Alternative Valuation Techniques For Predicting UK Stock Returns

Alternative Valuation Techniques For Predicting UK Stock Returns by Christian L. Dunis * and Declan M. Reilly ** (Liverpool Business School and CIBEF *** ) March 2004 Abstract Using daily data over the period 31st December 2000 to 31st December 2002 and 5 variables to categorise a panel of 689 stocks from the FTSE All-Share index along a value-growth dimension, this paper analyses the investment returns obtained from the best decile portfolios of "growth" stocks and "value" stocks. The results suggest that a value-growth factor is significant in the UK stock market no matter which of the five relative valuation techniques are used. Value stocks outperformed growth stocks, on average, for all five relative valuation techniques used during the period studied, both absolutely and after adjustment for risk. Value stocks also outperformed the market, on average, for all five relative valuation techniques, both absolutely and after adjustment for risk. The lowest market capitalisation decile portfolio was the best performing relative valuation technique, in terms of its risk-adjusted return, and the level of returns for all the value deciles were significant no matter which of the five methods of selecting the value decile portfolio was used. According to the correlations, there was a fairly high degree of similarities in the daily variation of the different value deciles stock returns. One can therefore use the term value across all of the techniques used to select value portfolios, as they are clearly detecting similar investment return characteristics. However, there is enough of a difference in the variation of returns to make it possible for an analyst of investment funds to detect if a certain portfolio is composed using a particular valuation technique. Keywords: cash flow/price ratio, dividend yield, growth portfolio, market capitalisation, price/book value ratio, price/earnings ratio, valuation techniques, value-growth spread, value-market spread, value portfolio. * Christian Dunis is Professor of Banking and Finance at Liverpool Business School and Director of CIBEF (E-mail: cdunis@totalise.co.uk). ** Declan Reilly is an Associate Researcher with CIBEF (E-mail: d.m.reilly@lycos.co.uk). *** CIBEF Centre for International Banking, Economics and Finance, JMU, John Foster Building, 98 Mount Pleasant, Liverpool L3 5UZ. 1

1. INTRODUCTION Capaul et al. (1993) examine whether value stocks outperform growth stocks in terms of their investment returns. They go about this task by defining value and growth stocks and analysing whether there is any significant difference in the level of investment returns. Value companies are ones whose securities can be purchased for prices that are low in relation to their estimated underlying values. Growth companies are ones with substantial growth prospects and this is usually recognised in the price. Their scheme used only one classificatory measure the current price per share divided by the most recently reported book value per share. Intuitively, a company s share price represents investors assessments of future prospects, while its book value represents accountants representation of its past costs. Therefore, the greater a company s prospects for future growth, the higher the ratio of price/book value. The motivation for this paper is to extend the format set out in Capaul et al. (1993) and attempt to identify significant differences, in terms of profitability, between value stocks and growth stocks. While they used only one classificatory measure, the price/book value ratio, to categorise securities along a value/growth dimension, this paper uses a total of five methods: I. The first classificatory measure is the same as the one in Capaul et al. (1993), the price/book value ratio. II. The second method uses the same means of appraisal as Basu (1977), the price/earnings ratio this assumes that a security s price represents investors assessments of future prospects, while its earnings figure represents accountants representation of its past profit/loss. Therefore, the higher the ratio of price/earnings, the greater a company s prospects for future growth, according to investors opinions. III. The third test looks at the cash flow/price ratio, following Lakonishok et al. (1994), amongst others - this again assumes that a security s price represents investors assessments of future prospects, while its cash flow represents the cash generated or lost by a firm over the reporting period, therefore giving an indication of financial strength. Therefore, the lower the ratio of cash flow/price, the greater a company s prospects for future growth, according to investors opinions. IV. The fourth approach, following Keppler (1991a), examines the dividend yield, i.e. the annual dividend to share price ratio: this supposes that the stock price embodies investors judgements of a company s future outlook, while its dividend is the amount paid out by a firm to its ordinary shareholders. Therefore, the greater a company s prospects for future growth, according to investors opinions, the lower the dividend yield. V. The final means of categorising stocks along a value/growth dimension is through the market capitalisation of firms, as used by Banz (1981) this assumes that investors have priced in their opinions of future prospects, therefore giving a value of the firm. The firms with a small market 2

capitalisation are seen as value companies, while firms with a large market capitalisation are seen as growth companies 1. A difficult question to solve, and one in which this paper aims to contribute some answers, is deciding which relative valuation technique to use, and which is best, in attaining superior returns. It is clear from the historical evidence that value stocks earn higher returns on average than growth stocks 2, and that is the same conclusion arrived at by this paper. It has also been shown that value stocks outperform the market 3 again this is backed up by our findings. The results in this paper show that there is a significant value-growth factor in the UK market. This means that companies that are classified as being value will, on average, outperform those that are deemed to be growth. This paper also shows that the value decile 4 portfolios, constructed using any of the relative valuation techniques, beat the market, in terms of its investment returns and after adjustment for risk, over the period between 31 st December 2000 and 31 st December 2002. The value decile portfolio selected using the market capitalisation technique is the best performing, according to its investment return over the period studied and its risk-adjusted return. Both the price/earnings ratio technique and the cash flow/price ratio technique are beaten by the market capitalisation technique, but they still have Sharpe ratios above one. These portfolios, in most investment managers considerations, would be attractive types of investment vehicles. Our paper further demonstrates that there is a fairly high correlation between the daily variations of the different value decile portfolios stock returns. This shows that the selection procedures used by the different relative valuation techniques are picking up on similar investment return characteristics within the companies invested. However, if the style of an investment fund is already known to be tilted towards value stocks, there is enough of a difference in the variations of the stocks returns so that the relative valuation technique used by an investment fund can be determined by analysing the historical covariance of the fund s returns with those of portfolios of value stocks composed using one of the five methods outlined. The paper is organised as follows. Section 2 presents a comprehensive literature review of relative valuation studies. Section 3 briefly discusses the earnings vs. cash flow debate as these two measures are quite important in 1 This maybe a very flimsy assumption as their will undoubtedly be value companies, according to other valuation techniques, that are classified as having a large market cap and hence a growth company. However, the assumption is there to see which relative valuation techniques are the most effective even if the selected portfolios contradict each other. 2 For example, see Basu (1977, 1983), Ibbotson (1986), Capaul et al. (1993) and Lakonishok et al. (1994). 3 For example, see Ibbotson (1981,1986), Oppenheimer (1984) and Levis (1989). 4 A decile is 10% of the sample, and when calculating investment returns of a decile portfolio there is equal weighting for each company within a decile. 3

our analysis. Section 4 contains the data and methodology followed in the paper. Section 5 presents our results while section 6 offers some concluding remarks. 2. LITERATURE REVIEW As mentioned above, past studies show that, in general, value stocks outperform not only growth stocks, but also the market as a whole. 2.1 Price/Book Value Ratio Capaul et al. (1993) examined the comparative investment returns of low price to book value stocks (value stocks) and high price to book value stocks (growth stocks) in the UK, the United States, France, Japan, Switzerland and Germany. The stocks were comprised of the Standard & Poor s 500 Index for the United States, and the Morgan Stanley Capital International indices for the other countries. The results showed that value stocks significantly outperformed growth stocks in every country, in terms of the cumulative investment return over the 11.5 period. The UK has a spread of 31.5% over the period. Capaul et al. (1993) state that: value stocks outperformed growth stocks on average in each country during the period studied, both absolutely and after adjustment for risk. Ibbotson (1986), DeBondt and Thaler (1987), Fama and French (1992) and Lakonishok et al. (1994) all found the same relationship to exist within their studies. Ibbotson (1986) found that stocks with a low price/book value ratio, the socalled value stocks, performed significantly better, in terms of returns, than stocks with a high price/book value ratio, the growth stocks, over their 18 year study on the New York Stock Exchange (NYSE). DeBondt and Thaler (1987) noted that an investment return revision seems to take place with the value stocks on the New York and American Stock Exchanges, as companies that performed poorly in the four years prior to portfolio formation subsequently performed best in the four years after portfolio formation. And vice versa. Fama and French (1992) examined the effects of price as a percentage of book value, together with market capitalisation, on investment returns for all non-financial New York Stock Exchange (NYSE), American Stock Exchange (ASE) and NASDAQ companies included in the CRSP 5 file. The results indicated that the companies with smallest market capitalisation at the lowest prices in relation to book value provided the best returns, a 23.0% annual investment return. Lakonishok et al. (1994) studied the effect of price as a percentage of book value on investment returns. This paper ranked all companies listed on the New York Stock Exchange and the American Stock Exchange according to their stock price as a percentage of book value and then sorted them into deciles. Again, the results indicated that, generally, the lower the price/book value ratio the higher the investment return. The results from their study also showed significant outperformance of the UK domestic market indices by the value portfolios. 5 Centre for Research in Security Prices. 4

With so much evidence suggesting value stocks produce better investment returns over time than growth stocks Lakonishok et al. (1994) took an interesting angle at the topic by examining the consistency of investment returns for low price/book value ratio companies. They did this by comparing the investment returns of value and growth portfolios over one-year, threeyear and five-year holding periods for the period between April 30 th 1968 and April 30 th 1990. The results showed that as the holding period of portfolios increased the low price/book value ratio (value) portfolio becomes the better choice of investment. Lakonishok et al. (1994) attempted to examine whether higher returns of low price to book value stocks were due to greater risk, as suggested by Fama and French (1992), Ball and Kothari (1989) and Chan (1988). Lakonishok et al. (1994) measured monthly investment returns in relation to price as a percentage of book value between April 30 th 1968 and April 30 th 1990. The low price/book value ratio stocks outperformed the high price/book value ratio stocks in the market s worst 25 months and in the other 88 months where the stock market declined, although not significantly. Furthermore, their results suggest, that the value strategy (low price to book ratio) does not expose investors to greater downside risk. 2.2 Price/Earnings Ratio Basu (1977) examined the relationship between price/earnings ratios and investment results. The study covered New York Stock Exchange listed companies from 1957 to 1971. The price/earnings ratios for all the stocks were calculated at year-end, ranked from highest to lowest price/earnings ratio, and then sorted into quintiles 6. The stocks in each quintile were equally weighted and the stocks were bought on 31 st December and sold after one year. The lowest price/earnings ratio quintile had the highest average annual rate of return at 16.3% over the 14-year period. This was 7.0% higher than the quintile with stocks among the highest price/earnings ratios. $1 invested in the lowest price/earnings ratio group over the 14-year period would have grown to over $8, whereas if that same $1 had been invested in the highest price/earnings ratio group over the 14-year period it would have grown to approximately $3.5 7. Ibbotson (1986), Oppenheimer (1984), Basu (1983) and Lakonishok et al. (1994) all found the same relationship. Oppenheimer (1984) examined the investment performance of the low price/earnings ratio stocks and followed the procedures laid out by Benjamin Graham 8. The criteria required the purchase of securities of companies in which the earnings yield 9 was at least twice the AAA bond yield, and the company s total debt (current liabilities and long-term debt) was less than its 6 A quintile is 20% of the sample, and there is equal weighting in each quintile. 7 That is less than half the cumulative return of the lowest price/earnings ratio quintile. 8 Benjamin Graham was a seminal figure on Wall Street and is widely acknowledged to be the father of modern security analysis and value investing. His most famous works include: Security Analysis (1934) and The Intelligent Investor (1949). 9 The earnings yield is the inverse of the price/earnings ratio. 5

book value. According to Oppenheimer (1984), Graham advised that the stocks that met the criteria were held for either two years, or until 50% price appreciation occurred, whichever came first. Oppenheimer (1984) tested stocks on the New York and American Stock Exchanges that met Benjamin Graham s criteria on each March 31 st from 1974 up until 1980. During this period, if the selection criteria had been employed, a mean annual return of 38% would have been achieved. This is significantly higher than the 14% mean annual return of the CRSP index of NYSE-AMEX securities over the same period. Every holding period in the test produces a positive return. The minimum return was 26.16%, while the maximum return was 46.68%. 2.3 Cash Flow to Price Ratio (or price/cash flow ratio) Keppler (1991b) examined the relationship between price to cash flow ratios and investment returns for companies throughout the world. The study covered the period between January 31st 1970 and December 31 st 1989 and assumed an equal weighted investment each quarter in 18 countries MSCI national equity indices. The 18 countries were: Australia, Austria, Belgium, Canada, Denmark, France, Germany, Hong Kong, Italy, Japan, the Netherlands, Norway, Singapore, Spain, Sweden, Switzerland, the United Kingdom and the United States. Every quarter, the country indices were ranked according to the ratio of price to cash flow and sorted into four quartiles. The total investment return was measured for each of the four quartile groups over the subsequent 3 months. Cash flow was defined as net earnings after tax, minority interests, dividends on preferred stock, and distribution to employees, plus reported depreciation on fixed assets for the latest available 12-month period. The study indicated that the most profitable strategy was investment in the lowest price to cash flow (or highest cash flow to price) quartile. This strategy produced a 19.17% compound annual return in local currencies (and 20.32% in U.S. dollars) over the period between January 31 st 1970 and December 31 st 1989. The least profitable strategy was investment in the highest price to cash flow (or lowest cash flow to price) quartile, which produced a 4.37% compound annual return in local currencies (and 5.63% in U.S. dollars). The comparable return over the January 31, 1970 through December 31, 1989 period for the Morgan Stanley Capital International World index was 12.45% in local currencies and 13.58% in U.S. dollars, showing that the value quartile (consisting of the high cash flow to price ratio stocks) outperformed the market over the period. Lakonishok et al. (1994) examined the effect of price/cash flow ratios on investment returns for all companies listed on the New York Stock Exchange and the American Stock Exchange over the period between April 30, 1968 and April 30 th, 1990. The decile portfolios were held for five years, and the average annual year-by-year investment returns, the average annual five-year returns and the average cumulative total five-year returns were calculated. The investment returns were equally weighted. The highest price/cash flow ratio (or lowest cash flow/price ratio) decile produced the lowest average annual return over the five-year period of 9.1%. This was 11% lower than that of the lowest price/cash flow ratio (or highest cash flow/price ratio) decile (20.1%). 6

2.4 Dividend Yield Keppler (1991a) examined the relationship between dividend yield and investment returns for companies throughout the world. The study covered the 20-year period between December 31 st 1969 and December 31 st 1989 and assumed an equally weighted investment each quarter in 18 countries MSCI national equity indices. The 18 countries were: Australia, Austria, Belgium, Canada, Denmark, France, Germany, Hong Kong, Italy, Japan, the Netherlands, Norway, Singapore, Spain, Sweden, Switzerland, the United Kingdom and the United States. Every quarter, the country indices were ranked according to dividend yield and sorted into four quartiles. The total investment return was measured for each of the four quartile groups over the subsequent 3 months. The most profitable strategy was to invest in the highest dividend yield quartile. The compound annual investment return for the highest yielding stocks was 18.49% in local currencies and 19.08% in U.S. dollars over the 20- year period. The least profitable strategy was investment in the lowest dividend yield quartile, which produced a 5.74% compound annual return in local currency (and 10.31% in U.S. dollars). The Morgan Stanley Capital International World Index return over the same period was 12.14% in local currency and 13.26% in U.S. dollars, showing that the value quartile (consisting of the highest dividend yield stocks) outperformed the market, over the period studied. Levis (1989) examined the association between dividend yield and investment returns from January 1955 through December 1988. Using a sample of 4,413 companies listed on the London Stock Exchange, all listed companies each year were ranked according to dividend yield and sorted into deciles. The highest dividend yield decile earned an annual investment return of 19.3%, whilst the lowest dividend yield decile had an annual investment return of 13.8%. Both were greater than the annual investment return for the FTSE All Share Index (13.0%) over the same thirty-four-year period. 2.5 Market Capitalisation Banz (1981) ranked all New York Stock Exchange listed companies according to market capitalisation each December 31st from December 31 st 1925 through December 31 st 1980 and sorted the firms into quintiles, and measured the annual investment returns, on a market capitalisation weighted basis, for each quintile. The quintile with the highest compound annual return was the group with the smallest market capitalisation firms, at 12.1% per annum. This was 3.2% higher than the compound annual return of the worst performing group, the large market capitalisation group, at 8.9%. The difference would not be very significant on an individual year s basis but over the whole of the period tested, from 1926 to 1980, the difference is huge. The value of $1 invested in the largest market capitalisation quintile on 31 st December 1925 at the end of 1980 would be worth $108.67. The value of $1 invested in the smallest market capitalisation quintile on 31 st December 1925 at the end of 1980 would be worth $524, almost five times the largest market capitalisation quintile s cumulative aggregate amount. 7

Reinganum (1983) studied the investment returns of all stocks listed on the New York and American Stock Exchanges from 1963 through 1980. All companies listed on the two stock exchanges were ranked according to market capitalisation and sorted into deciles at the beginning of each year. The results showed that the mean average annual returns declined as size of the market capitalisation decile increased. The smallest market capitalisation decile had an average annual investment return of 32.8%, whilst the largest market capitalisation decile had an average annual investment return of just 9.5%. This constitutes a small-large cap spread of 23.3% per annum. 3. THE EARNINGS VS. CASH-FLOW DEBATE This part discusses the theory behind the valuation techniques and any contentious issues concerning the financial variables retained. Due to the fact that the earnings figure and the cash flow figure of a firm are essentially attempting to capture similar information, the financial position of a firm, this section examines the differences between the two figures and which is the best indicator of performance. The dividend decision a firm has to undertake is also considered, with respect to how a firm s policy choice will affect the investment returns to an investor. Accounting earnings, it could be argued, can be manipulated at least for individual periods, through the use of creative accounting techniques and strategic allocations, especially at the equity level (i.e. net income). In this instance, the cash flow measure would be a better indicator, as it is a difficult figure to fiddle. No business accepts earnings as payment for goods and services delivered. They require cash. So, looking at the cash flow statement provides a good signal of whether a firm has a healthy financial position, in terms of being able to meet payments and use cash to start new projects. A firm with positive earnings and negative cash flows means that cash is leaking out of the company. This could be due to the fact that the firm has more cash outflows than inflows in terms of its operating activities, which would mean the business is seen to be performing badly despite the positive earnings. It could also be due to the financing of projects the firm is undertaking, in which case the firm s accounts would need closer examination. Despite the openness of the earnings figure to manipulation it is still an important figure. The whole purpose of using accrual accounting is to gain a true insight into how a business is performing by including goods and services that have been delivered and received (but not yet paid for, in terms of cash). As a firm starts receiving cash flows from its current operations it is faced with a decision. Should it reinvest the cash back into the business, or should it pay it out to equity investors? The importance of dividends (or lack of it) can clearly influence our results and, more generally, the level of returns of a firm. 8

The debate over the importance of dividend policy was first seriously introduced and discussed in Miller and Modigliani (1961), who suggest that both firm financing and dividend policy was irrelevant for firm investment decisions and independent of the value of the firm. They argued that firms that pay more dividends offer less price appreciation but must provide the same total return to stockholders, given their risk characteristics and the cash flows from their investment decisions. Therefore, if there are no taxes, or if dividends and capital gains are taxed at the same rate, investors should be indifferent to receiving their returns in dividends or price appreciation. A second school of thought argues that dividends create a tax disadvantage for the investors who receive them because they are generally taxed more heavily than the alternative of price appreciation (e.g. Damodaran, 1999). According to this principle, dividend payments should decrease firm value and reduce the returns to equity holders, when taking into consideration the various effects of the differing taxation rates. As a result, firms will be better off if they either retained their earnings that they were considering paying out as dividends or they repurchased their own stock. A third school of thought on dividend policy argues that dividends are good because some groups of investors are fond of dividends and that dividends also convey a positive signal to the outside world in terms of the firm s future prospects (e.g. Healy and Palepu, 1988). 4. DATA AND METHODOLOGY This section presents the data used and the methodology carried out to perform the tests. The primary objective of the tests was to ascertain whether value stocks outperformed, not only growth stocks, but also the market as a whole. To do this, the value and growth decile portfolios were constructed to enable comparisons to be carried out. 4.1 Data Set and Price Ratios The first step in the process was to select an appropriate data set. It was decided that all available stocks on the FTSE All-Share Index, from Thomson s Datastream would make up the universe of data in the studies and the market proxy would be the FTSE All-Share Index. The limitation of the Datastream databank meant that only the most up-to-date construction of the FTSE All-Share Index could be obtained. This meant that when the ranking of the stocks and composition of portfolios occurred there were inherent biases in the universe chosen, as stocks that may have been included in the FTSE All-Share one-year ago might have dropped out of the current set. This survivorship bias should be noted as it may skew the results to the positive side of returns 10, but the data set chosen is sufficiently large to eliminate most of this bias. The FTSE All-Share index is market capitalisation weighted and represents 98-99% of the UK market capitalisation and is a good proxy for the UK market 10 This could occur because companies that would have been included in the FTSE All-Share, say a year ago, may have gone out of business and therefore would not be included in the study. 9

as a whole. The FTSE All-Share index is the aggregation of the FTSE 100, FTSE 250 and FTSE Small Cap Indices. The composition of the FTSE All-Share on the 4 th July 2003, obtained from Thomson s Datastream, was used: at the time, 689 stocks made up the universe. Data on share price, market capitalisation, price/earnings ratio, cash flow/price ratio, market price/book value ratio and dividend yield was obtained for every stock included in the FTSE All-Share for daily data from 31 st December 2000 up until 31 st December 2002 (506 observations for each variable or 3036 observations for each company and over 2 million observations in total for our entire universe). As noted by Cottle et al. (1988), a basic principle in finance is that no investment regarding a common stock can properly be made except in the light of a specific price, which is usually the current market price but may be some anticipated or calculated figure. Comparing the price with earnings, dividends, asset value, and cash flows can be very useful for security analysts to make their investment decisions. The price ratios are usually calculated on the basis of the last full year s results or the latest balance sheet. However, some financial service firms, such as brokerage or fund management firms, may use estimated figures for some of the calculations, or perhaps calculate a figure with a mix of recent past results and estimated future results. Using forecasts would not be a good measure for this study as we want to be able to identify the true value of stocks, and more specifically underpriced stocks. Estimated figures are too open to manipulation from brokerage firms who may have conflicts of interests. 4.2 Methodology for Relative Valuation Tests Not all of the stocks in the universe had data on the financial variable (i.e. the relative valuation method) and/or price, so they were omitted. The remaining number of stocks on 31 st December 2000 was then ranked according to the financial variable retained. The ranking criteria are displayed in table 4.1. On 31 st December 2001, the remaining number of stocks, after readjustment for missing data, was also then ranked according to the financial variable. For both dates (31 st December 2000 and 31 st December 2001), the stocks were then sorted into deciles 11 according to the study s ranking criteria. Table 4.1 Ranking Criteria Study 12 From To price/book value ratio Highest 13 Lowest price/earnings ratio Highest 14 Lowest cash flow/price ratio Highest Lowest dividend yield Highest Lowest 15 market capitalisation Highest Lowest 11 A decile is 10% of the sample, and there is equal weighting in each decile. 12 When dealing with negative ratios the ratios that are less negative are ranked as being lower than those that are more negative. 13 A negative ratio is classified as being higher than a positive one. 14 A negative ratio is classified as being higher than a positive one. 15 Note that some companies choose to pay no dividends. If this is known, then they are included in the study. 10

The number of stocks in each study is shown in table 4.2: Table 4.2 Number of stocks comprising each study Number of stocks Study Year2001 Year2002 price/book value ratio 659 672 price/earnings ratio 571 576 cash flow/price ratio 546 558 dividend yield 663 675 market capitalisation 663 675 Each decile portfolio was equally weighted and the portfolios were assumed to be held for one year. From this we get two years of results, the advantage being that patterns that may emerge from one year may or may not be apparent in another year, so we have scope to analyse the trends during that period 16. The individual one-year holding periods are from 31 st December 2000 to 31 st December 2001 (also referred to as 2001), and from 31 st December 2001 to 31 st December 2002 (also referred to as 2002). The two-year holding period of a decile portfolio consists of holding the decile portfolio for 2001 and then switching to the new rebalanced 17 portfolio for 2002 (i.e. 31 st December 2000 to 31 st December 2002). Daily returns were calculated for every decile portfolio for an individual oneyear holding period. From this annualised returns for the financial variable s decile portfolios were calculated. Also, calculations were performed for the cumulative return of the portfolios over the test periods: annualised volatility, Sharpe ratio, maximum daily profit, maximum daily loss, maximum drawdown, percentage of winning and losing days, number of up and down days during the test period, average gain in up periods, average loss in down periods, average gain to loss ratio and T-statistic of the profit level of the decile portfolios over the test period. Once all the portfolios have been constructed and the returns obtained, a comparison between value and growth stocks can be constructed, and comparisons with the market s performance can be made. The decile portfolio used from each study to represent either the value or growth portfolio is outlined in table 4.3. Table 4.3 value and growth deciles Study value decile growth decile price/book value ratio Lowest Highest price/earnings ratio Lowest Highest cash flow/price ratio Highest Lowest dividend yield Highest Lowest market capitalisation Lowest Highest 16 Due to the labour intensive nature of constructing the portfolios with a panel of 689 stocks, it was felt that two years of daily data, although a relatively short period, would be sufficient for this research. 17 The portfolio is rebalanced according to the financial variable rankings on 31 st December 2001. 11

4.2.1 Comparison of value and growth stocks The comparison of value and growth stocks is performed by calculating the value-growth spread, which is the growth decile s return subtracted from the value decile s return. By examining the value-growth spread we can then see if a particular valuation technique is preferred, in terms of outperforming the growth decile. We can examine this trend by looking at the cumulative daily value-growth spread over the two years to see if the value decile portfolios significantly outperform the growth deciles. This can also be examined by looking at the annualised value-growth spread, the annualised volatility of the daily value-growth spreads, and the T-statistics associated with the value-growth spreads, to see if the spread is significant. The correlations between the value-growth spreads of the different relative valuation techniques can also be examined to see if there is a link in the spreads over time, arising from different valuation techniques. 4.2.2 Comparison of the value portfolios Assuming that the valuation techniques selecting the value stocks are better performing than the growth portfolios, and this is a pretty safe assumption in view of the existing literature, it would be prudent to see which technique is the best performing over the period. So, we will graphically make a comparison of the cumulative daily returns of the value deciles over the two years. The performance can also be assessed statistically by looking at the annualised returns, the annualised volatility of the daily returns, the riskadjusted returns (Sharpe ratio) and the T-statistics associated with the returns, to see if the returns are significant. It would also be beneficial to examine the correlation between the value decile portfolios arising from different valuation techniques. This will give us an indication of whether it really makes a difference when deciding which valuation techniques to use. Of course, this decision will primarily be based upon the overall return and risk characteristics of the portfolios, but if we have two similar performing portfolios it is useful to know if they are behaving in the same way over time, because the different valuation techniques could be, in effect, selecting stocks with the same characteristics, if not exactly the same stocks. 4.2.3 Comparison between value stocks and the market The valuation techniques investment returns are then compared to the market s 18 investment returns to see if any of the valuation techniques have significantly outperformed the market. The comparison of value and market stocks is performed by calculating the value-market spread, which is the market s return subtracted from the value decile s return. By examining the value-market spread we can then see if a particular valuation technique is preferred in terms of outperforming the market. We can examine this trend by looking at the cumulative daily value-market spread over the two years to see if the value decile portfolios significantly outperform the market. This can also be examined by looking at the annualised value-market spread, the annualised volatility of the daily value-market spread, and the T-statistics associated with the value-market spread, to see if the spread is significant. 18 As mentioned previously, the market proxy is the FTSE All-Share index price. 12

A correlation matrix can be used to see if any of the value-market spreads moves together. It would be beneficial to know this as it will give us an indication of how closely related the types of valuation technique are to each other, in terms of their performance compared to the market. The primary measures of performance for this research are the level of the risk-adjusted return (Sharpe ratio) for each value decile portfolio, the valuemarket spread and the value-growth spread. Therefore, comparisons of performance can be made between value decile groups belonging to different valuation techniques; between different styles of investing (value and growth); and also with the market proxy. 5. EMPIRICAL RESULTS 5.1 Value Investment Strategies We try to determine which technique is the best performing over the period. Figure 5.1 graphically makes a comparison of the cumulative returns of the value deciles over the two years. Figure 5.1 value portfolios 60.00% Value Portfolios against Market Low PE Ratio (Value) decile High Cash Flow/Price Ratio (Value) decile Market (FTSE All-Share Index) Low Price/Book Value Ratio (Value) decile High Div. Yield (Value) decile Low Market Cap. (Value) decile 40.00% Cumulative Percentage Return 20.00% 0.00% -20.00% 02/01/01 02/02/01 02/03/01 02/04/01 02/05/01 02/06/01 02/07/01 02/08/01 02/09/01 02/10/01 02/11/01 02/12/01 02/01/02 02/02/02 02/03/02 02/04/02 02/05/02 02/06/02 02/07/02 02/08/02 02/09/02 02/10/02 02/11/02 02/12/02-40.00% -60.00% Date Figure 5.1 shows that, over the 2-year period retained 19, when using market capitalisation as a technique of selecting a value decile, value stocks produce an absolute cumulative return of approximately 50% (49.64%). This was the largest cumulative return out of all the valuation techniques, suggesting that market capitalisation is the best performing valuation technique. Figure 5.1 also shows that, over the two years, when using the price/earnings ratio as a relative valuation technique, value stocks produced a cumulative return of over 35% (36.22%). The valuation technique of using the cash flow/price ratio shows a cumulative return (29.0%), which is less than the 19 The 2-year holding period consists of holding the decile portfolio for the year 2001 and then switching to the new rebalanced portfolio (rebalanced according to the financial variable rankings on 31 st December 2001) for 2002. 13

price/earnings ratio over the 2 years. The lowest cumulative returns produced by the value deciles were by the low price/book value ratio decile (7.5%) and the low dividend yield decile (8.74%). However, these two techniques were still much better performing than the market (FTSE All-Share Index), which produced a cumulative return of -40.26%. It is quite clear that, from figure 5.1, value stocks produced good levels of returns, for all of the valuation techniques, over the period studied. However, was this level of return significant in every case? Table 5.1 shows some evidence on this point. The first three columns show the annualised daily returns, annualised volatilities and Sharpe ratios for each of the valuation techniques and the market (FTSE All-Share Index). The final column shows the T-statistic associated with each of the spread returns. Table 5.1 value deciles and Market Descriptive Statistics Value decile (PE Ratio) Value decile (CF/P Ratio) Value decile (P/BV Ratio) Value decile (Div. Yield Ratio) Value decile (Market Cap Ratio) Annualised Returns Annualised Volatilities Sharpe Ratio T-statistic 18.11% 12.83% 1.411 22.400 14.50% 12.87% 1.127 17.884 3.75% 10.79% 0.347 5.512 4.37% 12.05% 0.363 5.762 24.81% 12.02% 2.064 32.766 Market -20.13% 22.59% -0.891-14.143 The results in table 5.1 show that the all the value decile portfolios were significantly different from zero, with T-statistics well above 2.0. The results also show the return provided per unit of risk (the Sharpe ratio) for holding the value deciles and the market over the study period. The value deciles constructed using the price/earnings ratio, cash flow/price ratio and the market capitalisation have a Sharpe ratio of above one. This should be an attractive proposition for investors as it means that over that period these value portfolios earned a higher return than the amount of risk that was borne. The lowest market capitalisation (value) decile is the best performing, in terms of the risk-adjusted return (2.064). In fact, the lowest market capitalisation decile produces a level of return that is more than twice the risk associated with holding the portfolio. The period from 31 st December 2000 through 31 st December 2002 saw a general downward trend in the market, however, value strategies showed a general upward trend during the period. This suggests that perhaps value strategies move in opposite directions to the market. Table 5.2 tests this possibility. Table 5.2 shows that the market is positively correlated with all of the value strategies. The average of the correlation between the market and the value portfolios is 0.6092. This suggests that the value deciles follow some of the 14

pattern of the market s direction and are influenced by the market variation to some degree but there is enough scope for diversification benefits. This is useful to know for a fund manager who holds the market portfolio, as the portfolio could have been tilted towards holding more of the value decile s stocks to gain a higher return and lower risk, whilst still retaining some of the market s variation of return characteristics. Table 5.2 Correlations value portfolios and the Market (using daily returns) Correlation Matrix (Value strategies and market) PE (Value) CF/P (Value) P/BV (Value) Div Yield (Value) Market Cap (Value) Market PE (Value) 1 0.8814 0.8620 0.8765 0.8814 0.5724 CF/P (Value) 0.8814 1 0.8636 0.8882 0.8516 0.6394 P/BV (Value) 0.8620 0.8636 1 0.8789 0.8592 0.6458 Div Yield (Value) 0.8765 0.8882 0.8789 1 0.8255 0.6559 Market Cap (Value) 0.8814 0.8516 0.8592 0.8255 1 0.5327 Market 0.5724 0.6394 0.6458 0.6559 0.5327 1 Table 5.2 also shows that the correlation between the different value strategies are highly positively correlated, suggesting that there is some commonalities in the characteristics of the value decile portfolios (and perhaps the selection of value stocks) for all the relative valuation techniques. The average correlation between the value decile portfolios is 0.8668. In summary, the lowest market capitalisation decile portfolio is the best performing valuation technique in terms of its risk-adjusted return. The price/earnings ratio and cash flow/price technique produced good Sharpe ratios (above 1) and the level of returns for all the value deciles were significant no matter how the decile portfolios were selected. There is a degree of commonality in the daily variation of the value deciles stock returns. This means we can use the term value across all of these techniques, as they are clearly detecting similar investment return characteristics. However, there is enough of a difference in the variation of the returns to make it possible for an analyst of investment funds to detect if a particular portfolio is composed using any of the valuation techniques 20. 5.2 Value-Growth Spreads Value-growth spreads are computed by subtracting the return on the growth decile portfolio from that on the corresponding value portfolio, for each of the valuation strategies. Several interpretations are possible when examining these spreads. The resulting spreads can be considered the return advantage (if positive) or disadvantage (if negative) resulting from holding value stocks rather than growth stocks. It can also be considered the return pick-up (if positive) or drop (if negative) associated with a prior decision to switch from growth to value stocks. It can also be regarded as the net return on a swap, whereby an investor agrees to receive the return on the value stocks in return for a promise to pay the return on growth stocks. Figure 5.2 shows the cumulative returns of the value-growth spreads for each of the different valuation techniques (price/book value ratio, price/earnings ratio, cash flow/price ratio, dividend yield and market 20 For an in-depth discussion on this point see Sharpe, W. F. (1988, 1992). 15

capitalisation). For ease of comparison all of the spreads are plotted on the same graph. Increases indicate periods in which the value decile portfolio outperforms the growth decile portfolio; decreases indicate periods in which the value decile portfolio underperforms the growth decile portfolio. Figure 5.2 Value-Growth Spreads 120.00% Cumulative Value-Growth Spreads Value-Growth Spread (PE Ratio) Value-Growth Spread (Cash Flow/Price Ratio) Value-Growth Spread (P/BV Ratio) Value-Growth Spread (Div. Yield) Value-Growth Spread (Market Cap) 100.00% Cumulative Percentage Return 80.00% 60.00% 40.00% 20.00% 0.00% -20.00% 02/01/01 02/02/01 02/03/01 02/04/01 02/05/01 02/06/01 02/07/01 02/08/01 02/09/01 02/10/01 02/11/01 02/12/01 02/01/02 02/02/02 02/03/02 02/04/02 02/05/02 02/06/02 02/07/02 02/08/02 02/09/02 02/10/02 02/11/02 02/12/02 Date Figure 5.2 shows that, over the 2 year-period retained, when using the price/earnings ratio or the market capitalisation as an indication of value and growth, the value stocks outperform the growth stocks by approximately 90% (92.1% and 88.96%, respectively). These were the biggest spreads of returns between value and growth stocks out of all the valuation techniques, suggesting that the valuation techniques of price/earnings or market capitalisation distinguishes the classification of stocks the most easily. Over the two-year period, the cumulative outperformance of value stocks over growth stocks, when using the cash flow/price ratio technique was 78.2%. However, at its peak (in September 2002), value stocks, using the cash flow/price ratio technique, outperformed growth stocks by almost 100%. The cash flow/price ratio relative valuation technique had the biggest valuegrowth spread for the majority of the test period, but fell away in the final few months of the two-year study. The smallest cumulative spread of returns over the two-year period between the value and growth stocks is obtained by using the price/book value ratio technique of relative valuation (39.98%). It is interesting to note the spikes around September 2001. This could indicate the possibility of value stocks significantly outperforming growth stocks in periods of crisis. During that time the FTSE All-Share Index fell by about 15-20% in just a few weeks, in the main following the terrorist attacks in New York. Overall, it is quite clear from figure 5.2 that value stocks did indeed outperform growth stocks, for all of the valuation techniques over the period studied. However, was this outperformance significant in every case? Table 5.3 shows some proof on this point. The first two columns show the annualised daily return spreads and annualised volatilities for each of the valuation techniques. The final column shows the T-statistic associated with each of the spread returns. 16

Table 5.3 Value-Growth Spreads Descriptive Statistics Value-Growth Spread (PE Ratio) Value-Growth Spread (CF/P Ratio) Value-Growth Spread (P/BV Ratio) Value-Growth Spread (Div. Yield) Value-Growth Spread (Market Cap) Annualised Returns Annualised Volatilities T-statistic 46.05% 14.07% 51.968 39.10% 17.41% 35.642 19.99% 11.15% 28.454 22.07% 13.02% 26.908 44.48% 20.75% 34.024 The results show that all the value-growth spreads were significantly different from zero, over the two-year period, with T-statistics well above 2.0. Table 5.3 also shows that the annualised volatilities of the value-growth spreads, especially for both the price/earnings technique and the dividend yield technique, are very close. This suggests the two valuation techniques could be distinguishing value and growth stocks in a similar way, in terms of its investment return spread. Table 5.4, below, confirms that there were some similarities in the selection of the portfolios. Table 5.4 Correlations of Daily Value-Growth Spreads Correlation Matrix (Value-Growth Spreads) PE Spread CF/P spread P/BV Spread Div Yield Spread Market Cap Spread PE Spread 1 0.6511 0.4203 0.6352 0.5569 CF/P spread 0.6511 1 0.4801 0.6887 0.3243 P/BV Spread 0.4203 0.4801 1 0.4026 0.5031 Div Yield Spread 0.6352 0.6887 0.4026 1 0.2629 Market Cap Spread 0.5569 0.3243 0.5031 0.2629 1 It shows that the value-growth spreads for the price/earnings technique and the dividend yield technique were positively correlated with each other. However, the figure was only 0.6352 suggesting a certain amount of independence from each other. The value-growth spreads of the cash flow/price technique and the dividend yield technique were positively correlated with each other. These two techniques showed the highest correlation amongst the five methods of relative valuation (0.6887). The lowest correlation between value-growth spreads over the two-year study was between the market capitalisation spread and the dividend yield spread (0.2629). The average of the off-diagonal values was 0.4925, indicating that the valuation techniques are distinguishing between value and growth stocks in a similar way but there are some major differences in the daily variations of the spreads. To summarise: value stocks outperformed growth stocks over the period covered, and the difference in performance was significant no matter how the decile portfolios were selected. From this, we can say that value and growth 17