Relative Valuation Methods for Equity Analysis
Much research3 in corporate finance and similar academic disciplines is tilted toward the use of discounted cash flow (DCF) methods. However, many analysts also make use of relative valuation methods, which compare several firms by using multiples or ratios. Multiples that are commonly used for such purposes include price/earnings, price/book, and price/free cash flow.
Relative valuation methods implicitly assume that “similar” firms are likely to be valued similarly by investors. Therefore, on average, we would expect that firms that are generally comparable are likely to trade at similar multiples, in terms of price/earnings, price/book, or various other metrics. If this assumption is approximately correct, then relative valuation methods can be used to identify firms that look “cheap” or “expensive” relative to their peers. When a particular firm’s multiples are extremely different from the rest of the universe, this may indicate a potential investment opportunity—though further analysis will likely be required to determine whether there are reasons why such a firm is valued differently from other companies that otherwise appear comparable.
The basis of relative valuation methods is to use one or several ratios to determine whether a firm looks “cheap” or “expensive” by comparison with generally similar firms. Relative valuation methods do not attempt to explain why a particular firm is trading at a particular price; instead, they seek to measure how the market is currently valuing multiple companies, with the underlying assumption that the average multiple for a group of companies is probably a reasonable approximation to overall market sentiment toward that particular industry. In other words, relative valuation work assumes that on average, the share prices of companies in a particular universe are likely to trade at similar multiples relative to their own financial or operating performance. Baker and Ruback (1999) provide a more formal presentation of these concepts. However, it is important to realize that at any particular time, some firms are likely to be trading at higher or lower multiples than would be justified under “fair value.”
Making effective use of relative valuation methods does require careful selection of “similar” companies. Sometimes this is relatively simple, for instance when an analyst is dealing with industries where there are a large number of roughly homogeneous firms providing goods or services that are approximately equivalent. However, sometimes there can be considerable difficulties in identifying “similar” companies, particularly if the firms under consideration are unusually idiosyncratic in terms of their product mix, geographical focus, or market position. In this entry, we will provide some tentative guidance about how to build a universe of comparable companies. However, ultimately this part of the process will depend on the skill and knowledge of the individual analyst; two different experts may pick different sets of “similar” firms, and thus generate different values from their relative valuation analysis.
BASIC PRINCIPLES OF RELATIVE VALUATION
Analysis based on relative valuation requires the analyst to choose a suitable universe of companies that are more or less comparable with one another. There is no standardized approach concerning how to choose such a universe of similar firms, and the process relies to some extent on an analyst’s personal judgment concerning the particular industry and geography involved. However, it is possible to lay out some general principles that combine practitioners’ insights with the results of academic inquiry.
Sources of Data
Relative valuation approaches can only be employed if there is sufficient information, produced on an approximately consistent footing, about the various companies that are the subjects of analysis. In most countries, companies that are publicly listed on stock exchanges are required by law and regulation to report their historical results publicly in a timely manner, or risk being delisted from the exchange. (There may be occasional exceptions to this general pattern, particularly for entities that are majority owned or controlled by their home country government. But such anomalies are not frequently observed except during crisis periods.) Consequently, it is almost always possible to obtain information about listed companies’ historical results. However, multiples based solely on historical data may not provide a complete picture, as most analysts would probably agree that forward-looking estimates are likely to provide more useful insights into the market’s opinion of a particular company (Valentine, 2011, p. 261).
Investment banks, rating agencies, and other firms can provide estimates of a firm’s future earnings, revenues, and other metrics, typically over the next two or three years. Various data providers such as Bloomberg or Thomson Reuters collect such information and use it as the basis for “consensus” estimates, which can be viewed as representing the market’s general opinion of a company’s future prospects. It is also possible to use a firm’s own in-house estimates for the companies under coverage, as these may incorporate insights that are not yet reflected in current pricing. However, for precisely this reason, in-house estimates should be used as a supplement rather than as a replacement for consensus figures.
It is conventional to consider more than one year of data, as there may be disparities in how the market is valuing results in the immediate future and in the slightly longer term. However, it is often difficult or impossible to obtain consensus estimates more than two or three years into the future. Consequently, relative valuation approaches generally focus on relatively short periods into the future, rather than seeking to gauge how the market is valuing expected performance five or ten years hence. (In this respect, relative valuation analysis can be viewed as somewhat limited by comparison with DCF approaches, which typically give considerably more attention to the relatively distant future.)
Number of Comparable Firms
In general, an analyst would like to use data from other firms that are as similar as possible. However, if the criteria for “similarity” are specified too stringently, then there may be too few firms included in the universe. And if the sample is too small, then the idiosyncrasies of individual firms may exert an excessive influence on the average multiple, even if the analyst focuses on the median rather than the mean when calculating the “average” multiple.
Generally speaking, we believe that it is desirable to have at least five or six comparable companies, in order to begin drawing conclusions about relative valuation for a particular industry. Conversely, there may be few benefits from considering more than 12 companies, particularly if the larger universe contains firms that resemble less closely the particular company that is the focus of the analyst’s attention.1 For most practical purposes, a group of between six and 12 comparable firms should be sufficiently large to produce usable results.
Basis for Selecting Comparable Firms
In an ideal situation, a universe of comparable companies would be similar in terms of size, industry focus, and geography. This tends to be easier when considering small or mid-sized firms—say, with market caps between $100 million and $10 billion (based on 2010 U.S. dollars). Firms that are below this size limit, in other words microcap stocks, may be more difficult to use for relative valuation purposes. Even if these firms are public, they may receive less coverage from research analysts, who typically are more interested in companies that are large, liquid, and already owned by institutional investors (see Bhushan, 1989).
Conversely, it can also be difficult to perform relative value analysis on companies with relatively high market capitalization. Many large firms are dominant players in their particular market niches, in which case they may be more likely to trade at a premium reflecting their higher degree of market power. Alternatively, large firms may be effectively a conglomerate of numerous smaller entities, each engaged in a specific activity, and there may be no other large or small firm that produces an approximately equivalent blend of goods and/or services.
When attempting to assess the relative value of firms that are large and/or complex, it can often be useful to assess “relative value” using two separate approaches. The first approach is to consider the firm as a complete entity and try to find other firms that are at least somewhat comparable in terms of size and complexity, even if their business mix is not precisely identical. In such cases, it can often be useful to consider similar firms that may be located in other countries, even though their different geographical positioning may affect their level of risk and thus the multiples at which they trade. The second approach is to use a sum-of-the-parts valuation method, which will be discussed in more detail later in this entry.
Geography and Clientele
Differences related to geographic location can affect the extent to which companies can be viewed as broadly similar. For instance, in the United States public utilities are predominantly regulated at the state level, and the public utility commissions in one state may operate quite differently from their counterparts elsewhere. Consequently, a public utility operating in one state may not be directly comparable with a public utility located in another state. In recent decades, there has been a wave of acquisition activity in the U.S. utility industry, so that now some utilities have operations in multiple states. In such instances, the valuation placed on a utility will presumably incorporate investors’ perceptions of the regulatory environment affecting each of its state-level operations. For relative value purposes, a group of multistate public utilities may not be very similar to a public utility that is operating in only one state.
Regional differences in regulatory regimes may only affect a subset of companies. However, firms in the same industry may well have quite different client bases and geographic exposures. For instance, one retailer may aim to sell a wide range of goods to a mass-market client base at the regional or national level, while another retailer might instead focus on selling a limited number of luxury products to the most affluent members of the global population. These two firms are likely to have substantially different product quality, cost bases, profit margins, and sensitivity to macroeconomic conditions. In particular, retailers of luxury goods to a global client base may have developed brands that transcend national borders, and a high proportion of their current and future revenues and profits may come from outside their home country. Under such conditions, it is possible that a suitable universe of comparable companies might include at least a few foreign firms, particularly if they have similarly broad geographic reach.
In past decades, analysts focusing on U.S. firms would probably have only rarely used foreign firms in their analysis of “comparable companies.” However, as both U.S. and foreign firms have become increasingly globalized, and as accounting standards around the world have gradually started to become more similar, we believe that for some types of relative value analysis, there may be benefits to including firms that are generally comparable in terms of size and product mix, even if their legal headquarters are not located in the United States. For more insights into these issues, see Copeland, Koller, and Murrin (2000, Chapter 18).
Many companies have “depositary receipts” in other markets, such as ADRs. Consensus estimates may be available for a firm’s local results and/or its depositary receipts. The estimates for the depositary receipts may be affected by actual or expected movements between the currencies of the two countries, which may bias the analysis. We therefore recommend that when calculating figures for companies that are listed in different countries, all multiples should be consistently calculated in terms of local currency throughout, in order to ensure that anticipated or historical currency fluctuations will not affect the results. A substantial number of non-US companies have a share price quoted in one currency, but report their financial results in another currency; to avoid potential mismatch-related errors in such cases, it may be prudent to convert all numbers into a single numéraire such as the US dollar.
Sector and Industry Characteristics
Some academic research has examined different ways of selecting a universe of comparable firms. Bhojraj, Lee, and Oler (2003) compared the effect of using four different industry classification methods, and concluded that at least for a universe of U.S. securities, the Global Industry Classification Standard (jointly developed and maintained by Standard & Poor’s and Morgan Stanley Capital International) appeared to do the best job of identifying firms with similar valuation multiples and stock price movements. Chan, Lakonishok, and Swaminathan (2007) compared the effect of using industry classification schemes with statistically based clustering approaches, and found that examining stocks in terms of industry membership seemed to give better explanatory power than working in terms of either sectors or subindustries. To our knowledge, there have not been any parallel investigations into the effectiveness of different industry classification schemes for cross-national analysis. The results of Phylaktis and Xia (2006) suggest that the importance of sector-level effects has been increasing in recent years, while the influence of country-level effects has waned slightly.
Technology and Intraindustry Diversity
As discussed above, some academic research has suggested that firms from similar industries tend to trade at similar multiples and to experience similar stock price movements. Industry membership therefore would seem to be a useful starting point for analysis. Thus, for instance, trucking companies and railroad companies both provide transportation services, but railroads will generally trade at different multiples from trucking companies because their cost structure and balance sheets tend to be quite different.
In some cases, there can be substantial variation even within a particular subindustry. For instance, “publishing” covers a wide variety of different business models, including daily newspapers, weekly magazines, publishers of textbooks and professional journals, printers of fiction or nonfiction books, and suppliers of financial data. Each of these individual industries is likely to have different sources of revenue, different technological requirements, different cost structures, and different rates of expected growth. Admittedly, the larger publishing houses may have operations spanning several different fields, but the relative contributions of each division to the firm’s overall revenues and profits may differ substantially. In such instances, relative value analysis may result in a wide range of valuation multiples, possibly with several different clusters reflecting each firm’s competitive position. We consider such difficulties in the next section.
There are also some industries in which technological differences are the principal basis on which relative values are assigned. For instance, small companies in the field of biotechnology may have only a handful of products, each of which could potentially be a great success or a dismal failure. Some companies of this type may be still at the prerevenue stage when they go public, so that their valuation is entirely based on the market’s expectations about the ultimate value of technology that has not yet generated actual sales. In such instances, relative value analysis might require particularly careful selection of companies that are truly comparable in terms of the market’s perception of their stage of development and the likelihood that their key products will ultimately be successful. Arguably, relative value analysis in such cases may not generate particularly useful results, because the spread of potential outcomes is so broad.
Bimodal and Multimodal Patterns
Sometimes the outcome of a relative value analysis will show that the valuation multiples are not evenly spread between low and high, but instead are bimodal or multimodal—in other words, there seem to be two or more clusters of results. We show an example of this in our hypothetical example below, which suggests that in a universe of seven firms, two are expected to achieve a return on equity (ROE) of 11% to 12% in FY0 and FY1, whereas the other companies are generally projected to deliver an ROE of 8% to 9%. Such differences may appear relatively minor, but if the market really does expect these outcomes, then the two companies with higher profitability may legitimately be expected to trade at a premium to their peers.
When a relative valuation table appears to have bimodal or multimodal characteristics, an analyst will generally be well advised to investigate further. In any given sector or industry, there may well be some firms that are truly capable of producing higher returns than their peers, perhaps as a result of better management, a stronger market position, or a more supportive regulatory environment. Relative valuation methods can identify potential outliers of this type, but cannot test whether the estimates themselves are reasonable.
One potentially useful approach is to extend the analysis further back into the past, using historical prices for valuation purposes, and if possible also using as-was projections for the relevant period. Such projections are now widely available from various different data vendors, including Bloomberg, FactSet, and Thomson Reuters. Consider the companies that are currently trading at a premium or a discount to their peers—did they also trade at a discount to their peers in the past? A logical extension of relative value analysis based on a single period is to gauge whether a particular firm persistently tends to trade at a lower or higher multiple than its peers, and then assess whether its current multiple is above or below what would be expected on the basis of prior periods. Damodaran (2006, Chapter 7, p. 244) notes that relative valuations frequently have low persistence over time. For industries in which this is the case, then relative valuation methods may indeed provide useful investment signals.
Choice of Valuation Multiples
Many relative valuation methods compare a company’s share price with some measure of its performance, such as earnings per share (EPS) or free cash flow per share. Other relative valuation methods compare a company’s share price with some measure of its size, such as book value per share. Block (1999) has reported that the majority of practitioners consider that when analyzing securities, measures of earnings and cash flow are somewhat more important than measures of book value or dividends. However, many practitioners will make use of various metrics in their work, in the expectation that the different multiples will provide varying perspectives. Liu, Nissim, and Thomas (2002) compared the efficacy of six different metrics for relative valuations of U.S. firms on a universe-wide basis. Liu, Nissim, and Thomas (2007) extended the analysis to seven different metrics applied to 10 different countries and multiple industries. Hooke (2010, Chapter 15) presents an example using eight different metrics applied to the single industry of temporary staffing companies. In a hypothetical example below, we use three different metrics for relative valuation analysis, and we believe that most practitioners would consider that between three and six different metrics is probably justifiable. It is certainly possible to have a much larger number of metrics (see Damodaran, 2006, p. 650), but the results may be harder to interpret.
A ratio such as price/earnings can be calculated in terms of share price/EPS, or alternatively can be interpreted as market cap/net income. For most purposes, these two ratios will be the same. However, share issuance or buyback activity may impair the comparability of figures expressed in terms of EPS. If there is any possibility of ambiguity, then we would generally recommend using market cap/net income.
For instance, a company may currently have 100 million shares outstanding, a current share price of $40, and expected earnings of $2 in FY0 and $3 in FY1. If the P/E ratio is calculated in terms of price/EPS, then the FY0 ratio is 20 and the FY1 ratio is 13.3. However, analysts may be expecting that the company will buy back and cancel 20% of its shares during FY1. If so, then the projected net income in FY1 would presumably be $240 million rather than $300 million. If the P/E ratio is calculated using market cap and net income, then the FY1 ratio would be 16.7 rather than 13.3. This hypothetical example indicates the importance of ensuring that the denominator is being calculated on a basis that reflects the historical or projected situation for the relevant period. (An investor might consider that if a firm’s management is indeed strongly committed to buying back its own shares, then this might indicate that the firm’s management views the shares as being undervalued. However, such considerations would presumably be included as a qualitative overlay to the relative valuation analysis.)
Choice of Numerator: Market Cap versus Firm Value
In some instances, the choice of numerator may have a significant impact on the multiple. For instance, many analysts will use price/sales ratios for valuation purposes. However, a firm’s revenues are generated from the total of its capital base, comprising both equity and debt.
Consider two companies, A and B, which both have a current market cap of $300 million and projected annual revenues of $600 million in FY0, so that they both have a current price/sales ratio of 2. But suppose that Company A has no outstanding borrowings, whereas Company B has net debt of $300 million. One could argue that Company B is actually rather less attractive than Company A, as apparently it requires twice as much capital to generate the same volume of sales. In effect, analyzing the company in terms of “firm value/sales” rather than price/sales would reveal that Company B is actually making less efficient use of its capital than Company A.
There is no single definition of “firm value” that is generally accepted by all practitioners. In an ideal world, one would want to have the market value of the firm’s equity capital and of the firm’s debt capital. However, because corporate bonds and bank loans typically are not traded in liquid markets, there may not be any reliable indicator of the market value of debt capital. Consequently, it is conventional to use market capitalization to estimate how investors are valuing the firm’s equity capital, but then to use figures from the firm’s most recent balance sheet together with the notes to the financial statements as a proxy for net debt. The broadest definition of which we are aware is the following:
In practice, for most firms, the biggest components of net debt are likely to be total short-term debt, total long-term debt, and cash and equivalents. In most cases, using an alternative definition of firm value will often have only a small impact on the calculated multiple.
Conceptually, it is possible to divide the income statement between the line items that are generated on the basis of total capital, and those that pertain solely to equity capital. For most firms, the separator between these two categories is Net Interest Expense or Net Interest Income. Analyzing relative valuation for banks and insurance companies can be somewhat more complex, as discussed in Copeland, Koller, and Murrin (2000, Chapters 21 and 22). Generally speaking, it is usually desirable that the numerator and denominator of a valuation metric should be consistent with each other (Damodaran, 2006, pp. 239–240).
Table 1 Hypothetical Relative Valuation Results
Industry-Specific Multiples
Analysts covering some industries may make use of information specific to that industry, such as paid miles flown for airlines, same-store sales for retailers, or revenue per available room for hotel chains. Such data can provide insights into how the market is valuing individual firms’ historical or expected operating performance. However, we consider that they should be viewed as a supplement to other multiples, rather than as a replacement for them, for two reasons: because it can be difficult to reconcile a company’s operating performance with its financial results, and also because there may be little or no intuition about what would be a “reasonable” estimate for long-run valuation levels (Damodaran, 2006, Chapter 7, pp. 237–238). Natural resource producers tend to be valued in terms of both their operating efficiency and the resources that they control, so it may be useful to include some measure of their reserves in the analysis (Hooke, 2010, Chapter 21). Many practitioners make use of efficiency metrics when using relative valuation approaches to assess some types of banks and other lending institutions (Hooke, 2010, Chapter 22).
HYPOTHETICAL EXAMPLE
Suppose that an analyst is seeking to gauge whether Company A is attractive or unattractive on the basis of relative valuation methods. Suppose that the analyst has determined that there are six other listed companies in the same industry which are approximately the same size, and which are also comparable in terms of product mix, client base, and geographical focus.2 Based on this information, the analyst can calculate some potentially useful multiples for all seven companies. A hypothetical table of such results is shown in Table 1. (For the purposes of this simple hypothetical example, we are assuming that all the firms have the same fiscal year. We will consider calendarization later in this entry.)
In this hypothetical scenario, Company A is being compared to Companies B through G, and therefore Company A should be excluded from the calculation of median and standard deviation, which would otherwise lead to double-counting. The median is used because it tends to be less influenced by outliers than the statistical mean, so it is likely to be a better estimate for the central tendency. (Similarly, the standard deviation can be strongly influenced by outliers, and it would be possible to use “median absolute deviation” as a more robust way of gauging the spread around the central tendency. Such approaches may be particularly appropriate when the data contain one or a handful of extreme outliers for certain metrics, which might be associated with company-specific idiosyncrasies.) The table has been arranged in terms of market cap, from largest to smallest, which can sometimes reveal patterns associated with larger or smaller firms, though there don’t appear to be any particularly obvious trends in this particular set of hypothetical numbers.
The table suggests that the chosen universe of comparable companies may be reasonably similar to Company A in several important respects. In terms of size, Companies B, C, and D are slightly larger, while Companies E, F, and G are slightly smaller, but the median market cap across the six firms is the same as Company A’s current valuation. In terms of P/E ratios, Company A looks slightly cheap in terms of FY0 earnings and somewhat cheaper in terms of FY1 earnings. In terms of P/FCF ratios, Company A looks somewhat expensive in terms of FY0 free cash flow, but only slightly expensive in terms of FY1 free cash flow. And finally, in terms of P/B ratios, Company A looks somewhat expensive in terms of FY0 book value, but roughly in line with its peers in terms of FY1 book value.
Analysis of the Hypothetical Example
So what are the implications of these results? First, Company A looks relatively cheap compared to its peer group in terms of P/E ratios, particularly in terms of its FY1 multiples. Second, Company A looks rather expensive compared to its peer group in terms of P/FCF and P/B ratios, particularly in terms of FY0 figures. If an analyst were focusing solely on P/E, then Company A would look cheap compared with the peer group, and this might suggest that Company A could be an attractive investment opportunity.
However, the analyst might be concerned that Company A looks comparatively cheap in terms of P/E, but somewhat expensive in terms of price/book. One way to investigate this apparent anomaly is to focus on ROE, which is defined as earnings/book value. Using the data in the table, it is possible to calculate the ROE for Company A and for the other six companies by dividing the P/B ratio by the P/E ratio—because this effectively cancels out the “price” components, and thus will generate an estimated value for EPS divided by book value per share, which is one way to calculate ROE.
The results suggest that Company A is expected to deliver an ROE of 10.8% in FY0 and 12% in FY1, whereas the median ROE of the other six firms is 8.7% in FY0 and 8.8% in FY1. Most of the comparable companies are expected to achieve an ROE of between 8% and 9% in both FY0 and FY1, though apparently Company C is expected to achieve an ROE of 11.5% in FY0 and 11.7% in FY1. (A similar analysis can be conducted using “free cash flow to equity,” which involves dividing the P/B ratio by the P/FCF ratio. This indicates that Company A is slightly below the median of Companies B through G in FY0, but in line with its six peers during FY1.)
These results suggest that Company A is expected to deliver an ROE that is substantially higher than most of its peers. Suppose that an analyst is skeptical that Company A really can deliver such a strong performance, and instead hypothesizes that Company A’s ROE during FY0 and FY1 may only be in line with the median ROE for the peer group in each year. Based on the figures in Table 1, Company A’s book value in FY0 is expected to be $15.38, and the company is projected to deliver $1.67 of earnings. Now suppose that Company A’s book value remains the same, but that its ROE during FY0 is only 8.7%, which is equal to the median for its peers. Then the implied earnings during FY0 would only be $1.35, and the “true” P/E for Company A in FY0 would be 14.9, well above the peer median of 12.75.
The analysis can be extended a little further, from FY0 to FY1. The figures in the table above suggest that Company A’s book value in FY1 will be $16.67, and that the company will generate $2.00 of earnings during FY1. But if Company A only produced $1.35 of earnings during FY0, rather than the table’s expectation of $1.67, then the projected FY1 book value may be too high. A quick way to estimate Company A’s book value in FY1 is to use a clean surplus analysis, using the following equation:
Based on the figures in the table above, Company A is expected to have earnings of $1.67 during FY0, and $2.00 during FY1. The implied book value per share is $15.38 in FY0, and $16.67 during FY1. According to the clean surplus formula, Company A is expected to pay a dividend of $0.38 per share in FY1.
Assuming that the true earnings in FY0 are indeed $1.35 rather than $1.67, and that the dividend payable in FY1 is still $0.38, then the expected book value for Company A in FY1 would be $16.35 rather than $16.67. Taking this figure and applying the median FY1 peer ROE, the expected FY1 earnings for Company A would be $1.42 rather than $2.00, and consequently the “true” P/E for FY1 would be 13.9 instead of the figure of 10.0 shown in the table. At those levels, the stock would presumably no longer appear cheap by comparison with its peer group. Indeed, Company A’s FY1 P/E multiple would be roughly in line with Company G, which has the highest FY1 P/E multiple among the comparable companies.
This quick analysis therefore suggests that the analyst may want to focus on why Company A is expected to deliver FY0 and FY1 ROE that is at or close to the top of its peer group. As noted previously, Company A and Company C are apparently expected to have an ROE that is substantially stronger than those of the other comparable companies. Is there something special about Companies A and C that would justify such an expectation? Conversely, is it possible that the estimates for Companies A and C are reasonable, but that the projected ROE for the other companies is too pessimistic? If the latter scenario is valid, then it’s possible that the P/E ratios for some of the other companies in the comparable universe are too high, and thus that those firms could be attractively valued at current levels.
Other Potential Issues
Multiples Involving Low or Negative Numbers
It is conventional to calculate valuation multiples with the market valuation as the numerator and the firms’ financial or operating data as the denominator. If the denominator is close to zero, or negative, then the valuation multiple may be very large or negative. The simplest example of such problems might involve a company’s earnings. Consider a company with a share price of $10 and projected earnings of $0.10 for next year. Such a company is effectively trading at a P/E of 100. If consensus estimates turn more bearish, and the company’s earnings next year are expected to be minus $0.05, the company will now be trading at a P/E of –200.
It is also possible for a firm to have negative shareholders’ equity, which would indicate that the total value of its liabilities exceeds the value of its assets. According to a normal understanding of accounting data, this would indicate that the company is insolvent. However, some companies have been able to continue operating under such circumstances and even to retain a stock exchange listing. Firms with negative shareholders’ equity will also have a negative price/book multiple. (In principle, a firm can even report negative net revenues during a particular period, though this would require some rather unusual circumstances. One would normally expect few firms to report negative revenues for more than a single quarter.)
As noted previously, averages and standard deviations tend to be rather sensitive to outliers, which is one reason to favor using the median and the median absolute deviation instead. But during economic recessions at the national or global level, many companies may have low or negative earnings. Similarly, firms in cyclical industries will often go through periods when sales or profits are unusually low, by comparison with their average levels through a complete business cycle. Under such circumstances, an analyst may prefer not to focus on conventional metrics such as Price/Earnings, but instead to use line items from higher up the income statement that typically will be less likely to generate negative numbers.
Calendarization
Some of the firms involved in the relative valuation analysis may have fiscal years that end in different months. Most analyst estimates are based on a firm’s own reporting cycle. It is usually desirable to ensure that all valuation multiples are being calculated on a consistent basis, so that calendar-based effects are not driving the analysis.
One way to ensure that all valuation multiples are directly comparable is to calendarize the figures. Consider a situation where at the start of January, an analyst is creating a valuation analysis for one firm whose fiscal year ends in June, while the other firms in the universe have fiscal years that end in December. Calendarizing the results for the June-end firm will require taking half of the projected number for FY0 and adding half of the projected number for FY1. (If quarter-by-quarter estimates are available, then more precise adjustments can be implemented by combining 3QFY0, 4QFY0, 1QFY1, and 2QFY1.)
Calendarization is conceptually simple, but may require some care in implementation during the course of a year. One would expect that after a company has reported results for a full fiscal year, the year defined as “FY0” would immediately shift forward 12 months. However, analysts and data aggregators may not change the definitions of “FY0” and “FY1” for a few days or weeks. In case of doubt, it may be worth looking at individual estimates in order to double-check that the correct set of numbers is being used.
Sum-of-the-Parts Analysis
When attempting to use relative valuation methods on firms with multiple lines of business, the analyst may not be able to identify any company that is directly similar on all dimensions. In such instances, relative valuation methods can be extended to encompass “sum-of-the-parts” analysis, which considers each part of a business separately and attempts to value them individually by reference to companies that are mainly or solely in one particular line of business (see Hooke, 2010, Chapter 18).
Relative valuation analysis based on sum-of-the-parts approaches will involve the same challenges as were described above— identifying a suitable universe of companies engaged in each particular industry, collecting and collating the necessary data, and then using the results to gauge what might be a “fair value” for each of the individual lines of business. But in addition to these considerations, there is an additional difficulty, which is specific to sum-of-the-parts analysis. This problem is whether to apply a conglomerate discount, and if so, how much.
Much financial theory assumes that all else equal, investors are likely to prefer to invest in companies that are engaged in a single line of business, rather than to invest in conglomerates that have operations across multiple industries. Investing in a conglomerate effectively means being exposed to all of that conglomerate’s operations, and the overall mix of industry exposures might not mimic the portfolio that the investor would have chosen if it were possible instead to put money into individual companies.
A possible counterargument might be that a conglomerate with strong and decisive central control may achieve synergies with regard to revenues, costs, or taxation that would not be available to individual free-standing firms dealing at arms’ length with one another. A skeptical investor might wonder, on the other hand, about whether the potential positive impact of such synergies may be partly or wholly undermined by the negative impacts of centralized decision making, transfer pricing, and regulatory or reputational risk.
For these reasons, an analyst might consider that it is reasonable to apply a discount to the overall value that emerges from the “sum of the parts.” Some practitioners favor a discount of somewhere between 5% and 15%, for the reasons given above. Academic research on spinoffs has suggested that the combined value of the surviving entity and the spun-off firm tends to rise by an average of around 6%, though with a wide range of variation (see Burch and Nanda, 2003). (Some analysts have suggested that in some particular contexts, for instance in markets where competent managers are very scarce, then investors should be willing to pay a premium for being able to invest in a conglomerate that is fortunate enough to have such executives. However, this appears not to be a mainstream view.)
Relative Valuation versus DCF: A Comparison
Relative valuation methods can generally be implemented fairly fast, and the underlying information necessary to calculate can also be updated quickly. Even with the various complexities discussed above, an experienced analyst can usually create a relative valuation table within an hour or two. And the calculated valuation multiples can adjust as market conditions and relative prices change. In both respects, relative valuation methods have an advantage over DCF models, which may require hours or days of work to build or update, and which require the analyst to provide multiple judgment-based inputs about unknowable future events. Moreover, as noted by Baker and Ruback (1999), if a DCF model is extended to encompass multiple possible scenarios, it may end up generating a range of “fair value” prices that is too wide to provide much insight into whether the potential investment is attractive at its current valuation.
Relative valuation methods focus on how much a company is worth to a minority shareholder, in other words an investor who will have limited or zero ability to influence the company’s management or its strategy. Such an approach is suitable for investors who intend to purchase only a small percentage of the company’s shares and to hold those shares until the valuation multiple moves from being “cheap” to being “in line” or “expensive” compared with the peer group. As noted above, relative valuation methods make no attempt to determine what is the “correct” price for a company’s shares, but instead focus on trying to determine whether a company looks attractive or unattractive by comparison with other firms that appear to be approximately similar in terms of size, geography, industry, and other parameters.
DCF methods attempt to determine how much a company is worth in terms of “fair value” over a long time horizon. DCF methods can readily incorporate a range of assumptions about decisions in the near future or the distant future, and therefore can provide a range of different scenarios. For this reason, most academics and practitioners consider that DCF methods are likely to produce greater insight than relative valuation methods into the various forces that may affect the fair value for a business. More specifically, DCF methods can be more applicable to situations where an investor will seek to influence a company’s future direction—perhaps as an activist investor pushing management in new directions, or possibly as a bidder for a controlling stake in the firm. In such situations, relative valuation analysis is unlikely to provide much insight because the investor will actually be seeking to affect the company’s valuation multiples directly, by affecting the value of the denominator.
Nevertheless, even where an analyst favors the use of DCF approaches, we consider that relative valuation methods can still be valuable as a “sanity check” on the output from a DCF-based valuation. An analyst can take the expected valuation from the DCF model and compare it with the projected values for net income, shareholders’ equity, operating cash flow, and similar metrics. These ratios drawn from the DCF modeling process can then be compared with the multiples for a universe of similar firms. If the multiples generated by the analyst’s DCF model are approximately comparable with the multiples that can be derived for similar companies that are already being publicly traded, then the analyst may conclude that the DCF model’s assumptions appear to be reasonable. However, if the multiples from the analyst’s model appear to diverge considerably from the available information concerning valuation multiples for apparently similar firms, then it may be a good idea to reexamine the model, rechecking whether the underlying assumptions are truly justifiable.
Relative valuation methods can also be useful in another way when constructing DCF models. Most DCF models include a “terminal value,” which represents the expected future value of the business, discounted back to the present, from all periods subsequent to the ones for which the analyst has developed explicit estimates. One way to calculate this terminal value is in terms of a perpetual growth rate, but the choice of a particular growth rate can be difficult to justify on the basis of the firm’s current characteristics. An alternative approach is to take current valuation multiples for similar firms and use those values as multiples for terminal value (see Damodaran, 2006, Chapter 4, pp. 143–144).
KEY POINTS
- Relative valuation methods tend to receive less attention from academics than DCF approaches, but such methods are widely used by practitioners. If relative valuation approaches suggest that a company is cheap on some metrics but expensive on others, this may indicate that the market views that company as being an outlier for some reason, and an analyst will probably want to investigate further.
- Choosing an appropriate group of comparable companies is perhaps the most challenging aspect of relative valuation analysis. Where possible, an analyst should seek to identify six to 12 companies that are similar in terms of size, geography, and industry. If this is not possible, then an analyst should feel free to relax one or more of these parameters in order to obtain a usable universe.
- Determining an appropriate set of valuation multiples is also important. Calculating a single set of multiples is likely to provide fewer insights than using several different metrics that span multiple time periods. It is conventional to use consensus estimates of future financial and operating performance, as these presumably represent the market’s collective opinion of each firm’s prospects.
- Most relative valuation analysis is performed using standard multiples such as price/earnings or firm value/sales. Under some conditions, using industry-specific multiples can be valuable, though there may be fewer consensus estimates for such data, and there may also be less intuition about what is the “fair” price for such ratios.
- Relative valuation methods are particularly useful for investors who aim to take minority stakes in individual companies when they are “cheap” relative to their peers, and then sell those stakes when the companies become “expensive.” Such methods are likely to be less directly useful for investors who will seek to influence a company’s management, or who aim to take a controlling stake in a company. For such investors, DCF methods are likely to be more applicable.
NOTES
1. By contrast, in an example of how to assess a small wine producer, the proposed universe of comparables consisted of 15 “beverage firms,” including both small and large caps, and covering specialists in beer, wine, and soft drink production. Arguably, some of these are unlikely to be very similar to the proposed target of analysis. See Chapter 7 in Damodaran (2006, pp. 249–252).
2. For further examples using real firms and actual figures, see Damodaran (2006, Chapters 7 and 8) or Hooke (2010, Chapter 15).
3. The material discussed here does not necessarily represent the opinions, methods, or views of Delaware Investments.
REFERENCES
Baker, M., and Ruback, R. (1999). Estimating Industry Multiples. Cambridge, MA: Harvard Business School.
Bhojraj, S., Lee, C. M. C., and Oler, D. K. (2003). What’s my line? A comparison of industry classification schemes for capital market research. Journal of Accounting Research 41, 5: 745–774.
Bhushan, R. (1989). Firm characteristics and analyst following. Journal of Accounting and Economics 11, 2–3: 255–274.
Block, S. B. (1999). A study of financial analysts: Practice and theory. Financial Analysts Journal 55, 4: 86–95.
Burch, T. R., and Nanda, V. (2003). Divisional diversity and the conglomerate discount: Evidence from spinoffs. Journal of Financial Economics 70, 1: 69–98.
Chan, L. K. C., Lakonishok, J., and Swaminathan, B. (2007). Industry classifications and return comovement. Financial Analysts Journal 63, 6: 56–70.
Copeland, T., Koller, T., and Murrin, J. (2000). Valuation: Measuring and Managing the Value of Companies, 3rd edition. New York: Wiley.
Damodaran, A. (2006). Damodaran on Valuation: Security Analysis for Investment and Corporate Finance, 2nd edition. New York: Wiley.
Hooke, J. (2010). Security Analysis on Wall Street: A Comprehensive Guide to Today’s Valuation Methods, 2nd edition. New York: Wiley.
Liu, J., Nissim, D., and Thomas, J. (2002). Equity valuation using multiples. Journal of Accounting Research 40, 1: 135–172.
Liu, J., Nissim, D., and Thomas, J. (2007). Is cash flow king in valuations? Financial Analysts Journal 63, 2: 56–68.
Phylaktis, K., and Xia, L. (2006). The changing roles of industry and country effects in the global equity markets. European Journal of Finance 12, 8: 627–648.
Valentine, J. J. (2011). Best Practices for Equity Research Analysts. New York: McGraw-Hill.