and Shapiro (1956) formula,1 a simplification of the discounted free cash-flows method, any synthetic valuation estimated multiple (Mest) derived from a sample of comparable companies can be assimilated to a capitalization factor applied to the expected free cash flow (CF) of the company to be valued (considering its discount and growth rates, hereafter referred as k and g):
where Mest = 1/(k−g).
Any estimated value derived from the application of that estimated valuation multiple necessarily yields a valuation error corresponding to the difference between the company’s estimated value (Vest) and its observed market value (Vobs):
where Error = Vobs – Vest.
Multiple-based valuation in practice
The multiples valuation method consists in applying multiples derived from the observed value of a sample of comparable companies (also referred as a “peer group”), in the form of synthetic multiples corresponding to the arithmetic mean or median, to the free cash flow of the company to be valued.
In practice, to calculate valuation multiples, the appraiser will use financial aggregates (published or estimated) that are the most representative proxy of both the company's and its peers’ ability to generate free cash flows over a long period: EBIT (earnings before interest and tax) and/or EBITDA (earnings before interest, tax, depreciation, and amortization) should prevail and to a lesser extent Sales (or Revenues) and Capital Employed (or Invested Capital).
Under these conditions, the value of any business can be expressed when it is observed on the market for a listed company or estimated in the form of a multiple applied to these financial aggregates (FA), considering the following generic formula:
where FA = λ × CF (that is, said to be proportional to CF).
The apparent simplicity and explicability of the multiples method is certainly the source of its success. It is thus possible to evaluate any privately held company by applying synthetic valuation multiples representative of the valuation level of a sample of comparable listed companies. Those multiples can also be used to measure a trading premium/discount of any listed company relative to its industry peers.
To identify these comparable listed companies it is customary to (1) consider companies within the same industry as the company to be evaluated, also referred as “industry peers”; then, within these industry peers, to (2) select those that are the most comparable in terms of profitability and growth prospects. To calculate those synthetic valuation multiples, it is customary to (1) consider financial aggregates directly issued from the financial statements of both the company and its peers, such as Sales, EBITDA, EBIT, and/or Capital Employed and (2) calculate the median or average of the multiples of the referred the “industry peer group.”
That being said, appraisers are often using their “professional” judgment, more or less formalized, in regard to both the selection of comparable listed companies and the choice of valuation multiples.
The apparent simplicity of that method and the place accorded to judgment lead some academics and professionals to believe that this method can be misapplied, resulting in a risk of unexpected valuation error and/or voluntary manipulation of the result of the evaluation.
Three questions are being raised by the community of appraisers:
-
What are the financial aggregates that should be used to calculate valuation multiples (Capital Employed, Sales, EBITDA or EBIT, actual, trailing, or forward-looking)?
-
What are the key performance indicators that should be used to select the most comparable listed companies within a peer group based on industry-based primary selection?
-
Should multiples be combined and with what weighting?
We propose here some answers to the first question, in the form of general results and a methodology that could be applied in each case by appraisers and that would prevail on professional judgment to choose the most relevant valuation multiples.
Related Literature
The relative performance of multiples has been regularly addressed in empirical research on the multiples method. Kim and Ritter (1999) examined the use of several multiples for valuing IPOs and found that the use of the price to earnings (P/E) multiple based on forecasted earnings leads to a superior accuracy relative to those of multiples based on book value, trailing earnings, cash flows, and sales. Baker and Ruback (1999) compared the performance of multiples based on EBITDA, EBIT, and sales. Their results show that industry-adjusted EBITDA performs better than EBIT and sales. Liu, Nissim, and Thomas (2002) studied the performance of a list of value drivers and found the general ranking for multiples: (1) forward earnings, (2) historical earnings, (3) cash flows and book value, and (4) sales.
Lie and Lie (2002) arrived to the following conclusions regarding the relative performance of multiples: (1) the total enterprise value/book value multiple generally yields better value estimates than sales and earnings multiples (total enterprise value/EBIT and total enterprise value/EBITDA), especially for financial but also for nonfinancial firms; (2) the use of forecasted earnings instead of trailing earnings improves the estimates of the P/E multiples; and (3) the EBITDA multiple generally yields better estimates than the EBIT multiple, except for pharmaceutical companies.
Liu, Nissim, and Thomas (2007) discovered for their sample a higher accuracy of earnings than those of operating cash flows and dividends. After the shift from reported numbers to forecasted ones, they obtained higher valuation performances for all multiples. They revealed that earnings still had the best performance in this case because their accuracy increased more than those of the other two measures.
These studies are generally based on U.S. and/or international data.
Herrmann and Richter (2003) examined different multiples (P/E, EV/EBIT, EV/EBITDA, P/B [price to book], EV/Invested Capital, and EV/Sales) for European nonfinancial firms. They asserted, consistent with the results of Liu, Nissim, and Thomas (2007), that (1) multiples based on earnings lead to the highest prediction accuracy and (2) sales multiples yield the lowest prediction accuracy. However, they discovered that the P/B multiple performed better than the EV/EBITDA multiple when comparable firms were selected based on ROEs and earnings growths instead of only the industry membership.
Another study that used European data and investigated the relative performance of different multiples is that of Schreiner (2007). He found that (1) equity value multiples outperform entity value multiples, (2) knowledge-related multiples outperform traditional multiples in science-based industries, and (3) forward-looking multiples outperform trailing multiples. The trailing multiples from Schreiner (2007) are in the following decreasing order from the viewpoint of accuracy: earnings multiples (P/E, P/Earnings before tax, P/EBIT, and P/EBITDA), cash-flows multiples (P/Operating cash-flow and P/Dividends), book value multiples (P/IC, P/B, P/Total assets), and gross income and sales multiples.
Data and Research Methodology
Since the market price of comparable listed companies can be considered one of the repositories that may be useful to investors in valuing a business (among others, intrinsic value derived from the DCF or the net revaluated asset methods), the multiples method is even more relevant if it reduces the difference between the company’s estimated value and its market value as if it was listed, or its observed value as it is itself listed.
Sample companies and data
The study we conducted compares the relative absolute valuation error (RAVE) resulting from the application of the most common valuation multiples for a sample of listed companies within industry peer groups.2 To set the sample, we used data from FactSet: market capitalization, actual financial statements (i.e., invested capital and net debt), and financial analysts’ Sales, EBITDA, and EBIT earnings estimates consensus.
As of March 31 of each year, we selected firm-years that satisfy the following criteria:
- (1)
Western European companies disclosing consolidated financial statements in IFRS GAAP
- (2)
All required data being nonmissing and non-negative for fiscal years 2006, 2007, and 2008 (i.e., all multiples can be observed and are positive)
- (3)
Each industry-year peer group including at least five companies
- (4)
All multiples lying within the 5th and 95th percentiles of each fiscal year.
The resulting sample includes 919 observations in 2006 representing 24 countries and 59 industries, 1,090 in 2007 (24 countries and 66 industries), and 1,192 in 2008 (27 countries and 70 industries).
Valuation multiples used for the study are those that appear the most frequently considered by financial analysts by reference to the studies of Chastenet and Jeannin (2007) and Harbula (2009): Enterprise Value (EV) to Sales, EBITDA, and EBIT (supplemented by the companies’ Capital Employed multiples: EV/CE) where financial aggregates refer to actual (subscript: 0), current (subscript: 1), and forecast (subscript: 2) data, as reported by financial analysts, both for the former (actual) and the latter (current and forecast), and consolidated by FactSet in its consensus.
Relative absolute valuation errors
For any company within a peer group, we defined the valuation error resulting from the application of industry-based multiples as the difference between its estimated value (Vest) and its observed value (Vobs) and the relative absolute valuation error (RAVE) as the absolute ratio between this difference and the company’s observed value:
Alternatively, any RAVE can be calculated by comparing any estimated multiple (Mest) and its corresponding observed multiple (Mobs):
For any company within a peer group, the lower the RAVE resulting from the application of a multiple is, the more the multiples method is relevant. It is thus possible to compare the relevance of different multiples by comparing the RAVE resulting from their application. Within a peer group or a sample of comparable companies the most relevant synthetic valuation multiples (EV/Sales, EV/EBITDA, or EV/EBIT) are those that minimize the average RAVE within the sample.
All things being equal, one can consider that within a peer group, the valuation multiple that should be used to identify mispriced companies would be the one that minimizes the RAVE of the companies within the peer group. All things being equal, one can consider that for a privately held company comparable to those composing the peer group, the more relevant multiples would be those that minimize the RAVE of the companies within the peer group.
To calculate the RAVE of each company within industry peer groups we followed the following steps:
- (1)
Setting industry peer groups by grouping companies from the same industry according to the best FactSet industry classification (i.e., level 2).
- (2)
Calculating the observed valuation multiples of each company (considering available data as at March 31 of each year, 2006, 2007, and 2008): market capitalization, net debt (short-term and long-term debt less cash and long-term investment), actual, current, and forecast Sales, EBITDA, and EBIT given by financial analysts’ consensus (see Table 1: descriptive statistics of multiples, as at March 31, 2008).
Table 1Distribution of Multiples and Related Relative Absolute Valuation Errors (RAVE), European Peer Groups, March 31, 2008
Calculating the synthetic multiples of each industry peer group, corresponding to the arithmetic or the harmonic mean of the observed multiples of their constituent companies.
- (4)
Calculating the RAVE of each company multiple, by comparing its observed value to its estimated value for each of the synthetic multiples previously calculated.
- (5)
Ranking the sample companies’ RAVE according to different statistics (quartiles, median, interquartile range, and mean of RAVE).
Table 1 presents the descriptive statistics of the sample for the multiples and RAVE as at March 31, 2008, and the ranking of multiples according to the different statistics (Figure 1 provides a graphical presentation of the distribution of RAVE).3
Citation: Business Valuation Review 34, 4; 10.5791/0882-2875-34.4.173
As part of academic work that resulted in the article, the validity of the results of the empirical study was based on achieving statistical parametric pairwise comparison of means, variances, and proportions (with similar results for the three samples studied as at March 31, 2006, 2007, and 2008).
Empirical Results
Our study allows us to classify the most commonly used valuation multiples according to their level of relevance (i.e., their ability to reduce the RAVE) and thus provide the appraisers with a set of general results that are listed below:
- (1)
For the calculation of peer group synthetic multiples the harmonic mean should be preferred to the arithmetic mean (based on Liu, Nissim, and Thomas 2007, an alternative to the use of the harmonic mean to reduce the valuation bias is the use of the median of the multiples).
- (2)
Appraisers should favor the use of EV to earnings multiples (EBITDA or EBIT) compared to multiples of Sales or Capital Employed (as earnings can be considered as better estimates of companies’ cash flows).
- (3)
When they have them, appraisers should give preference to the most forward-looking multiples (current and/or forecast) (as forward-looking earnings can be considered better estimates of companies’ future expected cash flows).
- (4)
When they do not have forward-looking multiples (often the case when valuing privately held companies), appraisers should generally favor the use of EBITDA over EBIT multiples (as EBITDA would be the best estimates of cash flows; this last result is confirmed for almost 75% of the industry peer groups in our sample).
Beyond these general results (some of which have already been published in the literature), one of the interesting aspects of the RAVE analysis is that it can be implemented by any appraiser in any valuation cases to identify and justify the preferred use of one multiple rather than another.
Case Studies
We address the particular case of the luxury and the cosmetics industries considering the valuation multiples observed as of April 2011. Table 2 and Figure 2 and Table 3 and Figure 3, respectively, show the descriptive statistics of multiples and RAVE for the peer groups that are most representative of the selected industries.
Citation: Business Valuation Review 34, 4; 10.5791/0882-2875-34.4.173
Citation: Business Valuation Review 34, 4; 10.5791/0882-2875-34.4.173
At the date of valuation (end of April 2011), in the luxury sector, considering both the mean and median of RAVE (16% and 12%, respectively), the multiple based of forecast EBITDA was the most relevant. In the cosmetics industry, considering both the mean and median of RAVE (6% and 5%, respectively), the forward-looking EV/EBIT was more relevant than the corresponding EV/EBITDA multiple. In that industry case, the relevance of EBITDA for the purpose of valuation based on the industry-based multiples method could not be considered as the most relevant choice.
Our two case studies demonstrate that the general results of our study can be reversed for some industries (when comparing the relevance of the EV/EBIT and EV/EBITDA multiples, for example). However, the RAVE approach can be used case by case and in many valuation contexts (e.g., M&A, Financial Analysis, Fairness Opinions, Impairment Tests).
Conclusion
Our study reveals that many of the most common valuation multiples do not all have the same level of relevance when one looks at the relative absolute valuation error (RAVE) that derives from their application. Our empirical results are of general consideration but need to be checked case by case. We show how the use of RAVE can be adopted in any valuation study.
The objective of this article is to recall that the multiples method, although it appears simple, is based on solid theoretical foundations like the DCF method and requires specific tools to improve the efficiency of the evaluation process based on a value relevance measure to mitigate the risks of unexpected errors or voluntary manipulation of results of valuation (especially regarding the selection of the multiples). The best multiple should be the one that minimizes RAVE within any peer group.
In this article, we do not address other issues attached to the multiples method raised in other empirical studies: what criteria should be used to select companies that are really the most comparable within an industry peer group (e.g., market segment, profitability, and/or growth prospects)? How can one justify the combination and the weighting of the valuation results derived from the use different multiples (such as Capital Employed and EBIT, or EBITDA and EBIT)?
Distribution of Relative Absolute Valuation Errors (RAVE), European Industry Peer Groups, March 31, 2008. Notes: The lower (higher) bound of the vertical line corresponds to the 1st (3rd) quartile of the RAVE distribution. When the rectangle in the middle of the line is gray, its bottom is the median and the top is the mean of the RAVE distribution; when it is dark, it is the opposite.
Distribution of Relative Absolute Valuation Errors (RAVE), Luxury Industry Peer Group, April 30, 2011. Notes: The lower (higher) bound of the vertical line corresponds to the 1st (3rd) quartile of the RAVE distribution. When the rectangle in the middle of the line is gray, its bottom is the median and the top is the mean of the RAVE distribution; when it is dark, it is the opposite.
Distribution of Relative Absolute Valuation Errors (RAVE), Cosmetics Industry Peer Group, April 30, 2011. Notes: The lower (higher) bound of the vertical line corresponds to the 1st (3rd) quartile of the RAVE distribution. When the rectangle in the middle of the line is gray, its bottom is the median and the top is the mean of the RAVE distribution; when it is dark, it is the opposite.
Distribution of Relative Absolute Valuation Errors (RAVE), European Industry Peer Groups, March 31, 2008. Notes: The lower (higher) bound of the vertical line corresponds to the 1st (3rd) quartile of the RAVE distribution. When the rectangle in the middle of the line is gray, its bottom is the median and the top is the mean of the RAVE distribution; when it is dark, it is the opposite.
Contributor Notes
Edouard Chastenet is an associate professor at the Business Administration Institute (IAE) of Université de Lyon, France. Alain Marion is a professor at the Business Administration Institute (IAE) of Université de Lyon, France.