Faculty of Economics and Business
Bachelor Thesis
Value and Growth investing: a case for Enterprise Value
Multiple
2014 July
Vytenis Susinskas
Student Number: 6157084
Study Program: BSc in Economics and Business
Specialization: Economics
Page 2 of 26
Table of Contents
1. Introduction
• 1.1. Research question 2. Literature review
• 2.1. Fundamental valuation metrics • 2.1.1. Enterprise value multiple • 2.2. Existing research
• 2.3. Relevance of the study 3. Design of the empirical study
• 3.1. Data
• 3.2. Methodology
• 3.3. First part of the analysis • 3.4. Second part of the analysis 4. Results
▪ 4.1. Results first part ▪ 4.2. Results second part 5. Summary and Conclusion
6. Discussion
Bibliography Appendix
A.1. Abbreviations
A.1.1. COMPUSTAT abbreviations A.2. Tables and Charts
Page 4 of 26
1. Introduction
1.1. Research question
Is the Enterprise multiple (EM) superior to Book-to-market (B/M) for value stock selection?
The stock market is a widely researched topic. The academic literature is full of articles on stock market efficiency, anomalies, investment styles, investor overreaction, other. The cornerstone of modern financial economics is efficient market hypothesis (EMH), the paradigm was formalized by (Fama 1970). Three forms of market efficiency were distinguished, the weak form of market
efficiency postulated that in a competitive market it is not possible to receive excess return basing investment decisions on information obtained from prior information, such as security prices or investment returns. The semi-strong form states that an excess return is not attainable basing a decision not only on the past information but also on all publicly available information, such as balance sheets of the companies, income statements, dividend changes, stock splits and other. Third, is the strong form that asserts that not only all historical, public, but privileged, inside information is fully reflected in current equity market prices. There is a perfect revelation of all private
information in market prices (Malkiel 1992). Usually one of the biggest challenges to the EMH is various anomalies, a pattern in the security price that is widely known, consistent and thus should be easily exploitable for economic value. Focusing on those anomalies that are relevant for research into market efficiency first is the "size effect," the apparent excess return that results from choosing companies with smaller market capitalization, while controlling for additional risk (Banz 1981). Secondly, there is the relation between price/earnings ratios and expected returns (Basu 1977). In Basu (1977) it was shown that during the time from April 1957 till March 1971, the low P/E portfolios earned on average a higher risk-adjusted and absolute rate of return than the stocks with high P/E. In a similar fashion, Chan, Hamao, and Lakonishok (1991) found that a significant relationship between fundamental variables and expected returns in the Japanese market exist, with the performance of the book-to-market equity ratio being especially noteworthy. Whether or not one can reap a sustainable excess return from these anomalies is a question unlikely to be settled in the exclusivity of academic setting, as paper profit calculations is done easier, than real world
simulation that would incorporate the full features of the process such as transaction costs, liquidity, "Black Swan" events, and others.[1] Another important factor to be mentioned is luck, in fact, it can play an important part, regular patterns in historical data can be discovered even if no recurrence exist, purely due to luck, although the possibility of so is tiny, it rises with the number of tries (Brown 1992). While debating the validity of EMH is beyond the scope of this paper, it is relevant to set a clear picture of the current state of the research and be aware of potential limitations of the research.
[1] “Black Swan” – according to Taleb (2010), is an event with three attributes. Firstly, it is a true outlier, “outside the realm of regular expectations” (Taleb, 2010: p. XXII prologue). Secondly, its impact is very significant. Thirdly, after the fact it is easily rationalized, as if it could have been predicted.
Page 5 of 26
Although, according to the Efficient Market hypothesis, security prices fully reflect available information in a fast and unbiased manner and thus provide true estimates of corresponding values, the research into possibilities of excess returns continues.
Drawing on the before mentioned research, there exists a value and growth investing distinction. Value stocks are those that have high ratios of book-to-market equity (B/M) (Fama 1992), price to earnings (P/E) (Basu 1977), or cash flow to price (C/P) (Lakonishok et al. 1994) and investing strategies that exclusively targets stocks with the lowest M/B, P/E and P/C ratios as value investing strategies, while the opposites being growth or glamour. The growing of academic interest in value and glamour investment strategies can be tracked back to Fama and French (1992) and Lakonishok, Shleifer, and Vishny (1994). The first study diminished the explanatory power of the Capital Asset Pricing Model (CAPM), subsequently academics moved to study the book-to-market equity ratio and the size of the company as the leading explanatory factors for the cross section of mean equities returns.[2] The work built on earlier studies of market inefficiencies such as (Basu 1977). Value investing is different from other possible market inefficiencies. Chan and Lakonishok (2004) review the state of research in the topic at the time and conclude that some of the market inefficiencies may lack any logical basis, such as day-of-the-week patterns.[3] Thus it is a legitimate reason to believe that such patterns are only statistical errors. However the value premium is related to investor behavior. Such as getting excessively excited about the new technologies and inflating the price of growth stocks. These behavioral characteristics will probably exist in the future, meaning that value investing will remain a profitable long-term investing strategy if value stocks are carefully picked. This research facilitates the careful selection of the best value stocks and so its conclusions will probably be relevant for decades to come.
Based on the assembled evidence from studies on the book-to-market equity ratio, P/E ratios and related anomalies, the accepted opinion is that value investment strategies, on average, tend to outperform growth investment strategies, (see Basu 1977, Capaul 1993, Fama and French 1992 and Lakonishok et al. 1994, whose work showed that for U.S. stocks there exists a strong value
premium in average returns). Capaul (1993) and Fama and French (1998) built on the before mentioned work to extend the value outperformance to international stocks. The previously
mentioned academic work has had a big impact on professional money management industry. Value and growth are two widely recognized distinctive investing styles adopted by professional money managers.
However, there has been a limited amount of research into different fundamental variables that would lead to forming value portfolios, which would outperform glamour portfolios. Most of the research was based on P/E (Basu 1977) and B/M (P/BV) (Fama and French 1970, 1992, 1998).
[2] Capital Asset Pricing Model (CAPM) – states that the excess return of a security is linearly related to the excess return on the efficient market portfolio, and the relation is measured by a factor beta. The factor measures the level of the systematic risk of a security. The model is expressed as follows: E( ) = + [E( (Damodaran, 2012). for abbreviations see appendix A.1. The CAPM was independently defined by Sharpe (1964), Lintner (1965), Treynor (1965) and subsequently revised and extended by Mossin (1966), Fama (1968), and Long (1974). The model is related to the topic discussed, however the approach used in this paper does not utilize CAPM as Basu (1977) or Fama and French (1992), that is why the CAPM is relegated to a footnote.
[3] Day-of-the-week – refers to the anomaly that daily stock returns on Mondays are lower and higher on Fridays (French 1980)
Page 6 of 26
Although, providing an exhaustive list of variables is beyond the scope of this paper, there is still something to add to the current state of research.
The paper is structured as follows; firstly the background is discussed in the section 1, secondly the relevant literature is reviewed, and results are discussed in section 2. Thirdly, the data for this study is described, along with the methods used to examine the data, and relevant testing is done in section 3. The methods include sorting the stocks into ten portfolios according each of the
fundamental metric, measuring subsequent returns, testing the predictive power of the metrics in a multiple regression analysis. Furthermore, section 4 will describe the results of testing and analysis. Lastly, in section 5 limitations of the paper will be discussed as well as suggestions for further research presented. The paper will conclude with a summary.
2. Literature review
This section will provide background information on fundamental equity valuation approach including “bottom up” and “top down.” Next, the existing research and findings will be presented. Furthermore, the relevance of the study will be discussed.
2.1. Fundamental valuation metrics
Fundamental analysis of equities is a distinct investment approach that utilizes various fundamental information about a company, to reach an investment decision. The principles of this analysis were first described in the book “Security Analysis” (Graham and Dodd 1934). There are two distinct ways to approach fundamental analysis: the "Top down" and the "Bottom up." The first approach is mainly concerned with general macroeconomic data, like the stage of the business cycle, interest rate and other macro data. The aim of this approach is, given the macro data, to identify an industry that would outperform other industries in this economic environment and to identify the stocks that would outperform in the particular industry.
The "Bottom up" approach of fundamental analysis is mainly geared towards estimating the fair value of a security and comparing it to the current market price. Stocks that appear undervalued would be a buy, independent of the current macroeconomic environment. Two of the most widely known proponents of this approach are Benjamin Graham and Warren Buffett.
The "Bottom up" approach has different methods for estimating the fair value of the stock. The main three are a) Asset based, comparing the book value to market value of a company's assets (Graham and Dodd 1934) b) Discounted cash flow valuation, estimating the cash flow the business generates for its stockholders and discounting it at a relevant rate (Williams 1938) c) Fundamental valuation or relative valuation models, calculating certain multiples stocks are trading at and comparing those to multiples of the other firms or over time (Damodaran 2012).
The last method for „Bottom up" approach is used in this paper, since this approach allows to directly distinguish between value and glamour stocks. The relative valuation has four principal methods:
Page 7 of 26
1. Relative asset multiple valuation: B/M (P/BV), others. 2. Relative revenue multiple valuation: P/S.
3. Relative earnings multiple valuation: P/E, PEG, others. 4. Relative cash flow multiple valuation: P/CF, P/EBITDA.
For this paper, a ratio from each of the methods listed above is chosen: P/BV, P/S, P/E, P/CFO (cash flow from operations). And an extra valuation multiple added EM (EV=enterprise
value=market value of equity + total debt + preferred stock value – cash and investments/EBITDA (Earnings before interest, taxes, depreciation and amortization)).[4]
2.1.1. Enterprise value multiple
Price-to-earnings P/E and Book-to-Market multiples are arguably the most used in academic setting (see Fama and French 1992, 1998), however it has serious flaws. Loughran (1997) reports that the empirical findings of before mentioned papers are driven by two features of the data: Exceptionally low returns on small growth/glamour stocks, and a seasonal January effect in the book-to-market data. “In the largest size quintile of all firms (accounting for 73% of the total market value of all publicly traded firms), book-to-market has no significant explanatory power on the cross-section of realized returns during the 1963–1995 period. Thus, book-to-market as such would have less importance to money managers than the literature would have led us to believe” (Loughran, 1997: p. 249). Furthermore, Enterprise value multiple (EM) utilizes the information about a stocks balance sheet, for example, debt, preferred stock, cash, this provides more information about a true
company costs than just the market value of equity. Additionally, EM provides a meaningful way to compare different companies with differing degrees of leverage (Damodaran, 2012). There are more reasons to prefer EM, the EBITDA, as Damodaran (2012) noted that differences in depreciation methods across companies will affect net income, but not the operating income before depreciation. To sum up, portfolios sorted by Enterprise value multiple should provide a more consistent return on average.
2.2. Existing research
Value stocks are those that have low ratios of market-to-book (M/B or P/BV), price to earnings (P/E), price to cash flow (P/CF), or other fundamental variables. Investing strategies that
exclusively targets stocks with the lowest P/BV, P/E and P/CF ratios are value investing strategies while the opposites are glamour. The growing of academic interest in value and glamour investment strategies can be tracked back to (Fama and French, 1992) and (Lakonishok et al., 1994). These value stocks, therefore, are priced at a discount to their fundamental multiples. The underpricing could due to higher risk intrinsic to these stocks, or their relative outperformance could be a result of data snooping, or could be easily explained by January effect and controlling for micro market value of equity securities, as discussed in section 2.1.1.
Page 8 of 26
One of the earliest prominent studies that has been done on value and glamour stock characteristics include the research of Basu (1977), and Banz (1981). In the first paper, a data sample of 753 firms was collected of firms that traded on the NYSE between September 1956 and August 1971. Stocks were ranked according to their price to earnings ratio, and five portfolios were formed. The returns on those portfolios were tested using the CAPM. A significant negative relationship between P/E ratios and average returns for NYSE stocks were found. This relation could not be explained by the Capital asset pricing model. In the second paper the author finds that smaller firms have higher returns, on average, even when risk-adjusted, this negative relation between a stock return and the market value of equity was called Size effect. The study showed that the relation is evident in at least forty years of data and provides evidence that the CAPM is misspecified. Moreover, the size effect is not linear and is more evident in the smallest firms’ portfolio while being almost
undetectable between the average and large firm size portfolios.
Chan et al. (1991) explores the relationship between securities return and fundamental variables in Japan. The data used is comprehensive, high quality, spanning from 1971 to 1988, containing the first and second sections of the Tokyo Stock Exchange. The set of fundamental variables used includes earnings yield, market value of equity, cash flow yield and book value of equity ratio. Their study finds a significant relationship between the fundamental variables and equities return in Japan during the period studied, with the book-to-market ratio being the most statistically
significant of all the variables studied. Secondly, the cash flow yield has a positive and generally statistically significant impact on the return. Furthermore, the existence of the size effect is confirmed, however the results are sensitive to the model used. The results of the earnings yield study were the most obscure. If the yield would be used in isolation or with size variable only, it shows a positive and significant relationship. If however, the book value of equity ratio is added to the model, the earnings yield becomes statistically insignificant.
Capaul et al. (1993) extends the study of the value premium to France, Germany, U.K., Switzerland, Japan and U.S. They document that during the period from 1981 – 1992, stocks with low price-to-book ratios provide a risk-adjusted superior return to stocks with high price-to-price-to-book ratios. In a one of the most authoritative studies done of the U.S. stock markets Fama and French (1992) studying the 1963 – 1990 time period finds that firms‘ market value of equity and the price-to book ratio capture all the cross-sectional variation in average equity returns associated with market beta, size, price-to-book ratios and price-to-earnings ratios
Lakonishok et al. (1994) studied the stocks on New York Stock Exchange (NYSE) and American Stock Exchange (AMEX) during the period of April 1968 to April 1990 and established three propositions. Firstly, that value stocks do outperform glamour stocks given the universe of the stocks studied. Secondly, this outperformance may be because market participants overestimated consistently future growth rates of various fundamental metrics of the glamour stocks. Thirdly, using simple approaches to risk, value strategies appear to not contain any more significant risk than glamour stocks.
In a newer study, Fama and French (2012) examine four regions (North America, Europe, Japan, and Asia Pacific) during the 1989 – 2011 period. They find value premium in average security
Page 9 of 26
returns across all of the regions as well as higher returns from smaller value stocks in all of the regions except Japan.
While all of this research universally documents higher average returns for value stocks, the reason for these returns is ambiguous. It could be explained by the higher intrinsic risk of these value stocks, or it could be a case of data-snooping, or influenced by the behavioral biases of the market participants (Lakonishok et al., 1994).
2.3. Relevance of the study
Following the review of the existing research, one can observe that the majority of the research has examined the relationship between investment returns and fundamental valuation metrics such as price-to-earnings, price-to-sales and book-to-market value. Usually, only one fundamental metric and its impact on the subsequent return was examined. Only occasionally, the fundamental metric was used together with a market capitalization, otherwise known as the size effect, jointly to examine the effect on investment return.
What has not been done is a direct comparison of these fundamental metrics, P/E. P/S, P/CF, B/M, EM and their relation to investment returns as well as investing approach utilizing a combination of these metrics. In this research, stocks are sorted by each of the fundamental metrics into ten
portfolios, giving a total of 50 distinct portfolios. For convenience purposes these metrics are expressed as follows, P/E, P/S, P/CFO, P/BV (price to book-value) and EV/EBITDA being the enterprise multiple. This would allow two things: firstly portfolio Nr. 1, being the portfolio with the 10% of the lowest values of the metric would be value stocks and portfolio Nr. 10 would mean glamour stocks, secondly this would allow a direct comparison between the metrics. Portfolios will be rebalanced annually.
This research will hopefully contribute to the existing literature by providing some perspective and comparison of the fundamental metrics, which could help in the future while building strategy that delivers the coveted consistent long-term excess returns.
3. Design of the empirical study
This section will describe the data used for the estimations, along with the methodology and the description of the carried out tests.
3.1. Data
The sample selection roughly follows (Fama, French 1992) and includes all firms from New York Stock Exchange (NYSE), NASDAQ, and American Stock Exchange (AMEX) with available stock price from the Center for Research in Security Prices (CRSP) and accounting information from the
Page 10 of 26
merged CRSP/COMPUSTAT database. The data are accessed through Wharton Research Data Services (WRDS). Fama and French exclude financial firms from their sample because the high leverage of those firms does not mean the same for financial. Since in this research the role of leverage for a subsequent return is not analyzed, financial firms are not excluded from the sample, this would also provide a more realistic investable universe. To avoid a back-filling bias, a stock are required to have at least two years of accounting data to enter the sample
Accounting information is measured from December 1963 till December 2012. The start date is chosen, because data for pre-1962 have a serious selection bias and is tilted towards big historically successful firms (Fama, French 1992).
Utilizing conservative approach of Fama and French, accounting data for fiscal year-ends in calendar year t - 1 is matched with the returns for June of calendar year t to returns of June t + 1. The minimum 6-month gap was chosen, because although firms are required to file their accounting reports (10-K) with the Securities and Exchange Commission within 90 days, on average 19.8% do not comply (Fama and French 1992).
There are 186,698 firm year observations that were retrieved from CRSP/COMPUSTAT database. Additional data screens were done: 20,699 firm year observations were removed due to missing or negative BV (book value of equity); 937 firm year observations were removed due to missing market value of equity; 17,670 firm year observations were removed due to missing or 0 stock price at either June t or June t+1; 17,974 firm year observations were removed due to missing or negative income or P/E; 20,256 firm year observations were removed due to missing or negative EBITDA. This was done for consistency reasons since papers like (Fama and French 1992) remove firm year observations with negative BV (book value of equity). Negative EBITDA firms would be
mathematically categorized as low EM firms, yet one could easily argue the opposite. Additionally, 6085 firm year observations with market value of equity of less than ten million dollars were removed. Although, this is quite arbitrary screener, stocks with lower capitalization are very thinly traded and bid – ask spread would be too high for real world trading. However, there is no research to be found at what market value of equity a stock would be liquid enough for practical trading. Finally, firm year observations are winsorized at 1% and 99% to restrict the impact of outliers, leaving the sample with 95,919 firm year observations. Table I reports the equally weighted statistics for the 95,919 firm year observations over the 1963-2012 period.
3.2. Methodology
The first part of the analysis will be a comparison of the returns during the 1964 - 2014 year period between value and glamour stocks.[5] Value strategies are sorted according to price-to-earnings ratio, price-to-sales, price-to-book value, price-to-cash flow, and the enterprise value multiple.
[5] The discrepancy between the periods is because the securities are sorted into deciles according to accounting variables in December year t – 1 (2012), utilizing conservative approach, one assumes that these values become available to the public only in June year t (2013), so this is the starting stock price. The end stock price then is recoded in June year t + 1 (2014).
Page 11 of 26
Second part uses the multiple regression analysis to establish the predictive power and significance of these multiples in explaining the corresponding equity returns.
3.3. First part of the analysis
Similarly to (Fama and French, 1992) and (Lakonishok et al., 1994), ten portfolios are formed based on the deciles of the five fundamental valuation metrics, leading to a total of fifty portfolios. These metrics are the price to earnings ratio (P/E), price-to-sales ratio (P/S), price-to-cash flow (P/CF), price-book-value (P/BV), and the enterprise value multiple (EM). The first decile would contain the securities with the lowest metrics, classifying them as value stocks, the tenth decile would contain the stocks with the highest ratios classifying them as glamour stocks. Portfolios are composed of equally weighted securities. Portfolio formation begins in 1963 and ends in 2012. Utilizing the conservative 6-month gap to allow the fundamental variables to be reported as explained in section 3.1., this means that the portfolios are sorted to deciles on year t – 1 at the end of the year values, starting equity price is recorded on year t June 30, and end equity prices are recorded on year t + 1 June 30 allowing for a full year. As explained in (Fama and French, 1992) the use of December accounting values and market values is objectionable due to a different fiscal years of the firms in question, however the test mixes firms with different fiscal year ends and usually test using a smaller sample with only firms with December fiscal year ends yields the same result (Fama and French, 1992).
The annual returns of the portfolios are computed for the 1963 – 2014 period. Following convention the yearly return is computed as follows:
= ( - + ) / (1)
where is the return of the stock for the company during the period from t June till t + 1 June, is the price of the company in t + 1 June, is the price of the company in t June, and being the dividends company paid during the period of June t to June t + 1. As stocks would be equally weighted the total return would be calculated using standard academic notation as follows:
∑ (2)
Comparing the differences in the returns of the portfolios labeled as value and glamour, one can infer which strategy would have yielded superior results in the sample studied.
3.4. Second part of the analysis
In this part, the predictive power of the fundamental metrics will be tested using multivariate OLS regression with the security returns being the dependent variable. The independent variables will be price to earnings ratio (P/E), price-to-sales ratio (P/S), price-to-cash flow (P/CF), Price-book-value (P/BV), Enterprise value multiple (EM) and the natural logarithm of the market value of equity (size). These are the same variables as used in the first part of the analysis with an extra variable
Page 12 of 26
size which would help to control the effect of the stock market value of equity on the return (Chan et al., 1991). The regression model would be:
size + (3)[6]
To check for multicollinearity White’s test for multicollinearity will be performed. To avoid biased test statistics the OLS regression will be carried out using robust standard errors. This part of the analysis will be carried out using statistical package STATA 13 MP.
4. Results
In this section, the results of the analysis described above will be introduced and examined.
4.1. Results first part
As describe in section 3.3. fifty portfolios were formed according to the deciles of the price to earnings ratio (P/E), price-to-sales ratio (P/S), price-to-cash flow (P/CF), price-book-value (P/BV), and the enterprise value multiple (EM). Value strategies are represented by the first decile portfolios and glamour by the tenth. The results of the analysis are presented in the Appendix A.2. Table III. As could be seen value portfolios outperformed glamour portfolios during the period studied no matter which metric was used for securities selection. Table III a) presents the difference between the average returns of portfolios formed from stocks belonging to decile 1 and decile 10. As we can see the value stocks selected according to P/CFO outperformed the glamour stocks by 10,857 % on average during the period studied, giving it the highest difference. However, the sample price-to-cash flow ratio is not as complete as for the other ratios with only 58,395 firm year observations, the ratio with the highest value premium with a complete sample is price-to-book value earning an average premium of 10,5839 %. The lowest difference between the value and glamour portfolios was while selecting the stocks according to EV/EBITDA or enterprise value multiple 6,2819 %. Graphical representation for the results can be seen in the Appendix A.2. Charts I to V. The results of this part of the analysis confirm the existence of the value premium and shows that the P/BV and P/CFO are the indicators with the highest value premium during the period studied.
4.2. Results second part
The second part is concerned with the predictive power of fundamental metrics. Firstly, the White test for heteroskedasticity is done, and according to the statistic one can reject the null hypothesis and declare that the standard errors are indeed heteroscedastic. The results can be seen in the Appendix Table IV a). To remedy this an OLS regression with Robust Standard Errors is carried
Page 13 of 26
out. As discussed in section 3.4., the dependent variable is the average return of a security during the 1963-2014 period of study. The independent variables will be price to earnings ratio (P/E), price-to-sales ratio (P/S), price-to-cash flow (P/CF), Price-book-value (P/BV), Enterprise value multiple (EM). To control for the market value of equity, which was found significant as discussed in section 2.2., see (Chan et al. 1991), natural logarithm of the market capitalization is added. The results of the regression are presented in Table IV b) for complete six variables but only for a limited sample of observations due to P/CFO having some missing values and Table IV c) for a complete sample of observations but without the price-to-cash flow multiple. The first regression shows that P/E, EV/EBITDA, and P/CFO are all insignificant in predicting the average return, with P/S, P/BV and the size factor being significant metrics. One can notice as well that P/E ratio show a positive coefficient meaning that securities with higher P/E should earn a higher return, which is at odds with the literature reviewed (Basu, 1977). The observation could be explained by the fact that the outperformance of this metric was one of the first discovered, and it has been fully exploited. The second regression confirms the insignificance of the price-to-earnings ratio, the other ratios namely, price-to-sales, price-to-book, enterprise value, and size being significant even at 1% significance level. With the P/BV having the most explanatory power on the corresponding return. Furthermore, a correlation analysis is carried out, and results are represented in Table II, which shows the highest correlation coefficients of EV/EBITDA, although they do not indicate perfect correlation, they could mean there is a problem with the way the metrics are structured or selected for this analysis.
To summarize the results of this section, the existence of a value premium is confirmed in this sample, with price-to-operating cash flow multiple yielding the highest difference in return, although a more exhaustive dataset is needed to study the variable more thoroughly.[7] Not far behind is the price-to-book value ratio, still evident even after 22 years (Fama and French 1992). The EV/EBITDA ratio, which this research was set to prove as a superior metric to P/BV, under delivers, which could be attributed to sample selection error, as financial firms were not removed from the sample, as high debt and cash held of such firms can obscure the metric.
5. Summary and Conclusion
This paper has studied the value versus glamour stock performance during the 1963 – 2014 year period. The securities included all firms from New York Stock Exchange (NYSE), NASDAQ, and American Stock Exchange (AMEX) with available stock price from the Center for Research in Security Prices (CRSP) and accounting information from the merged CRSP/COMPUSTAT database. Additionally, it has conducted a multiple regression analysis to estimate the predictive power of these metrics on the average stock return for the period.
[7] The accounting dataset used for this study (retrieved from CRSP/COMPUSTAT) contained only 58,395 firm year observations of price‐to‐operating cash flow multiple (P/CFO). As for the other multiples the database contained 95,919 firm year observations.
Page 14 of 26
Stocks were sorted according to their fundamental valuation multiples: price to earnings ratio (P/E), price-to-sales ratio (P/S), price-to-cash flow (P/CF), Price-book-value (P/BV), and Enterprise value multiple (EM). With value stocks being the ones with the smallest ratios and glamour with the highest. Past research has shown a significant negative relationship between P/E ratios and average return (Basu, 1977), a significant relationship between the fundamental variables and equity return in Japan, with the book-to-market ratio being the most statistically significant of all the variables (Chan et al. 1991), as well as firms market value of equity and the price-to book ratio capture all the cross-sectional variation in average equity returns (Fama and French, 1992).
In the first part of the analysis, value portfolios outperformed glamour portfolios during the period studied no matter which metric was used for security selection, and that the P/BV and P/CFO are the indicators with the highest value premium during the period studied.
The second part of analysis utilized the multivariate regression to estimate the explanatory power of these fundamental indicators. Two different regressions were done. The first regression showed that P/E, EV/EBITDA, and P/CFO are all insignificant in predicting the average return, with P/S, P/BV and the size factor being significant metrics, and the P/E ratio showing a positive coefficient meaning that securities with higher P/E should earn higher return, which is at odds with the literature reviewed (Basu, 1977). The second regression confirmed the insignificance of the price-to-earnings ratio, the other ratios namely, price-to-sales, price-to-book, enterprise value, and size being significant even at a 1% significance level. With the P/BV having the most explanatory power on the corresponding return
To conclude, within this sample of firms from the U.S. and given the time period of 1963 – 2014, value strategy based on price-to-cash flow and price-to-book ratios would have earned a
significantly higher average return than glamour strategy.
6. Discussion
The present analysis studied the period from 1963 till 2014 returns, this may not be a realistic holding period for the average investor. Further research could dissect the study period into shorter time-frames.
Furthermore, as explained in section 4.2. in a further research all financials firms should be
excluded from the sample as high debt and cash held of such firms can introduce significant outliers into the enterprise value ratios. Additionally, arbitrary data screener was used as explained in section 3.2. of removing the stocks with market value of equity of less than ten million. A future research could focus on establishing at what market value of equity a stock would be liquid enough for practical trading purposes.
Additionally, for further research the geography of the sample should be expanded as is common practice in the most current research into the topic (Fama and French 2012).
Page 15 of 26
For the modeling part, in this research the firm year observations with negative P/E ratios were removed, however in Lakonishok et al. (1994), the ratios were successfully incorporated into the model using dummy variables. Finally, a different regression model could be used, in example the Seemingly unrelated regressions (SUR), utilized by (Chan et al., 1991).
Page 16 of 26
Bibliography
Banz, R., 1981, "The relationship between return and market value of common stock," Journal of Financial Economics 9, 3–18.
Basu, S., 1977, “Investment performance of common stocks in relation to their price-earnings ratios. A test of market efficiency," Journal of Finance, vol. 32, no. 3 (June), pp. 663-682. Brown, S. Goetzmann, W. Ibbotson, R. and Ross, S., 1992, "Survivorship bias in performance studies," Review of Financial Studies 5, 553–80.
Chan, Louis K. C., Yasushi Hamao, and Josef Lakonishok. 1991. "Fundamentals and Stock Returns in Japan,” Journal of Finance, vol.46, no.5 (Dec), pp. 1739-64.
Chan, Louis K. C., and Josef Lakonishok. 2004. “Value and Growth Investing: Review and Update,” Financial Analysts Journal, Vol. 60, No. 1 (Jan. – Feb.), pp. 71-86.
Capaul, C., Rowley I., William F. Sharpe, 1993, "International Value and Growth Stock Returns," Financial Analysts Journal, Vol. 49, No. 1 (Jan. - Feb.), pp. 27-36.
Damodaran, A. 2012, “Investment Valuation: Tools and Techniques for Determining the Value of Any Asset," 3ed, Wiley finance series.
Damodaran, A. 2012, “Value Investing: Investing for Grown Ups?” Available at SSRN:
http://ssrn.com/abstract=2042657
Fama, Eugene F., 1968, “Risk, Return and Equilibrium: Some Clarifying Comments,” Journal of Finance, vol. 23, no. 1, (Mar.), pp. 29-40.
Fama, Eugene F., 1970, "Efficient Capital Markets: A Review of Theory and Empirical Work," Journal of Finance (May), pp. 383-417.
Fama, Eugene F. and Kenneth R. French, 1992, “The Cross-Section of Expected Stock Returns.” Journal of Finance, vol. 47, no. 2 (June), pp. 427-465.
Fama, Eugene F. and Kenneth R. French, 1998, “Value versus Growth: The International Evidence.” Journal of Finance, Vol. 53, no. 6 (Dec.), pp. 1975-1999.
Fama, Eugene F. and Kenneth R. French, 2012, “Size, value, and momentum in international stock returns.” Journal of Financial Economics 105, 457-472.
French, Kenneth R., 1980, “Stock returns and the weekend effect,” Journal of Financial Economics, Vol. 8, pp.55-70.
Graham, B. and David L. Dodd, 1934, "Security analysis," first edition, Mc-Graw Hill, New York. Grossman, Sanford J., Stiglitz, Joseph E., 1980, “On the Impossibility of Informationally Efficient Markets," The American Economic Review, Vol. 70, No. 3 (Jun.), pp. 393-408.
Page 17 of 26
Lakonishok, J., Shleifer A., Vishny, Robert W., 1994, "Contrarian Investment, Extrapolation, and, Effect?" Journal of Financial and Quantitative Analysis, 32, pp 249-268.
Lintner, J., 1965, “The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets,” The Review of Economics and Statistics, Vol. 47, no. 1, (Feb.), pp. 13-37.
Long Jr., John B., 1974, “Stock prices, inflation, and the term structure of interest rates,” Journal of Financial Economics, vol. 1, no. 2, (July), pp. 131-170.
Loughran, T., 1997, “Book-to-Market across Firm Size, Exchange, and Seasonality: Is There an Effect?” Journal of Financial and Quantitative Analysis, Vol. 32, (Sep), pp. 249-268.
Malkiel, B., 1992, Efficient Market Hypothesis, in Newman, P., M. Milgate, and J. Eatwell (eds.), New Palgrave Dictionary of Money and Finance, London: Macmillan.
Mossin, J., 1966, “Equilibrium in a Capital Asset Market,” Econometrica, Vol. 34, no. 4 (Oct.), pp. 768-783.
Sharpe, William F., 1964, "Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk," Journal of Finance, Vol. 19, no. 3 (Sep.), pp. 425-442.
Taleb, Nassim N., 2010, "The Black Swan: the impact of the highly improbable," Random House Trade Paperback Edition.
Treynor, J.L., (1965), "How to Rate Management of Investment Funds”, Harvard
Business Review, Vol. 43, No. 1, pp. 63-75.
Page 18 of 26
Appendix
A.1. Abbreviations
P/E
PEG
Stock price / Earnings per share or market value of equity / Equity income – dividends on preferred stock + deferred taxes
P/E/expected rate of earnings growth
P/S Stock price / Sales per share
B/M P/BV
Book value of equity / Market value of equity Inversed
P/CF P/CFO
Market value of equity / Cash flow from operations
EM EV=enterprise value=market value of equity + total debt + preferred stock value – cash and investments/EBITDA=Earnings before depreciation, interest, taxes and amortization
CAPM E(R expected return on security i; R = risk-free rate; E(R = expected return on market portfolio; β beta of security i. (Damodaran, 2012)
A.1.1. COMPUSTAT abbreviations Market value
of equity
MKVALT
PRCC_F (price fiscal year end) *CSHO (comon shares outstanding)
EM EM = EV/EBITDA
EV = MKVALT + DLC+DLTT + PSTKRV – CHE DLC+DLTT (total year‘s debt)
PSTKRV (preferred stock value) CHE (cash adn short-term investments)
P/BV P/BV = MKVALT /(CEQ+TXDITC)
CEQ (book value of equity)
TXDITC (balance sheet deferred taxes and investment tax credit) P/S MKVALT/SALE
P/E P/E = MKVALT/(IB – DVP +TXDI)
IB(income before extraordinary items) DVP(dividends on preferred stock) TXDI (income statement deferred taxes.
EBITDA OIBDP (operating income before depreciation). Alternative: EBITDA = SALE – COGS – XSGA COGS (cost of goods sold)
XSGA (selling, general, and administrative expenses)
Page 19 of 26
A.2. Tables and Charts Table I - Descriptive statistics
Descriptive statistics of variables used in the analysis. RR2 – represent annual revenues in year t = 1965, ..., 2014. MV is the market value of equity at December t – 1. PS is MV/Sales ratio at the end of t – 1. PE is MV/Earings ratio at the end of t – 1. PCFO is the MV/Cash flow from operating activities ratio at the end of t – 1. PBV is MV/BV (book value of equity) at the end of t – 1. EVEBITDA is EM at the end of t – 1.
Page 20 of 26
Table II – Correlations
Correlations with the significance levels. . RR2 – represent annual revenues in year t = 1965, ..., 2014. MV is the market value of equity at December t – 1. PS is MV/Sales ratio at the end of t – 1. PE is MV/Earings ratio at the end of t – 1. PCFO is the MV/Cash flow from operating activities ratio at the end of t – 1. PBV is MV/BV (book value of equity) at the end of t – 1. EVEBITDA is EM at the end of t – 1. MV is market value of equity at December t – 1.
0.0000 0.0000 0.3790 0.0000 0.0000 0.0399 MV 0.0836 0.0182 0.0036 0.1305 0.0390 -0.0066 1.0000 0.0000 0.0000 0.0000 0.0000 0.0000 RR2 -0.0451 -0.0207 -0.0211 -0.0505 -0.0451 1.0000 0.0000 0.0000 0.0000 0.0000 EVEBITDA 0.5742 0.4727 0.2282 0.4038 1.0000 0.0000 0.0000 0.0000 PBV 0.4583 0.2397 0.2241 1.0000 0.0000 0.0000 PCFO 0.2221 0.1260 1.0000 0.0000 PE 0.2638 1.0000 PS 1.0000 PS PE PCFO PBV EVEBITDA RR2 MV
Page 21 of 26
Table III – Returns
AR – represent the average return of the corresponding decile portfolio, st. dev. – the standart deviation of that portfolio
Table III a)
Difference between the return of value and glamour portfolios.
diff = - - representing the average return on the decile one porfolio, return on the tenth decile
P/S P/E P/BV EV/EBITDAP/CFO
Page 22 of 26
Table IV – OLS regression Table IV a) Total 67.03 34 0.0006 Kurtosis 1.02 1 0.3115 Skewness 12.57 6 0.0503 Heteroskedasticity 53.43 27 0.0018 Source chi2 df p Cameron & Trivedi's decomposition of IM-test
Prob > chi2 = 0.0018 chi2(27) = 53.43
against Ha: unrestricted heteroskedasticity White's test for Ho: homoskedasticity
Page 23 of 26 Table IV b) Table IV c) _cons .1558898 .0134911 11.56 0.000 .1294472 .1823324 size -.0102418 .0023789 -4.31 0.000 -.0149043 -.0055792 PCFO -.0003924 .0002326 -1.69 0.092 -.0008482 .0000635 EVEBITDA -.0016818 .001143 -1.47 0.141 -.0039222 .0005585 PBV -.0051458 .0018676 -2.76 0.006 -.0088064 -.0014853 PS -.0061226 .0025909 -2.36 0.018 -.0112007 -.0010445 PE .0001607 .0001418 1.13 0.257 -.0001172 .0004387 RR2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .77787 R-squared = 0.0025 Prob > F = 0.0000 F( 6, 58388) = 37.18 Linear regression Number of obs = 58395
_cons .1385591 .0071298 19.43 0.000 .1245847 .1525335 size -.0061516 .0014167 -4.34 0.000 -.0089283 -.0033749 EVEBITDA -.002399 .0007397 -3.24 0.001 -.0038487 -.0009492 PBV -.01133 .0014184 -7.99 0.000 -.0141099 -.00855 PS -.0056232 .0018986 -2.96 0.003 -.0093445 -.001902 PE .000074 .0001125 0.66 0.511 -.0001465 .0002945 RR2 Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .67902 R-squared = 0.0037 Prob > F = 0.0000 F( 5, 95913) = 105.91 Linear regression Number of obs = 95919
Page 24 of 26
Chart I
Average annual return (P/S Deciles, June 1964 – June 2014)
Chart II
Page 25 of 26
Chart III
Average annual return (P/CFO Deciles, June 1964 – June 2014)
Chart IV
Page 26 of 26
Chart V
