The value premium effect and risk: an industry level
perspective
R.F. (Ruurd) van der Leij Student number: 1460951 Msc Business Administration Specialization: Finance
University of Groningen
Faculty of Economics and Business Course code: EBM866A20
Supervisors:
Abstract:
This paper examines the relation between industry affiliation, the value premium effect and risk. Value and growth portfolios are constructed based on the book to market, earnings to price and cash flow to price ratios within ten European industries for the years 2000-2010. Significant value premiums are found in the majority of the ten industries. Moreover, this paper shows a significant value premium effect on both firm and industry level, of which the industry effect is most prominent. Furthermore, differences in size and significance level of the value premiums between industries are found, making it worthwhile to include an industry perspective when building asset pricing models. A positive relation between risk and the value premium effect is found on an industry level, however no uniform relation on a firm level. This makes room for the alternative explanation that the value premium effect could be a result of judgmental biases of investors.
JEL classification: C12, C22, G12
Table of Contents
I. Introduction
4II. Literature review
6A. The value premium effect 7
B. The value premium effect and risk 8
C. The value premium and behavioral explanations 10
III. Methodology
13A. Portfolio formation and returns 13
B. Explanatory variables 14
C. Intra and inter-industry value premium effects 15
D. Value premium and risk 16
IV. Data
17A. Construction of the dataset 17
B. Descriptive statistics of the independent variables 18
C. Descriptive statistics of the returns 20
V. Results
22A. Value premiums and industry affiliation 22
B. Intra and inter-industry value premium effects 26
C. Consistency of the value premium effect between industries 28
D. Variation in time 30
E. Value premiums and risk 31
F. Research imitations and ideas for future research 34
VI. Conclusion
35References
37I. Introduction
Results from academic research have been widely used in equity investment strategies. At the same time, ‘real life’ abnormalities encountered by portfolio managers have formed a base for new academic research. In finance literature, several market anomalies have been documented. Banz (1981) found that smaller firms show higher stock returns than would be expected from the capital asset pricing model (CAPM). De Bond and Thaler (1987) investigated long term return patterns and found long term reversals, the so called winner-loser effect, which they attributed to the overreaction of investors. Basu (1977) found a negative relation between price earnings ratios and the CAPM predicted returns. In line with these market anomalies, the topic of the value premium effect has received a lot of attention from the financial academics as well (Chan and Lakonishok, 2004). Investment managers classify firms with high ratios of book to market equity (B/M), earnings to price (E/P) or cash flow to price (C/P) as value stocks and stocks with low ratios as growth stocks. The effect that value stocks generate higher average returns is called the value premium effect (Fama and French, 1998).
The study on value premiums goes back to 1934 with the publication of Graham and Dodd, stating that value strategies outperform the market (Graham and Dodd, 1934). Throughout the years, several studies have proven the existence of the value premium effect in different countries and throughout different periods in time1. But while there is some agreement that value strategies outperform growth strategies, the reason why they have done so is more controversial (Chan and Lakonishok, 2004). Researchers have come up with two alternative explanations for the value effect. The first states that the high B/M ratio of value stocks also implies higher risk, because they are more prone to financial distress and therefore have a higher discount rate (Fama and French, 1992). Firms with high B/M ratios (value firms) are shown to have earning problems and relatively high levels of financial leverage. The higher returns that these value firms generate are considered to be a rational result of the higher financial distress risk of these firms (Banko, Conover and Jensen, 2006). Advocates of this risk based view have proposed various adjustments to the capital asset pricing model trying to capture and explain this premium in a risk related way (Phalippou, 2008; Fama and French, 1998).
1
The alternative explanation draws on behavioral considerations and agency costs, including psychological biases to explain the effect (Chan and Lakonishok, 2004). They state that the value premium is not the result of systematic risks, but the inability of investors to process and evaluate information correctly (Brouwer, 1996). Studies in psychology have suggested that individuals tend to use simple heuristics for decision making, which opens up the possibility of judgmental biases in investment behavior. In particular, investors may extrapolate past performance too far into the future. Investors become overly optimistic and overvalue stocks that have done well in the past (growth stocks) and undervalue stocks that have shown bad performances in the past (value stocks) (Phalippou, 2008; Brouwer, 1996; Fama and French, 1998).
This paper will look at the value anomaly using an industry perspective. The industry perspective has been used in researching other market anomalies like the winner/loser effect and the momentum effect (Moskowitz and Grinblatt, 1999; Wouters, 2006). Several studies identified a relation between industry wide price reactions and firm specific announcements, supporting the view that industry affiliation may be associated with the value premium effect (Kothari, Shanken and Sloan, 1995). Looking at the value premium effect using an industry perspective can be useful for incorporating the value premium effect in asset management. When the effect turns out to be significantly present in a limited number of industries, it is important to include these findings into the asset pricing models. However, if the value effect is not detectable on industry level and turns out to be a firm level phenomenon, the industry effect has a limited role (Banko, Conover and Jensen, 2006). Furthermore, this paper analyses whether higher risk of value portfolios is a solid explanatory factor for the value premium effect. Focusing on the relationship between the value premium effect, industry affiliation and risk is relatively new. The paper of Banko, Conover and Jensen (2006) which investigates U.S. industries, is by my knowledge the only paper with a similar focus. By focusing on the value premium effect using a European industry perspective, this paper contributes to the existing literature because it creates the option for comparison with U.S data and provides additional insight of the European equity market. The goal of this paper is to determine whether industry affiliation plays a role in the value premium effect and whether risk is a good explanation for this effect.
categorization, making it possible to build portfolios of sufficient size within each industry. Thomson’s Datastream is used for collecting the financial information for a period of ten years, from 2000 till 2010.
The substantial research conducted on the value premium topic, has generated a large methodological base. For the identification of value and growth portfolios, several ratios have been used. For example the book to market ratio, earnings to price ratio, cash flow to price ratio, dividend payout ratio and the Tobin’s Q index. In line with the paper of Fama and French (1998), this paper will use the B/M, E/P and C/P ratio for constructing value and growth portfolios.
By building value and growth portfolios of firms within the ten industries, based on their rank on the above mentioned ratios, this paper findings show a significant value premium effect on both firm an industry level. Furthermore, differences in size and significance level of the value premiums between the industries are found, making it worthwhile to include an industry perspective when building asset pricing models. By looking at financial leverage, earnings uncertainty and betas the relation between risk and the value premium effect has been tested, finding a positive relation between risk and the value premium effect on industry level but no uniform relation on firm level. This makes room for the alternative explanation that the value premium effect could be a result of judgmental biases of investors.
The remainder of this thesis is organized as follows. Section two describes the related literature about the value premium effect and the existing explanations. Section three will describe the methodology, followed by a description of the data in section four. The results of the tests will be presented in section five, followed by the discussion and conclusive remarks in the sixth part of the paper.
II. Literature review
The value premium effect
revealed that value portfolios based on the B/M, E/P and C/P ratio outperformed growth portfolios in both Japan and the United States. Chan and Lakonishok (2004) showed a value premium of 18% in the United States for the period of 1963-1990. Fama and French (1992) sorted stocks on the NYSE, Amex and Nasdaq markets into ten portfolios based on the B/M ratio of the stocks, ranking the highest decile as value portfolios and the lowest decile as glamour portfolios. Comparing the monthly average returns of the portfolios, they found a 1.53% difference in favor of the value portfolio. Lakonishok, Shleifer and Vishny (1994) expanded the research by looking at buy and hold returns over several years after portfolio formation, consequently taking a more long term perspective. They used the same portfolio categorization method based on the B/M ratio, finding similar results as Fama and French (1992). On average the value portfolios outperformed the growth portfolios by 10,5% per year over the five years after the portfolio formation. Chan, Hamao and Lakonishok (1991) looked at the value premium effect in Japanese stocks. Based on the B/M, E/P and C/P ratio they found results that were very similar to the U.S. market, finding a positive difference between value and growth portfolios on all the ratios. Capaul, Rowly and Sharpe (1993) expanded the research to France, Germany, Switzerland and the U.K, finding higher returns for value stocks in all the selected countries. Fama and French (1998) looked at even more countries by including Italy, the Netherlands, Belgium, Sweden, Australia, Hong Kong and Singapore. They found pervasive evidence of the effect in 12 of the 13 countries. They also expended their research with a sample for emergent markets and confirmed the effect in these countries as well. Based on the above mentioned findings, it can be stated that the value premium effect is persuasive in many stock markets all over the world.
overly confident or pessimistic about the future of an industry, instead of a particular firm, which may create conditional patterns in industry returns. Furthermore, they state that momentum strategies are not well diversified because the winners and the losers tend to be from the same industry, so large differences in the anomaly between industries can exist (Moskowitz and Grinblatt, 1999).
There are other studies that have identified the industry effect to be a contributing factor in firm financial performances. Kothari, Shanken and Sloan (1995) looked at reasons for explaining variations in expected returns, finding a significant positive relation between industry B/M ratios and industry stock returns. In a similar way, Banko, Conover and Jensen (2006) looked at the relation between the value effect, industry affiliation and firm risk, finding evidence of the value effect in 15 of the 21 selected industries. They identify the value effect at both the firm and industry level, however much stronger on the firm level.
Chan, Lakonishok and Sougiannis (2001) present evidence that the value effect is detectable in different industries, due to the difference in the amount of intangible assets an industry has. Investments in research and development expenditures are hard to value and can result in misvaluations, potentially enlarging the value effect. They found evidence that the value effect is stronger in those industries where the intensity of R&D spending is higher. Similarly, Fama and French (1997) found in their research on the industry costs of equity that their factor risk loadings on the B/M ratio vary across industries. This implies that these cross industry variations can have an impact on the return premiums of high B/M (value) firms and hence the value effect.
Based on the discussion above, it is expected that there are value premiums detectable in the different industries and that these value premiums are significantly different form each other.
H1: there is a significant difference in the value premium effect between industries
The value premium effect and risk
literature. The second objective of this paper is therefore to see how the value effect and risk characteristics are related in the industries.
The classical school of finance used to believe that investors are fully rational and higher returns were related to higher risks (Wouters, 2006). Fama and French (1992) expanded the research by introducing a set of micro-economic factors to explain the risk profile of an asset. They state that the value premium can be seen as a reward for the systemic risk that value stocks are facing. Value strategies are considered to be fundamentally riskier than growth strategies, because they are more sensitive for financial distress. In their opinion, the value premium is related to firm leverage, measured by the B/M ratio, which they consider to be a proxy for financial distress. Distressed firms are defined as firms which are less efficiently run, have higher financial leverage and have lower accessibility to external funds. Distressed firms are more sensitive and vulnerable to changes in the economy and are less likely to survive unfavorable economic conditions, making them riskier than other firms. For example, firms with poor past earnings and high financial leverage may have limited accessibility to external funding, leading to cash flow problems. In times when economic conditions are poor, these firms are more likely to get into financial difficulty (Wouters, 2006). Banko, Conover and Jensen (2006) also find supporting results for the rational explanation. They state that the value premium can be seen as a result of investors requiring a higher return for stocks from firms in relatively distressed conditions. Value firms are shown to have higher returns and consequently also higher earnings uncertainty, higher leverage and higher risk of financial distress.
activity. They use four variables, including the B/M, E/P and the C/P ratio and find as well that value strategies do not show underperformance in times of low economic activity, supporting the statement that the value premium effect cannot be explained by risk differences alone. Furthermore, they looked at the variability of their portfolios, but the differences in the standard deviations of the stocks could only explain a small part of the return differences. Capaul, Rowley and Sharpe (1993) found a significant value premium effect in six developed countries. An interesting result of their study is that in most cases, the value stocks had a lower beta than growth stocks, indicating that value stocks were not riskier compared to growth stocks.
Based on the above, it can be stated that the relation between risk and the value premium is not unambiguous. This makes it difficult to define expectations about the risk characteristics of industries and the value premium effect. In line with Banko, Conover and Jensen (2006), who looked at the value premium in U.S industries in relation with risk based pricing decisions of investors, it is expected that there is a greater value premium in industries with greater financial distress. They find a positive relation between risk measures and B/M ratio, stating that value firms and industries are riskier than their growth counterparts and therefore earning higher returns. Therefore it is expected that risk is higher for value firms and industries relative to growth firms and industries and that there is a significant relation between the returns attributed to the B/M ratio and risk measures.
H2: High B/M (value) firms and industries have higher risk measures compared to low (growth) firms and industries
The value premium effect and behavioral explanations
Evidence exists that company earnings are close to a random walk, with earnings growth rates being predictable one to two years into the future. However, the large price to earnings ratio differences between growth and value stocks seems to reflect expectations of investors that past growth differences persist much longer than can be reliable predicted from past data. Accordingly, value stocks generate higher returns compared to growth stocks, because the market slowly realizes that their extrapolations are wrong and earnings growth rates for value stocks are higher than initially expected and vice versa for growth stocks (La Porta et al., 1997). Extrapolating past performance of stocks too far into the future is called the extrapolation bias. When these growth rates are projected to the future, without looking at the lack of persistence of these growth rates, favorable sentiment is created for growth stocks (Chan and Lakonishok, 2004).
Lakonishok, Shleifer and Vishny (2004) explain that higher returns for value stocks can best be explained by the preference of both individual and institutional investors for growth strategies and their dislike for value strategies. The reason that individual investors might have focus on growth strategies is that they extrapolate past growth and base their selection on non-rational basic assumptions like familiarity and media coverage of the stock. They mistake well run companies with good investments, regardless of price (Solt and Statman, 1989). This is supported by brokers, who recommend good companies with stable earnings as being a good investment (Lakonishok, Shleifer and Vishny, 2004).
these investors are inclined to invest in companies that have a high current or expected future level of profit, regardless of their stock price, makes them overvalue growth stocks an undervalue value stocks. Shefrin and Statman (1985) follow the same line of reasoning in their article and also include the phenomenon of regret avoidance as an explanatory factor. They explain regret avoidance as investors resisting the realization of a loss because it proves that their original judgment turned out to be wrong. This feeling of regret may be exaggerated by having to admit the mistake to others, making it a safer option for investors and analysts to follow the main stream. This is in line with another phenomenon called herding. Investors and analyst follow the actions of the main group, which could result in asset prices that deviate from their rational value (Olsen, 1996).
Another important factor that is related to the fact that investors are more driven towards growth stocks, is that they have shorter time horizons for their investments than are required for value strategies to pay off. The pressure of underperforming their peers by investing in “risky” value strategies is seen as being too high. A value strategy that takes 3 to 5 years to pay off, may underperform the market in the meanwhile, making it too risky from a career perspective. Especially because the value stocks are by themselves more difficult to justify to their sponsors, as was mentioned before (Lakonishok, Shleifer and Vishny, 2004).
Also the fact that investors are rewarded on a commission base is considered to be influential on the judgmental bias. Investors have a preference for recommending stocks that are successful and generate trading commissions. Besides, growth stocks are considered to enjoy more media coverage, are easier to bring to the market and therefore generate a higher demand for analysts’ reports compared to value stocks. All these elements are taken into consideration by analysts and investors, influencing their buy and sell decisions. The result of this could be that analysts and investors don’t make the optimum investment decisions, but act more in their own interest. These considerations could result in underpricing of value stocks and overpricing of growth stocks relative to their fundamentals (Chan and Lakonishok, 2004).
values stocks, though misprice the same risk due to personal biases. However, if evidence will be found that value stocks and value industries do not have higher risk compared to growth firms and industries, a rational risk based explanation becomes less likely and the focus on explaining the value premium effect should focus more on the behavioral side.
III. Methodology
Portfolio formation and returns
For dividing the stocks into different industries the NYSE industry classification benchmark (ICB) is used, which comprises the stocks into ten industries. Data of stock exchanges of 27 countries in Europe has been collected. Following Fama and French (1998) three ratios are used to categorize the stocks into growth and value portfolios, the book to market (B/M) earnings to price (E/P) and cash flow to price (C/P) ratio. Firms with ratios falling in the top 30% are considered value stocks for that specific ratio. Firms with the ratios in the low 30% are growth stocks, leaving the remaining 40% categorized as medium stocks (Fama and French, 1998). Portfolios are formed at the end of each calendar year from 1999 till 2009 and returns are calculated from July the following year, in order to make sure that all relevant information is incorporated in the returns. The portfolios are rebalanced each year to reflect changes in the relative ratios (Brouwer, Van der Put and Veld 1996).
Holding discount rates and payout ratios constant, a high C/P ratio (value stock) indicates a low expected growth of cash flows and vice versa for a low C/P ratio (growth stocks) (Brouwer, Van der Put and Veld, 1996).
Fama and French (1992) have been followed in deriving the B/M ratio. The book equity is measured at fiscal year-end in year (t-1). The market equity used for calculating the B/M ratio is measured in December of year (t-1). As a result the return interval runs from July of year t through June of year t+1. Portfolio returns are derived as monthly equal weighted returns for the different firms in the portfolio.
The earnings and cash flows in the E/P and the C/P ratios are for the fiscal year-end in year (t-1). The price for the E/P and C/P ratio is measured in December of year (t-1). Both the accounting data as the total returns are generated by Thomson’s DataStream. The generated stocks prices at the end of each month are all adjusted for stock splits and dividend payments and are converted to continuously compounded returns, for these can easily be compared across assets and are time additive (Fama and French, 1992; Brooks, 2004). The formula to calculate the continuously compounded return is as follows:
ln / (1) Where:
Explanatory variables
To isolate the returns associated with the B/M, E/P and C/P ratios, size and beta are included as control variables (Banko, Conover and Jensen 2006). Size is represented by the market capitalization of stocks, calculated as the price per share times shares outstanding at the end of June of year t (Fama and French, 1992). Betas are the full period betas calculated over the ten year period using the S&P 500 Europe returns as a benchmark.
!"#,"%
Where:
' (
) *&500 .
/0',)1 / 23 ( *& 500 . /) / 4 (
Inter and intra-industry value premium effects
The first objective is to see whether there are value premiums in the ten industries. Annual returns are calculated for value and growth portfolios for all three ratios over the 2000-2010 period. The t-test is used to test whether the difference between the value and growth stocks differs significantly from zero (Brooks, 2004). In order to test whether the value premiums between the different industries are significantly different from each other, two groups are formed including the three industries with respectively the highest and lowest average value premiums. A t-test is performed to test whether these groups are significantly different from each other.
The OLS regression is the following: 5 67 /85 7 9 /8:;<5 7 =*>57 ? 57 @ /8:;<5 7 ∈5 (4) Where: 5 B 3 /8 4 6 , 9 44 /85 4 /8 4 4 /8C;<5 4 /8 4 *>5 4 ( > 4 4 5 4 2 4 4 / *&500 . D /8C;<5 4 /8C;<5 67 /85 7∈E/FGHIJK ∈5
Value premiums and risk
The second objective is to test whether there is a relationship between the value premium effect and risk. If B/M is an effective proxy for financial distress risk, the value premium effect should be present in the majority of our industries (Banko, Conover and Jensen, 2006). Again OLS regressions are used, though this time separate regressions are estimated for each industry. Identical as in the pooled OLS regression, the portfolio returns are the dependent variable, portfolio B/M and industry B/M are the independent variables and market capitalization and beta are included as control variables. Natural logarithms of both B/M and size ratios are used as well as White’s modified standard error estimates in order to deal with the heteroskedasticity problem.
The OLS regression for each industry is the following:
5 67 /85 7 9*>57 = 57 ∈5 (5)
Where:
5 B 3 /8 4
, 9 44
/85 4 /8 4 4
*>5 4 ( > 4 4
5 4 2 4 4 / *&500 . D
∈5
To see how B/M ratios change over time, a general view of the consistency is generated by ranking the ten industries based on their average annual B/M ratio into value, growth and medium industries. Finally the direct relation between risk and the value effect is examined. The industries are differentiated using their relative B/M ratio for each year. Consequently, the ten industries are ranked every year within this industry categorization, individual firms are then again categorized as being value and growth firms making a total of 6 portfolios. To look at the relation between risk and the intra and inter-industries value premiums, earnings uncertainty, beta and financial leverage are used as measures of risk (Banko, Conover and Jensen, 2006). Financial leverage is measured by using the book value of debt at fiscal year-end of t-1 divided by the market value of equity at calendar year-end of year t-1. Earnings uncertainty is measured by the standard deviation of the earnings to price ratio. Finally the betas are included.
IV. Data
Construction of the dataset
The span of this paper covers the period of 2000-2010. The portfolios are formed at the end of each calendar year from 1999 to 2009, and the returns are calculated for the following year, starting in July. The total number of firms included in this research is 7,174 but the type of data provided of each firm can change by year. However, in order to build the different portfolios it is not required that the same firms have data on all ratios. As mentioned, Thomson’s Datastream is used to generate the stock prices at the end of each month. These prices are all adjusted for stock splits and dividend payments.
Descriptive statistics of the independent variables
The prices generated by Datastream are converted to continuously compounded returns for these can easily be compared across assets and they are time additive(Fama and French, 1992; Brooks, 2004). In line with the Fama and French (1998) paper, the portfolios are built in December of each year based on the sorted values of the B/M, E/P and C/P ratios. The value portfolios include the firms whose ratio is among the highest 30% for a given industry. Growth portfolios include the firms with the lowest 30% ratios. All firms are equally weighted in the industry portfolios. The market portfolio includes all the firms. In total, there are 10 industries, covering 10 years times 12 months, generating a total of 1,200 monthly returns for both value and growth portfolios.
mean market cap Beta B/M B/M E/P E/P C/P C/P
Industry sample size in million € Std. Std. Std.
Technology 508 473 0,82 0,48 0,88 0,15 0,66 0,05 0,35 Telecom 49 8216 0,57 0,52 0,47 0,10 0,14 0,12 0,32 Health care 221 1467 0,66 0,56 0,65 0,09 0,22 0,03 0,16 Consumer service 543 997 0,72 0,65 0,95 0,12 0,42 0,09 0,27 Utilities 84 3695 0,70 0,71 0,46 0,10 0,27 0,11 0,21 Industrials 952 588 0,88 0,73 1,06 0,14 0,52 0,11 0,30
Oil & gas 131 4756 0,71 0,78 0,85 0,12 0,19 0,08 0,18
Consumer goods 555 1105 0,69 0,82 0,85 0,13 0,50 0,11 0,40
Financials 775 1687 0,73 0,90 1,00 0,17 0,52 0,11 0,55
Basic Materials 243 1148 0,99 0,90 0,97 0,13 0,27 0,11 0,24
Table I shows the descriptive statistics for the ten European industries based on the NYSE categorization. B/M ratio is book to market equity, where book equity is measured at fiscal year-end at t-1 and market equity is measured in December of year t-1. E/P and C/P represent the earnings to price and cash flow to price ratio, where earnings and cash flow are measured at the fiscal year end of year t-1 and price is measured in December of year t-1. Std. shows the standard deviations of the ratios. Size is measured as price times shares outstanding (market cap) at the end of June of year t, in million euros. Beta is the full period beta calculated as the variance of the stocks and the covariance with the S&P 500 Europe index. The values for sample size, market capitalization, B/M, E/P, C/P ratios and their standard deviations are annual medians.
industry and the consumer goods (including the automotive industry) are in the high end whereas more service based industries like health care and consumer service are in the lower end.
The standard deviation of the B/M ratios shows whether there is variation in the B/M ratios within the industry. It shows that the telecom, utilities and health care industries have the most within-industry consistency compared to the financials and industrial industries, which show the least within-industry consistency. The E/P and C/P descriptives are quite similar to each other, making it more difficult to identify value and growth related industries based on these industries. However, the standard deviation of the two ratios do seem to support the high within-industry variation of the telecom and health care industry, as well as the low within industry consistency of the industrials and especially the financial industry.
Looking at the betas of the industries, it shows that they are in a small range of each other and are all below one, which is lower than expected when looking at the beta. The relative differences between the industries are more in line with the expectations. Basic materials, the industrials and the technology industry have relative high betas, compared to the lower betas of the telecom, health care and utilities industries.
Descriptive statistics of the returns
B/M C/P E/P
value growth value-growth value growth value-growth value growth value-growth
Mean 15,60% 9,64% 5,96% 13,30% 10,12% 3,18% 17,63% 11,24% 6,39%
Std. 19,54% 15,75% 9,14% 15,16% 13,34% 11,38% 18,42% 13,74% 10,42%
min -14,58% -14,13% -1,41% -11,83% -17,95% -5,37% -13,23% -16,51% 2,02%
max 40,98% 30,79% 13,66% 30,76% 36,13% 8,89% 38,17% 28,07% 20,11%
Table II Annual returns for value and growth portfolios
Graph I is the representation of the value premium per year based on the B/M ratio, when averaged over all the ten industries. The average return per year of the S&P 500 Europe is included as a benchmark. It shows that the maximum value premium of 13.66% reported in table II is realized in 2002, the minimum value of -1.41% in 2010. Comparing both lines, it shows that for the years 2001-2002 and 2003-2007 they follow the same direction, but in the other years they move apart, so based on this information it can’t be stated that the value premium is related to the market performance. However, it is interesting to see that the economic downfall in the years 2009-2010 is followed by the decreasing value premium even ending up below zero in the year 2010.
V. Results
Value premiums and industry affiliation
B/M C/P E/P
industry value growth v-g value growth v-g value growth v-g
All industries 15,60 9,64 5,96 13,30 10,12 3,18 17,63 11,24 6,39 (19,57) (15,66) [3,87]*** (15,16) (13,34) [2,60]* (18,42) (13,74) [3,02]** Industrials 17,31 10,11 7,20 15,87 10,71 5,16 18,42 12,19 6,24 (20,20) (15,58) [5,90]*** (16,53) (14,75) [3,59]*** (20,17) (14,69) [3,98]*** Utilities 11,20 7,20 4,00 12,43 7,83 4,60 14,03 7,58 6,59 (16,13) (14,80) [1,61] (11,78) (13,92) [1,54] (13,36) (13,71) [2,69]*** Consumer goods 15,53 9,37 6,16 14,14 8,04 6,10 16,29 9,50 6,79 (16,79) (13,39) [3,28]*** (14,47) (13,62) [4,01]*** (19,23) (12,85) [3,91]*** Financials 14,81 7,95 6,87 10,41 9,74 0,67 15,92 9,33 6,59 (17,74) (16,35) [3,46]*** (14,91) (12,17) [0,60] (19,22) (12,24) [4,01]*** Basic materials 16,90 14,11 2,79 14,37 12,51 1,85 22,96 12,04 10,93 (18,65) (15,59) [0,98] (19,24) (17,75) [0,81] (20,43) (18,64) [4,75]*** Telecom 8,81 5,99 2,83 9,48 5,61 3,86 10,78 7,34 3,44 (23,88) (16,06) [0,61] (22,18) (16,64) [0,88] (24,71) (20,90) [0,96] Consumer service 14,05 6,46 7,59 12,12 6,34 5,78 16,34 7,95 8,39 (18,54) (14,70) [3,28]*** (13,56) (11,61) [3,87]*** (17,28) (12,29) [4,79]*** Health care 14,97 8,40 6,58 6,51 11,19 -4,69 15,51 10,05 5,46 (17,03) (15,46) [3,18]*** (20,07) (13,92) [-1,83]* (14,77) (15,71) [2,08]**
Oil & gas 28,96 17,39 11,57 24,49 17,46 7,03 23,85 21,84 2,01
(23,62) (21,30) [2,57]** (22,07) (16,87) [2,52]** (25,99) (18,18) [0,67]
Technology 13,46 9,43 4,03 9,71 9,37 0,33 16,40 12,62 3,79
(22,80) (13,68) [1,18] (17,61) (16,83) [0,08] (20,40) (17,01) [1,87]*
Table III: Annual industry returns for the value and growth portfolios for: 2000-2010
results show that in all the industries, the value portfolios outperform the growth portfolios. On average with 6.0 % for the M/B ratio, 3,2% for E/P and 6,4% for C/P ratio. This is in line with the findings of Fama and French (1998) who as well find strong evidence of a consistent value premium in international stock returns. They find an average value premium of 7.68% for their B/M portfolios. Also compared to the article of Lakonishok, Shleifer and Vishny (1994), the results are similar. They calculated annual returns for B/M portfolios, finding returns varying from 11% for the growth oriented stock and 17,3% for the value stocks, compared to an average of 9,6% for the growth portfolios and 15,6% for the value portfolios of table III.
Based on the B/M ratio, it shows that 6 of the 10 industries have significant value premiums (most of them with a significance level of 1%) and that there is never a significant growth premium. Similar results are presented in the other ratios, with 5 significant value premiums for the C/P ratio, 7 for the E/P ratio and only one growth premium. So based on these findings it can be stated that there are value premiums in the majority of the industries. In order to see whether the value premiums are significantly different from each other, two groups have been identified. A group including the three industries with the highest average value premium and a group with the lowest three industries. A t-test is performed to compare the means of the two groups showing significant difference on a 5% significance level for the E/P ratio and 1% significance level for the B/M and the C/P ratio. This indicates that the value premiums differ per industry, making it interesting to take into consideration for asset investment decisions.
The standard deviations of the annual returns vary from 13% to 26%, which are lower when compared to the article of Fama and French (1998) which used the same methodology to test the difference in the value premium effect for different countries. Their standard deviations show a range of 13% up to 57% and an average of 30%, which means that the volatility of industry returns seems to be lower than the volatility of the country returns.
All industries Basic materialsConsumer goods Consumer service Financials Health care Industrials Oil & gas Technology Telecom Basic materials 0,60 Consumer goods 0,81 0,71 Consumer service 0,44 0,13 0,69 Financials 0,62 -0,07 0,47 0,72 Health care 0,69 0,23 0,61 0,38 0,37 Industrials 0,82 0,67 0,92 0,44 0,32 0,74
Oil & gas 0,32 0,45 0,07 -0,27 -0,06 0,20 0,08
Technology 0,38 -0,12 0,29 0,14 0,23 0,50 0,54 -0,44
Telecom 0,33 -0,04 -0,09 -0,08 0,47 -0,16 -0,11 0,02 0,01
Utilities 0,21 0,03 -0,06 -0,14 0,18 -0,17 -0,07 0,18 0,00 0,20
Table IV Correlations of the value premium on industry B/M value growth returns: 2000:2010
Inter and intra-industry value premium effects
Table III showed already that there is a persuasive value premium effect in the industries. In order to further test this relationship and to see how the value premium effect is related to firm and the industry level, table V is included. Table V shows the result of a pooled cross sectional time-series regression model that has been used. Five regressions are modeled in order to test the relationship between returns and both portfolio B/M ratios as well as industry B/M ratios. Other variables that have proven to be influential on return, namely size and Beta are included as control variables (Banko, Conover and Jensen, 2006).
Table V Relation between industry affiliation and the value premium effect
Table V shows the regression results of monthly portfolio returns on portfolio B/M, industry B/M and control variables. B/M and industry B/M are the natural logarithms of B/M for portfolio P (value, medium and growth) and the industry B/M. For both measures, Book equity is measured at fiscal year-end in year t-1 and market equity is measured at the year-year-end of year t-1. Size is the natural logarithm of the market capitalization of portfolio P in July of year t. Beta is the full period beta, measured as the variance of the portfolio returns with the covariance of the S&P 500 Europe index. The industry B/M residuals are the residuals of the industry B/M portfolio regressed on the B/M portfolio. The portfolios are reformed annually based on their B/M ratio. Each regression has 3600 observations (10 industries x 10 years x 12 months x 3 portfolios). t-statistics are in parentheses. *, ** and *** indicate t values significant at respectively 10%, 5% and 1% level.
model intercept B/M Industry Size Beta Industry B/M
B/M residual 1 0,0127 0,0056 (15,94)*** (6,42)*** 2 0,0149 0,0174 (18,55)*** (8,19)*** 3 0,0156 0,0027 0,0148 (18,34)*** (2,98)*** (6,61)*** 4 0,0143 0,0030 0,0150 -0,0010 0,0168 (2,27)** (2,28)** (6,64)*** (-1,41) (2,35)** 5 0,0139 0,0055 -0,0013 0,0150 0,0172 (2,19)** (4,56)*** (-1,69)* (2,08)** (6,77)***
both portfolio B/M and industry B/M are regressed simultaneously. Both variables remain significant on a 1% level, although the t-values have dropped slightly. The coefficients themselves drop as well, slightly in the case of the industry B/M variable, but much more for the portfolio B/M variable. The coefficients and significance level remain more or less the same in the fourth model, which includes the two control variables size and beta. In line with other research (Fama and French, 1992), size has a negative coefficient, implying a negative relation between size and return, however not significant. Looking at the overall picture, it can be stated that the value premium effect is related to both firm as well as industry characteristics. The industry B/M coefficient remains more or less the same size in all three regressions, which proves that the industry B/M variable is very robust. The coefficient and the significance level of the portfolio B/M level drop after including our control variables. The coefficients are even cut in halves. So both variables are significant, but based on the results the industry B/M proves to be the better explaining variable, for it has the higher coefficient as well as a better significance level.
By testing the B/M coefficient on both the portfolio as well as the industry level, there is the risk that both variables are highly correlated and say merely the same thing. The correlation between the two variables is 0,44 which is lower when compared to the 0,55 which Banko, Conover and Jensen (2006) find in their study on U.S. data. However, to prevent multicollinearity problems with the coefficients, model (5) is included in order to make the test more robust. Model (5) uses the residuals of the industry B/M instead of the original industry B/M values. By regressing the portfolio B/M ratio as the independent variable against the industry B/M ratio as the dependent variable, the residuals of the regression represents that part of the regression that cannot be explained by the portfolio B/M ratio and is therefore the additional part of the industry B/M ratio. Model (5) shows that the coefficient of the portfolio B/M coefficient goes up and that the industry B/M remains the same. This makes the finding that the B/M coefficient is related to both firms as well as industry characteristics even more robust.
outcomes. So where the results of this study show that inter-industry variation in B/M is more important in explaining returns, the article of Banko, Conover and Jensen (2006) suggest that intra-industry variation in B/M ratios is more important.
Similar to the findings of this study are the outcomes of the article of Kothari, Shanken and Sloan (1995). They looked at the relationship between B/M ratio over the period of 1947-1987 based on both COMPUSTAT and S&P industry data and found a significant positive relation between returns and industry B/M ratio based on COMPUSTAT data, however don’t find significant results when using S&P data. Although Kothari, Shanken and Sloan (1995) can’t give a conclusive explanation for the fact that they do not find the same results when using S&P data, Banko, Conover and Jensen (2006) link this result to the fact that the value premium effect is more an intra than an inter-industry phenomenon. The use of inter-industry data and broader industry classification would diminish the prominence of the value premium effect. As mentioned, the findings of this paper prove otherwise. This paper categorizes the firms into 10 different industries, compared to 45-75 in the paper of Kothari, Shanken and Sloan (1995). Using such a broad industry categorization should result in smaller industry B/M coefficients, which is also not the case. So based on the above mentioned findings, the results of this paper show that even after controlling for other relevant factors like size and beta, both inter and intra-industry variation in B/M are relevant in explaining stock returns and the value premium effect, of which the B/M coefficients on an industry level are most robust.
Consistency of the value premium effect across industries
Now it is known that there is a significant relation between the B/M ratio and the value premium effect on both the intra as well as the inter-industry level, it is interesting to see how this develops when looking at the effect across the sample of industries.
Table VI presents the results of the OLS regressions whithin each of the ten industries. For each indivdual industry, a separate regression is estimated testing the relation between the portfolio return (value, medium and growth) as the dependend variable and the portfolio B/M, market capitalization and beta as explanatory variables.
effect is present on an industry level. The fact that the utilities, basic materials and the oil and gas industry don’t show significant results, indicate that the effect is not universal. This means that the effect is not the same in all industries, these differences make it worthwile to look at the industry affiliation when building asset pricing models.
Table VI The value premium effect across industries
Table VI shows the regression results of monthly portfolio returns on portfolio B/M and control variables by industry. Portfolio B/M is the natural logarithm of B/M for portfolio P (value, medium and growth). For both measures, Book equity is measured at fiscal year-end in year t-1 and market equity is measured at the year-end of year t-1. Size is the natural logarithm of the market capitalization of portfolio P in July of year t. Beta is the full period beta, measured as the variance of the portfolio returns with the covariance of the S&P 500 Europe index. The portfolios are reformed annually based on their B/M ratio. Each regression has 360 observations (10 years x 12 months x 3 portfolios). t -statistics are in parentheses. *, ** and *** indicate t-values significant at respectively 10%, 5% and 1% level.
Industry intercept BE/ME Size Beta
Industrials -0,085884 0,024672 -0,000116 0,200296 (-1,41) (2,30)** -0,02 (2,64)*** Utilities -0,011938 0,005956 -0,001965 0,09013 (-0,25) (1,09) (-0,51) -0,65 Consumer goods -0,069695 0,021624 -0,002072 0,239764 (-1,30) (2,1)** (-0,35) (2,83)*** Financials -0,005067 0,017874 -0,006003 0,138154 (-0,20) (2,54)** (-1,91)* (2,63)*** Basic materials 0,02600 -0,002462 -0,014836 0,16900 (0,89) (-0,44) (-1,91)* (1,83)* Telecom 0,00845 0,007466 -0,005056 0,069119 (0,13) (2,00)** (-1,20) (0,91) Consumer service -0,076502 0,020559 -0,000621 0,205002 (-1,57) (2,76)*** (-0,14) (2,85)*** Health care -0,049945 0,016112 0,005329 0,074346 (-1,48) (2,72)*** (1,59) (1,46)
Oil & gas 0,097061 0,006677 -0,002873 -0,088836
(1,20) (1,43) (-1,02) (-0,80)
Technology -0,106212 0,025534 0,006811 0,132102
(-2,28)** (2,77)*** (0,92) (2,46)**
significant on a 1% level, confirming that there are significant differences between the B/M ratios of the different industries. Furthermore a t-statistic is calculated between the coefficients across industries. The oil and gas, utilities and basic materials industries are included in the low coefficient group, compared to industrial, technology and consumer goods industries in the high efficient group. The t-test showed a significant difference between the two groups at a 1% level. The fact that the B/M coefficient shows significant results across industries and the fact that there is a large amount of variation in these coefficients, means that industry affiliation should be taken into consideration when implementing asset pricing models that include the B/M ratio (Banko, Conover and Jensen, 2006).
Variation in time
Results from U.S. industries show that industry B/M ratios can change in time (Banko, Conover and Jensen (2006). Table VII shows a simple representation of the development of the B/M ratio over time in the different industries. The industries are ranked on a yearly basis. The median, minimum and maximum values are derived as well as the range the highest and lowest rank. A high rank means that the industry is more growth oriented and vice versa for a low rank.
Risk and the value premium effect
The temporal fluctuations of table VII are taken into account when looking at both the inter as the intra-industry value premium effect and the relation between the value premium effect and risk. Based on the yearly B/M ratios (1-10) the ten industries are categorized into value, growth and medium B/M industries. After that, the firms within the industries are categorized into growth and value portfolios. Table VIII shows the results of this two step sorting process, resulting in six portfolios for each variable. Besides the returns of the portfolios, also three risk measures are included. Earnings uncertainty, measured by the standard deviation of the earnings per share divided by the price. Financial leverage, measured by the book value of debt divided by the market value of equity. And finally, the beta of the portfolios, measured as the full period beta. If there should be a relation between the value premium and risk, the value industries should show higher values on the three risk measures to indicate the relation between the value premium and risk at the inter-industry level. To indicate the relation between the value premium and risk at the intra-industry level, the value portfolios inside the industry categories should also show higher values.
The output of table VIII shows that the returns on both the inter as well as on the intra-industry level returns are higher for the value portfolios compared to their growth counterparts, however the results are not significant.
Industry median minimum maximum range
Industrials 6 5 9 4 Utilities 5 2 7 5 Consumer goods 8 4 10 6 Financials 10 8 10 2 Basic Materials 9 2 10 8 Telecom 4 1 6 5 Cons service 5 2 7 5 Health care 2 1 8 7
Oil & gas 4 1 10 9
Technology 2 1 9 8
Table VII Temporal consistency of B/M by industry
The risk measures that are included, show mixed results. The debt to equity ratio used to indicate leverage increases from growth to value industries, as is expected. This is also the case for the earnings uncertainty variable, which also increases from 14,65% for growth industries to 18,65% for value industries. Both variables show highly significant results. The results for the beta variable show that the beta levels are rather low. Even more important, they show higher betas for growth industries compared to value industries. However, the difference between the value and growth industries are not significant. Although these three variables provide mixed results, based on the significant increase in both the leverage and earnings uncerainty ratios from growth industrues (low industry B/M ratios) to value industries
Cl a s s i fi ca ti on
Return Le ve ra ge Be ta
I ndus try Fi rm Indus try Fi rm I ndus try Fi rm Indus try Fi rm Indus try Fi rm
va l ue va l ue 11,44% 16,14% 18,65% 17,08% 2,81 2,48 0,62 0,62 growth 10,53% 21,30% 2,86 0,69 va l ue-growth 5,61% -4,22% -0,38 -0,07 (0,77) (-3,91)*** (-1,27) (-3,16)*** me di um va l ue 9,44% 14,56% 15,89% 18,24% 1,67 1,77 0,64 0,62 growth 9,30% 15,31% 1,66 0,71 va l ue-growth 5,26% 2,92% 0,11 -0,08 (-0,83) (-2,65)** (-0,97) (-0,83) growth va l ue 8,42% 13,51% 14,65% 16,46% 1,55 1,02 0,72 0,69 growth 8,85% 12,92% 1,39 0,85 va l ue-growth 4,66% 3,54% -0,37 -0,17 (0,72) (-3,60)*** (-0,05) (-1,51) va l ue -growth 3,02% 4,00% 1,26 -0,11 (-0,73) (-4,48)*** (-3.63)*** (-0,83)
Table VIII Risk and return characteristics
Ta bl e VIII s hows the re l a ti on be twee n the va l ue pre mi um a nd ri s k. The i ndus tri e s a re cl a s s i fi ed a nnua l l y i nto va l ue , me di um a nd growth i ndus tri e s ba s e d on the ra nk of the i ndus try B/M. Fi rms wi thi n the i ndus tri es a re then cl a s s i fi e d i n va l ue a nd growth portfol i os ba s e d on the ra nk of the fi rm B/M. Re turns a re a ve ra ge a nnua l returns for yea r t. Ea rni ngs unce rta i nty i s the s ta nda rd devi a ti on of e a rni ngs /pri ce ra ti o, where the e a rni ngs a re from the e nd of yea r t, pri ce i s the ma rke t va l ue of e qui ty a t the e nd of yea r t-1. Le ve ra ge i s fi na nci a l l e ve ra ge , defi ne d a s tota l book va l ue of debt a t fi s ca l yea r-end of ye a r t-1 di vi ded by the ma rke t va l ue of e qui ty a t the yea r-end of t-1. Beta i s ca l cul a ted a s the va ri a nce/cova ri a nce of s tocks wi th the S&P Europe i ndex. Indus try s hows the res ul ts of re s pecti vel y the va l ue a nd growth i ndus tri es . Fi rm s hows the res ul ts of the va l ue a nd growth portfol i o i ns i de the i ndus try ca tegori za ti on. t -s ta ti s ti cs whe ther the va l ue a nd growth portfol i os a re s i gni fi ca nt di ffe re nt form e a ch othe r a re i n pa renthes es . *, ** a nd *** i ndi ca te t va l ue s s i gni fi ca nt a t re s pecti vel y 10%, 5% a nd 1% l evel .
(high industry B/M ratios) one could say there that value industries show higher risk levels than growth industries, which supports the view that the significant industry B/M variable from table V reflects a premium that investors require as a compensation for higher risk.
Looking at the relation between the risk measures and the value premium within the industry classification, this paper doesn’t find results in this direction. As stated above, the returns for value firms are higher compared to their growth counterparts, and also the value premium increases slightly from 4,66% to 5,61% though these findings are not explained by the results of the three risk measures. The beta coefficients are higher for growth firms in all three the cases and also the leverage ratio of the firms, show higher ratios for growth firms than for value firms in both value and growth categorized industries. The earnings uncertainty ratio shows significant resuls that inside the growth and medium ranked industries the uncerainty ratio increased from growth to value firms, however Inside the value ranked industries this relation turns the other way around. Therefore based on the three risk measures within the industries, it cannot be stated that value firms whithin industries are riskier than growth firms. The proposition that value firms, relative to growth firms, have higher required returns bacause of their higher risk is not supported by our findings. The significant coefficient on firm level B/M ratio in table V can therefore not be explained by risk, consequently another explanation has to be taken into account.
Research limitations and ideas for future research
Table VIII shows that two of the three risk measures indicate higher risk measures for value industries yet shows no relation between risk and the value premium effect on a firm level. Therefore it may be worthwhile to look at the alternative explanation, including more behavioral based arguments for the value premiums. It could well be that judgmental errors and behavioral biases of investors do a better job explaining the higher returns of the value portfolios. People can make judgemental errors and extrapolate past growth rates of low B/M (growth) stocks to persist in the future, or can identify well run firms with good investments, or simply invest in stocks with high media and analyst coverage (Lakonishok, Shleifer and Vishny, 1994).
Because we do find evidence of industry affiliation and value premiums and it might be interesting to look at the relation between the value premium and these behavioral explanations. There was no IBES Datastream license for analyst coverage, making it hard to test the relation between analyst coverage and the value premium in this paper. Furthermore, it would also be interesting to use the methodology of La Porta et al. (1997) on an industry level. By looking at the returns around the earnings anouncements for value and growth portfolios, they found that investors incorrectly focus on past growth as a basis for growth forecasts. When the real company facts came out on the anouncement days, investors tended to be disappointed in the real performance of growth stocks and vice versa for value stocks. This evidence supports the argument that expectational errors are at least part of the reason for the superior return of value stocks (Chan and Lakonishok, 2004). Copying this methodology on industry level, might provide additional evidence of the expetetional explanation of the value premium effect.
VI. Conclusion
This paper examined the role that industry affiliation plays in the value premium effect and looked whether risk is a good explainatory factor for these value premiums. This paper analysed 7,174 firms in 27 European countries. The firms are divided into ten industries, categorized by the NYSE industry index. The book to market, earnings to price and cash flow to price ratios are used for constructing the value and growth portfolios. The results show that value portfolios tend to have higher returns than growth portfolios in all ten industries. The difference between the average returns of high B/M (value) and low B/M (growth) portfolios is 5.96% per year (t=3.87). All ten industries showed a value premium, of which six of the ten were significant on at least a 5% significance level and none of the industries showed a significant growth premium. There are similar results when the portfolios are sorted on earnings to price and cash flow to price ratios.
To test the relation between the value premium and industry affiliation, pooled OLS regressions are used. After controling for size and beta, positive and significant results have been found, indicating that both inter and intra-industry variation in B/M are relevant in explaining stock returns. Consequently, the value premium effect is present at both industry and firm levels. However, contrary to the U.S. data the results of this paper finds that the inter-industry relation shows a higher and more consistent coefficient and a higher significance level, indicating that the inter-industry variation in B/M ratios is the most important variable of the two and that much of the value premium effect can be attributed to B/M differences on industry level.
Further investigation on both intra, inter-industry value premiums and risk, showed that value industries have higher value premiums than growth industries. Moreover, it shows that inside these value and growth industries value firms earned higher returns compared to their growth counterparts. Testing the consitency of the value premium effect in industries over time showed that many of the industries have moved along the value/growth spectrum during the ten years, implicating that it is important to take these fluctuations into account and reclassify the industries periodically.
References
Banko, John C., Mitchell C. Conover, and Gerald R. Jensen, 2006, The relationship between the value effect and industry affiliation, Journal of Business 79, 2595-2616.
Banz, Rolf W., 1981, The relationship between returns and market value of common stocks, Journal of Financial Economics 9, 3-18.
Barberis, Nicholas, Andrei Shleifer, and Robert W. Vishny, 1998, A model of investor sentiment, Journal of Finance 49, 307-343.
Basu, S., 1977, Investment performance of common Stocks in relation to their price- earnings ratios: A test of the efficient market hypothesis, Journal of Finance,
32, 663-682.
Brooks, Chris, 2008, Introductory Econometrics for Finance, Second Edition (Cambridge University Press, Cambridge).
Brouwer, Iwan, Jeroen van der Put, and Chris Veld, 1996, Contrarian investment strategies in a European context, working paper, Tilburg University.
Capaul, Carlo, Ian Rowly, and William F. Sharpe, 1993, International value and growth stock returns, Financial Analysts Journal, pp. 27-36
Chan, Louis K. C., Yasushi Hamao, and Josef Lakonishok., 1991, Fundamentals and stock returns in japan, Journal of Finance 46, 1739-1764.
Chan, Louis K. C., Josef Lakonishok and Theodore Sougiannis, 2001, The stock market valuation of research and development expenditures, Journal of Finance 56, 2431-2458.
Daniel, Kent, and Sheridan Titman, 1997, Evidence on the characteristics of cross sectional variation in stock returns, Journal of Finance 52, 1-33.
De Bondt, Werner F.M., and Richard H Thaler, 1987, Further evidence on investor overreaction and stock market seasonality, Journal of Finance 42, 557-581.
Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns, Journal of Finance 47, 427-465.
Fama, Eugene F., and Kenneth R. French, 1997, Industry costs of equity, Journal of Financial Economics 43, 153-193.
Fama, Eugene F., and Kenneth R. French, 1998, Value versus growth: The international evidence, Journal of Finance 53, 1975-1999.
Foster, Douglas F., Tom Smith, and Robert E. Whaley, 1997, Assessing goodness-of-fit of asset pricing models: The distribution of the maximal R2, Journal of Finance 52, 591–607.
Graham, Benjamin, and David Dodd, 1934. Security Analysis, (McGraw Hill, New York).
Hirschleifer, David, 2001, Investor psychology and asset pricing, Journal of Finance 56, 1533-1597.
Kothari, S.P., Jay Shanken and Richard G. Sloan, 1995, Another look at the cross-section of expected stock returns, Journal of Finance 50, 185-224.
Lakonishok, Josef, Andrei Shleifer, and Robert W. Vishny, 1994, Contrarian investment, extrapolation, and risk, Journal of Finance 49, 1541-1578.
Moskowitz, Tobias J., and Mark Grinblatt, 1999, Do industries explain momentum? Journal of Finance 54, 1249-1290.
Olsen, Robert A., 1996. Implications of herding behavior for earnings estimation, risk estimation and stock returns, Financial Analyst Journal 52, 37-41.
Phalippou, Ludovic, 2008, Where is the value premium? Financial Analysts Journal 64, 41-47.
Shefrin, Hersh and Meir Statman, 1985, The disposition to sell winners too early and ride losers too long: Theory and evidence, Journal of Finance 40, 777-790.
Shleifer, Andrei and Robert W. Vishny, 1997, The limits of arbitrage, Journal of Finance 52, 35-55.
Shleifer, Andrei, 2000, Inefficient markets: An introduction to behavioral finance (Oxford university Press, New York)
Solt, Michael E., and Meir Statman, 1989, Good companies, bad stocks. Journal of Portfolio Management 15, 39-44.
Appendices
Appendix B NYSE Industry classification benchmark
Industry Supersector Sector Subsector
Basic Materials Basic Resources Forestry & Paper Forestry
Paper
Industrial Metals Aluminum
Iron & Steel
Nonferrous Metals
Mining Coal
Diamonds & Gemstones
General Mining
Gold Mining
Platinum & Precious Metals
Chemicals Chemicals Commodity Chemicals
Specialty Chemicals
Consumer Goods Automobiles & Parts Automobiles & Parts Auto Parts
Automobiles
Tires
Food & Beverage Beverages Brewers
Distillers & Vintners
Soft Drinks
Food Producers Farming & Fishing
Food Products
Personal & Household Goods
Household Goods Durable Household Products
Furnishings
Home Construction
Nondurable Household
Products
Leisure Goods Consumer Electronics
Recreational Products
Toys
Personal Goods Clothing & Accessories
Footwear
Personal Products
Tobacco Tobacco
Consumer Services Media Media Broadcasting & Entertainment
Media Agencies
Publishing
Retail Food & Drug Retailers Drug Retailers
Food Retailers & Wholesalers
General Retailers Apparel Retailers
Broadline Retailers
Home Improvement Retailers
Specialized Consumer Services
Specialty Retailers
Travel & Leisure Travel & Leisure Airlines
Gambling
Recreational Services
Restaurants & Bars
Travel & Tourism
Financials Banks Banks Banks
Financial Services
Equity Investment
Instruments Equity Investment Instruments
General Financial Asset Managers
Consumer Finance Investment Services Mortgage Finance Specialty Finance Nonequity Investment Instruments Nonequity Investment Instruments
Real Estate Diversified Real Estate
Investment Trusts
Hotel & Lodging Real Estate
Investment Trusts
Industrial & Office Real Estate
Investment Trusts
Mortgage Real Estate
Investment Trusts
Real Estate Holding &
Development
Real Estate Services
Residential Real Estate
Investment Trusts
Retail Real Estate Investment
Trusts
Specialty Real Estate
Investment Trusts
Insurance Life Insurance Life Insurance
Nonlife Insurance Full Line Insurance
Insurance Brokers
Property & Casualty Insurance
Reinsurance
Health Care Health Care Health Care Equipment & Services
Health Care Providers
Medical Equipment Medical Supplies Pharmaceuticals & Biotechnology Biotechnology Pharmaceuticals
Industrials Construction & Materials Construction & Materials
Building Materials & Fixtures
Heavy Construction
Industrial Goods & Services
Aerospace & Defense Aerospace
Defense
Electronic & Electrical
Equipment
Electrical Components & Equipment
Electronic Equipment
Diversified Industrials Industrial Engineering Commerical Vehicles & Trucks
Industrial Machinery
Industrial Transportation Delivery Services
Marine Transportation
Railroads
Transportation Services
Trucking
Support Services Business Support Services
Business Training &
Employment Agencies
Financial Administration
Industrial Suppliers
Waste & Disposal Services
Oil & Gas Oil & Gas Alternative Energy Alternative Fuels
Renewable Energy Equipment
Oil & Gas Producers Exploration & Production
Integrated Oil & Gas
Oil Equipment, Services
& Distribution
Oil Equipment & Services
Pipelines
Technology Technology Software & Computer Services
Computer Services
Internet
Software
Technology Hardware &
Equipment
Computer Hardware
Electronic Office Equipment
Semiconductors
Telecommunications
Equipment
Telecommunications Telecommunications Fixed Line
Telecommunications
Fixed Line Telecommunications
Mobile
Telecommunications
Mobile Telecommunications
Utilities Utilities Electricity Alternative Electricity
Conventional Electricity
Gas, Water &
Multiutilities
Gas Distribution
Multiutilities
Water
