An Analysis of the Value Premium:

(1)

An Analysis of the Value Premium:

Evidence from European stock markets

ABSTRACT

This study examines the value premium in the major European stock markets in the period from 1995 to 2018. Using the price-to-earnings ratio (P/E) and ratio of market-to-book value

of equity (M/B) as value-growth indicators, value-weighted and equally-weighted portfolios are constructed for Belgium, France, Germany, Italy, the Netherlands, Sweden, Switzerland and the UK. Over the full sample period, no persistent value premium is found based on both indicators. Moreover, the capital asset pricing model (CAPM), Carhart four factor model and Fama French five factor model do not explain the portfolio returns. Furthermore, results are

also not in line with the overreaction hypothesis as driver of the returns.

Key words: Value Premium

Word count: 9415 (excluding appendices)

Student name: Lars van Grol Student number: s2541548

Study program: MSc Finance, University of Groningen Contact: l.t.van.grol@student.rug.nl

(2)

2

1. Introduction

A significant portion of the academic literature within the field of finance and asset pricing has been devoted to the value premium anomaly. The value premium can be defined as the (positive) spread between (risk-adjusted) returns of value stocks and growth stocks. Value stocks trade at a relatively low price compared to their fundamentals, while growth stocks, conversely, trade at high prices relative to their fundamentals. Generally, they are classified according to certain accounting ratios such as the earnings-to-price ratio (E/P), the cash flow-to-price ratio (CF/P), the ratio of book value of equity to market value of equity (B/M), or the inverses of these ratios (Fama and French, 1998). Investing in value stocks as a means to achieve superior returns was first introduced by Graham and Dodd (1934). In essence, they argued that these stocks were undervalued and they would revert to their intrinsic value over time (i.e. mean reversion). Later, in the second half of the 20th century, various scholars documented the outperformance of value stocks relative to growth stocks in the U.S. stock market (see e.g., Nicholson, 1960; Basu, 1977, 1981; Lakonishok et al., 1994). Subsequently, international evidence of a persistent value premium was also found in global stock markets (see e.g., Capaul et al., 1993; Fama and French, 1998; Athanassakos, 2009).

One might argue that the outperformance of value stocks relative to growth stocks is not in line with the (strong form) efficient market hypothesis (EMH), as the hypothesis states that asset prices reflect all available information fully and instantaneously (Fama, 1970, 1991) and therefore it would not be feasible to beat the market (i.e. yield superior risk-adjusted returns) persistently. Therefore, the value premium is often classified as a market anomaly. The value premium gained considerable attention when Fama and French (1992) argued that a value factor, proxied by the book-to-market value of equity, combined with a size factor captured the cross-sectional variation in average U.S. stock returns. This led to the introduction of the Fama-French three-factor model (Fama and French, 1993), which was the beginning of the factor models in the asset pricing literature with well-known extensions of the original three-factor model such as the Carhart four-three-factor model (Carhart, 1997), the Fama- French five-factor model (Fama and French, 2015) and the q-five-factor model (Hou et al., 2015). The emergence of the great number of factors proposed by scholars since then, sometimes referred to as the “factor zoo” (Cochrane, 2011), signals the enormous interest in factor models and asset pricing theory. Today, factor models play an important role in portfolio management of (large) funds and the value factor is still employed in most commonly used models.

(3)

3 et al., 1994). These behavioural-based explanations are based on the notion that irrationality of market participant causes miss-pricings in the stock market. Value stocks are undervalued and growth stocks are overvalued, arguably due to overreaction towards past events (De Bondt and Thaler (1985) and extrapolation of past earnings (Lakonishok et al., 1994). When stock prices eventually rebound, value stocks yield higher (risk-adjusted) returns than growth stocks.

Hence, besides establishing whether the premium exists at all, the question that remains is whether the premium is actually a violation of the EMH caused by market irrationality or whether it is a compensation for bearing more systematic risk (i.e. risk premium). It is the discrepancy in interpreting the value premium, besides its persistence, that has given rise to interest in the phenomenon and its underlying drivers.

1.1. Research questions

A large number of the prior literature regarding the value premium was published decades ago. As noted above, in the beginning the focus was primarily on U.S. stock market data. Later, also global stock market data was examined. This thesis will test the value premium in a comprehensive European stock market setting, including more recent stock return data. The sample will contain data from the following major European stock markets: UK, Germany, the Netherlands, France, Italy, Sweden, Switzerland and Belgium. The sample will be divided into subsamples on a country level. Therefore, I construct the first research question:

1) Do value stocks outperform growth stocks in the major European stock markets?

The sample is based on stock market data from 1995 to 2018, including periods of economic expansions and contractions. An examination of the value premium over time could offer insights of how it is affected by general economic conditions and the market environment. Therefore, I construct the second research question:

2) How did the value premium develop through time?

Although the existing literature has touched upon multiple dimensions of reasoning, a comprehensive and economically satisfying explanation about the underlying drivers of the value premium has yet to be found, if the premium even exists at al. Therefore, I construct the third and fourth research questions:

(4)

4 So, the aim of this thesis is not only to establish whether the value premium evidently exists in the data set, but it also focuses on the underlying determinants of the value premium and its development through time. Results of this thesis show that there is no persistent value premium in the major Europeans stock markets for the period 1995 to 2018. Portfolios have been constructed according to the P/E ratios and the M/B ratios and have both been value-weighted and equally-value-weighted. In some years a positive premium is documented over the full sample; however, this is not persistent in the full sample period in any of the eight countries. The years that show a positive premium in the sample are at the end of (severe) bear markets or the subsequent year. The (growth) portfolio returns cannot be explained by the CAPM (Sharpe, 1964; Lintner, 1965), Carhart four factor model (Carhart, 1997) or the Fama French five factor model (Fama and French, 2015), as the regressions result in

significant non-zero alphas for all three models. Returns must be driven by risk factor which are not specified in these models. Moreover, the returns are also not driven by the

overreaction hypothesis, as De Bondt and Thaler (1985, 1987) argue. Hence, the driver of the premium is left unexplained.

The remainder of this thesis is organised as follows. In section 2 the academical literature regarding the value premium is reviewed and discussed. Section 3 describes the data and the methodology of the study. Section 4 describes the results. Section 5 provides the conclusion and recommendations for future research.

2. Literature Review

In this section I will provide a comprehensive overview of the academical literature regarding the value premium. The literature reviewed and its discussion will be the foundation for the rest of this thesis. This chapter starts with a review of the (multi-)factor model literature and the importance of the value factor and the premium in multi-factor models. The second section focuses on defining value and growth stocks and their characteristics. Thereafter, I review the academic empirical evidence - mostly supporting the existence - of the value premium. Finally, I will discuss the risk-based explanations and behavioural explanations for the premium provided by prior research.

2.1. (Multi-) factor models

(5)

5 French, 1993; Carhart, 1997). However, factors can also be based on macroeconomic

variables. Recent developments in the literature have led to the emergence of competing factors, the “factor zoo” (Cochrane, 2011), as attempts to enhance the explanatory power of standard asset pricing models or to enhance unbiased fund performance evaluation (Mateus et al., 2019).

One of the most well-known models in the asset pricing literature is the capital asset pricing model (CAPM) (Sharpe, 1964; Lintner, 1965). In (the basic version of) the CAPM there is a positive relationship between average stock returns and market risk, represented by the market beta β. Although this one-factor model has received considerable support in other literature (see e.g., Fama and Macbeth, 1973) and is still popular and commonly used primarily due to its simplicity, it has also been subject to empirical contradictions. Resulting anomalies, such as the value premium, has given rise to extensions of the CAPM.

Fama and French (1992) argue that the market beta itself in the basic CAPM does not explain the cross-section of average U.S. common stock returns. Based on empirical inconsistencies related to firm size (Banz, 1981) and book-to-market equity (Stattman, 1980; Rosenberg et al., 1985; Chan et al., 1991), Fama and French (1992) suggested that these two variables combined did significantly improve the explanation of the cross-section of average U.S. common stock returns for their sample period (1963-1990). This led to the introduction of the Fama-French three-factor model (Fama and French, 1993), a multi-factor model that

expanded the classical CAPM with a size factor (small minus big, SMB) and a value factor (high minus low, HML). The former captures the historical outperformance of small firm stocks relative to large firm stocks (where market equity proxies for firm size), while the latter captures the historical outperformance of stocks with high book-to-market ratios relative to stocks with low book-to-market ratios (that is, the ratio of a firm's book equity to its market equity). This means that these factors should capture additional elements of systematic risk that were not captured by the classical CAPM.

(6)

6 alternative multi-factor model that takes a different perspective than the Fama-French three-factor model and the Carhart four-three-factor model. It has also dropped the value three-factor, but focuses more on the economic interpretation of the factors. They argue that their model, that includes a market factor, a size factor, an investment factor and a profitability factor,

outperforms the prior models in capturing many asset-pricing anomalies.

So, the value factor had a prominent role in the influential multi-factor models proposed by scholars in the previous decades (Fama and French, 1993; Carhart, 1997), whereas recent contributions to the existing literature suggest that the value premium is captured by other factors (Fama and French, 2015; Hou et al., 2015).

2.2. Value versus growth stocks

Graham and Dodd (1934) are considered the founders of value investing. Their early ideas regarding the definitions of value and growth stocks have been the foundation in classifying these stocks in later literature. According to Graham and Dodd (1934), value stocks (i.e. stocks that qualify for their value investing strategy) are stocks that rank relatively low on certain multiples such as price-to-earnings (P/E), price-to-book (P/B) and price-to-cash flow (P/CF). Thus, the stock price at which those securities trade can be considered low relative to their fundamentals. On the other hand, stocks that trade at high prices relative to their

fundamentals are considered growth stocks. This has been the foundation for defining value and growth stocks in other literature (see e.g., Lakonishok et al., 1994; Fama and French, 1998; Chan and Lakonishok, 2004). Another multiple that is widely used to classify value and growth stocks is the ratio of book value of equity to its market value of equity (that is, book-to-market, B/M). In essence, this ratio is equivalent to the price-to-book ratio.

Moreover, it is important to note that also the inverses of these ratios (e.g. earnings-to-price instead of price-to-earnings) are often used in the academic literature.

Moreover, Chan and Lakonishok (2004) suggest that: “value stocks tend to have a past history of poor performance (relative to growth stocks), with respect to growth in earnings, cash flows and sales”. Poor performance could also be seen in the light of maturity, another factor driving value stocks (Graham and Dodd, 1934). When a firm reaches the stage of maturity in its life cycle, growth prospects stabilize and investment opportunities relative to competitors diminish. Fama and French (1995, 1998) argue that stocks with a low stock price relative to its book value, thus value stocks, signal persistently low earnings relative to their book equity and link that to relative (financial) distress. Moreover, growth stocks may be valued based on the expected growth opportunities perceived by investors. However,

(7)

7

2.3. Value premium

In the second half of the last century, various scholars started to examine the performance of value and growth stocks. Initially, value and growth stocks were primarily classified by the price-earnings multiple. Nicholson (1960) and Basu (1977, 1983) were among the first to link the price-to-earnings ratio to stock performance in the U.S. stock market. Basu (1977)

documented that low P/E portfolios earned, on average, higher risk-adjusted returns than high P/E portfolios. These results were strengthened by Basu (1983), who found similar significant evidence for the premium while also controlling for firm size. Moreover, Jaffe et al. (1989) examined the anomaly within a considerable larger sample period and confirmed the positive E/P effect for U.S. stock returns.

During the same period Rosenberg et al. (1985) fuelled interest in another value-growth indicator, as they documented higher returns, on average, for firms with high ratios of book equity to market equity relative to firms with low ratios. Fama and French (1992) used B/M to explain the cross-section of stock returns and documented a significant positive

relationship between B/M and average U.S. stock returns from 1963 to 1990. Subsequently, this led to the construction of the value factor (HML) in their proposed three-factor model (Fama and French, 1993). Moreover, Lakonishok et al. (1994) presented the similar results with U.S. common stock data. Their value portfolios, formed with stocks scoring high on B/M, C/P and E/P, outperformed growth portfolios (with stocks scoring low on B/M, C/P and E/P).

(8)

8 Hypothesis 1: Value stocks outperform growth stocks in the major European stock markets in the period from 1995 to 2018.

2.4. Risk-based explanations

Besides the examination of the value premium, various scholars have also attempted to explain the observed phenomenon. The literature can be roughly be divided into two types of explanations: risk-based and behavioural-based. The fundamental idea underlying risk-based explanations is that value stocks earn higher returns as they are fundamentally riskier relative to growth stocks. Fama and French (1992, 1993, 1998) and Liew and Vassalou (2000) adhere to this argumentation and suggest that the higher level of perceived risk for value stocks is based on some sort of relative (financial) distress. They argue that value stocks have poor prospects and persistently low earnings, which increases the risk for investors. The premium arises because value stocks are compensated with higher returns for bearing this additional risk. In line with this reasoning are the results of Chen and Zhang (1998), who examined the risk characteristics of value and growth stocks in multiple international stock markets and found that value stocks are indeed under distress, with high levels of leverage and uncertain future earnings. Moreover, Fama and French (1995) study whether stock price behaviour is consistent with the behaviour of earnings. Results show that high B/M stocks are persistently in financial distress as past earnings are depressed and future earnings are uncertain.

Therefore, they are arguably riskier. Conversely, growth stocks with low B/M ratios exhibit persistent high earnings and are less risky.

Black and McMillian (2006) examined the value premium in relationship to the effects of changing economic conditions (shocks). They find that value portfolios have larger risk premiums and greater asymmetry in volatility relative to growth portfolios (i.e. negative shocks increase volatility more than positive shocks). This suggests that the value portfolios are inherently more riskier than the growth portfolios, thereby supporting the rational risk interpretation of the value premium.

Hypothesis 2: The outperformance of value stocks relative to growth stocks can be attributed to a compensation for bearing additional systematic risk.

2.5. Behavioural-based explanations

On the other hand of the value premium spectrum are the scholars that contribute the outperformance of value stocks compared to growth stocks to irrational investor behaviour instead of fundamental risk characteristics. The basic idea underlying this explanation is that growth stocks are overpriced and value stocks are under-priced by irrational expectations of market participants. In line with the notion of irrational investor expectations is the

(9)

9 on a sample that consisted of monthly return data from the New York Stock Exchange

(NYSE) from 1926 to 1982, De Bondt and Thaler (1985) formed loser portfolios with the most extreme “losing” stocks and formed winner portfolios that contained the most extreme “winning” stocks. Loser portfolios significantly outperformed winner portfolios. The CAPM-betas of the extreme portfolios were examined; however, these were surprisingly larger for winner portfolios than for loser portfolios. In other words, if the CAPM-beta correctly captures fundamental risk, loser portfolios were significantly less risky. Therefore, De Bondt and Thaler (1985) argue that the outperformance of the loser portfolios is consistent with the overreaction hypothesis, implying that investors overweight recent information and

unexpected new events, while they underweight fundamental prior information. The “winner-loser” effect was further examined by De Bondt and Thaler (1987) and they found additional evidence that the outperformance of loser portfolios could not be attributed to risk captured by the CAPM-betas. As their results do show predicted earnings reversal patterns, they support the overreaction hypothesis. The same pattern is also observed for portfolios ranked on market value compared to book value.

Lakonishok et al. (1994) examined the outperformance of value strategies and found that their results were in line with the contrarian investment model (i.e. betting against naive investors). The future growth rates of growth stocks (named “glamour” stocks in their

research) were consistently overestimated by investors relative to value stocks. This is in line with Bauman and Miller (1997), who suggest that the value premium is a systematic

overestimation of earnings per share (EPS) for stocks with high past EPS growth rates and high prices relative to EPS and cash flow per share (thus growth stocks). Consequently, growth stocks earn lower average stock returns than value stocks.

Hypothesis 3: The outperformance of value stocks relative to growth stocks can be attributed to irrational investor behaviour.

Thus, it can be concluded that the outperformance of value stocks relative to growth stocks seems persistent in international stock markets, based on various value-growth indicators. The question that remains is what the underlying driver of this premium is. Is the higher average return a compensation for bearing more systematic risk, as predicted by classical asset pricing theories? Or is it a direct result of irrationality of investors that causes mispricing of value and growth stocks? Or can the premium be contributed to both?

3. Data & Methodology

(10)

10 growth stocks. Moreover, I will discuss the method used for portfolio construction and

measuring portfolio returns. Finally, the last two chapters outline the empirical methodology used to test the underlying determinants of the results.

3.1. Data

The sample consists of stock data from the following eight major European stock markets: Belgium, France, Germany, Italy, the Netherlands, Sweden, Switzerland and the UK. The data is collected from Thomas Reuters Datastream (TRD) for the period from 1995 to 2018. Data from before 1995 is limited in TRD; therefore, the beginning of this year will be the starting point. The sample includes data from stocks that have ceased to exist, so that the sample is free of survivorship bias. For all countries, all the domestic stock exchanges are included; however, firms with no data during the entire sample period are excluded from the sample, just as non-equity listings or equity listings which are not major listings. Similar to Fama and French (1992), I exclude firms from the financial service sector as their

characteristics (e.g. high levels of leverage) differ significantly from non-financial service firms. In line with the screening method proposed by Schmidt et al. (2019), I adjusted data from firms that were listed in the sample with their domestic currency before the introduction of the Euro (i.e. stocks that were included for the first year(s), but ceased to exist before the introduction of the Euro), according to the fixed exchange rate of the domestic currency relative to the Euro. Firms that were listed with other than domestic currencies were excluded from the sample.

From TRD I collected the monthly total return index (RI) for all stocks in the sample. This measure accounts for the theoretical growth in stock value over a certain holding period and reinvestments of dividends in the same stock (at the closing price on the ex-dividend date). Monthly individual stocks returns were derived from this return index based on the following equation:

𝑅_𝑖,𝑡 = (𝑅𝐼_𝑖,𝑡− 𝑅𝐼_{𝑖,𝑡−1}) 𝑅𝐼⁄ _{𝑖,𝑡−1} (1)

Where 𝑅_𝑖,𝑡 stands for the monthly stock return of stock i at month t and 𝑅𝐼_𝑖,𝑡 for the total return index of stock i at month t. Similar to Schmidt et al. (2019), monthly stock returns greater than 300% or smaller than -55% are outliers and treated as missing. This should diminish the potential influence of (extreme) outliers. Moreover, the market values for all stocks were collected from TRD as well. This is simply the share price multiplied by the number of ordinary shares in issue at any specific moment. I also collected the price-to-earnings ratio (P/E):

(11)

11 Where 𝑃 𝐸⁄ _𝑖,𝑡 stands for the price-to-earnings ratio of stock i at time t, 𝑃_𝑙̇,𝑡 for the price of stock i at time t and 𝐸𝑃𝑆_𝑖,𝑡 for the most recent annualised rate of earnings per share of stock i at time t (the annualised rate may reflect the last financial year or it may be derived from an aggregation of interim period earnings). Finally, I collected the ratio of market-to-book value of equity (M/B) for all stocks from TRD:

𝑀 𝐵⁄ _𝑖,𝑡 = 𝑀𝑉_𝑖,𝑡⁄𝐵𝐸_𝑖,𝑡 (3)

Where 𝑀 𝐵⁄ _𝑖,𝑡 stands for the ratio of market value of equity to book value of equity of stock i at time t, 𝑀𝑉_𝑖,𝑡 for the market value of equity of stock i at time t and 𝐵𝐸_𝑖,𝑡 for the book value of equity of stock i at time t.

3.2. Portfolio construction

Before the portfolios are formed, the stocks need to be classified and sorted according to their value-growth indicators. As discussed before, stocks can be classified as either value or growth stocks based on various accounting ratios. Ratios regarding earnings-to-price, book-to-market equity (similar to book-to-price), cash flow-to-price and dividend yield have been proposed by the prior academic literature (Fama and French, 1998). Most commonly used are the earnings-to-price ratio and the ratio of book to market equity, or the inverses of these ratios (see e.g., Capaul et al., 1993; Lakonishok et al., 1994; Bauman and Miller, 1997). Therefore, in this research, value and growth stocks will be separated based on how they rank on P/E and M/B. Using two separate multiples and rankings instead of one ensures that the research has a more comprehensive foundation and therefore it strengthens the results. Stocks with negative ratios will be excluded from the sample, due to the difficulty of interpreting them. Besides, Basu (1977) argues that their exclusion has no significant impact on the results.

(12)

12 ratio of its market value of equity relative to the total market capitalization for that country. For the latter it means that all the individual stocks within the portfolios have the same weight. Fama and French (1993) argue that the value-weighted approach corresponds to realistic investment opportunities and should therefore be preferred; however, the equally-weighted approach might produce different insights and increases robustness. Therefore, this study includes both.

3.3. Regression analysis

After the returns of the value and growth portfolios have been collected, I control them for systematic risk. Models that are frequently used for regression analyses in the prior academic literature are the one-factor CAPM and multi-factor models such as the Fama-French three-factor model (Fama and French, 1993), Carhart’s four-three-factor model (Carhart, 1997) and the Fama-French five-factor model (Fama and French, 2015). First of all, the excess returns of the value and growth portfolios will be regressed on the market premium, according to the ex post variant of the CAPM proposed by Jensen (1968):

𝑅_𝑖,𝑡 − 𝑅𝐹_𝑡= 𝑎_𝑖 + 𝑏_𝑖(𝑅𝑀_𝑡− 𝑅𝐹_𝑡) + 𝑒_𝑖,𝑡 (4) Where 𝑅_𝑖,𝑡 – 𝑅𝐹_𝑡 is the excess return of the portfolio, 𝑎_𝑖 the alpha of the portfolio, 𝑏_𝑖 the

market beta, 𝑅𝑀_𝑡− 𝑅𝐹_𝑡 the market factor measured as the excess return of the market portfolio and 𝑒_𝑖,𝑡 the residual error of the portfolio. However, as various scholars argue that the market beta in the CAPM on itself does not capture all the systematic risk of securities (see section 2.1), a multi-factor model for the regression analysis is preferred. Therefore, in this thesis, I also employ the Carhart four-factor model to test whether this model can explain the cross-section of returns in this sample. This approach is extensively used in the literature and industry for measuring the performance of portfolios. In contrast to the different Fama and French models it contains a momentum factor that captures different cycles and therefore may control for some performance attributes that could be the explanation for any

outperformance. The model tests the excess returns of the portfolios against four factors: a market factor, a size factor, a value factor and a momentum factor:

𝑅_𝑖,𝑡− 𝑅𝐹_𝑡= 𝑎_𝑖+ 𝑏_𝑖(𝑅𝑀_𝑡− 𝑅𝐹_𝑡) + 𝑠_𝑖(𝑆𝑀𝐵_𝑡) + ℎ_𝑖(𝐻𝑀𝐿_𝑡) + 𝑚_𝑖(𝑀𝑂𝑀_𝑡) + 𝑒_𝑖,𝑡 (5) Where 𝑆𝑀𝐵_𝑡 is the size factor measured as the return difference of small (cap) minus big (cap) stocks, 𝐻𝑀𝐿_𝑡 the value factor measured as the return difference of high B/M stocks minus low B/M stocks and 𝑀𝑂𝑀_𝑡 the momentum factor measured as the return difference of past winners minus past losers. Moreover, 𝑠_𝑖, ℎ_𝑖 and 𝑚_𝑖 are the coefficients of the

(13)

13 (1993) with an additional profitability and investment factor:

𝑅_𝑖,𝑡− 𝑅𝐹_𝑡= 𝑎_𝑖+ 𝑏_𝑖(𝑅𝑀_𝑡− 𝑅𝐹_𝑡) + 𝑠_𝑖(𝑆𝑀𝐵_𝑡) + ℎ_𝑖(𝐻𝑀𝐿_𝑡) + 𝑟_𝑖(𝑅𝑀𝑊_𝑡)

+𝑐_𝑖(𝐶𝑀𝐴_𝑡) + 𝑒_𝑖,𝑡 (6)

Where 𝑅𝑀𝑊𝑡 is the profitability factor measured as the return difference of the most

profitable stocks minus the least profitable stocks and 𝐶𝑀𝐴_𝑡 the investment factor measured as the return difference of stocks that invest conservatively and stocks that invest

aggressively. Moreover, 𝑟𝑖 and 𝑐𝑖 are the corresponding coefficients, respectively. Fama and French (2015) argue that this model does a better job in explaining returns than their three-factor model. The three-factors are the Fama/French European Factors, obtained from Kenneth R. French’s data library. As the data library does not contain factors per individual country, the same general European factors are used for all regressions. These factors are widely used in the academic literature to explain asset returns and control for systematic risk. However, I am aware that this could create some biased results as some return generating factors might be country specific. Consequently, finding some significant alphas could be due to model misspecification. The risk-free rate, 𝑅𝐹_𝑡, is proxied by the one-month Euribor rate. As the Euribor rate exists from 1999 onwards, the one-month Euro LIBOR rate has been used for the period from 1995 until 1998.

3.4. Overreaction hypothesis

Besides the regression analysis in which we test whether the cross-section of excess returns in our sample is fully captured by the factor models resulting only in insignificant alphas, I also employ an empirical test to determine whether the overreaction hypothesis is the driver for the value premium. The test will be based on the method of De Bondt and Thaler (1985, 1987), with some modifications. The method is based on the hypothesis that firms with low P/E values are considered to be temporarily undervalued, as investors overreact with

excessive pessimism to recent bad news or information. Conversely, high P/E firms are considered overvalued and should eventually decline in price as they adjust to their intrinsic value. The same explanation should hold for return differences between firms with different market-to-book values of equity (i.e. firms with low M/B ratios could be considered

undervalued and firms with high M/B ratios overvalued).

(14)

14 of next calendar year, the procedure is repeated. According to De Bondt and Thaler (1985), this outperformance becomes larger as the holding period increases. However, as testing all the portfolios for multiple holding periods is beyond the scope of this thesis, I only examine the one-year return difference.

4. Results

This section will examine and discuss the empirical results of the tests proposed in the methodology section. Firstly, the summary statistics of the sample is discussed. The second chapter elaborates on the return results from the value and growth portfolios. Thereafter, the empirical results from the regression analysis and the “overreaction hypothesis” test are discussed.

4.1. Summary statistics

Appendix A contains the number of stocks included in the sample per country, on a yearly basis. Overall, an upward trend in the number of stocks is visible for the first few years of our sample period. For most countries, this trend reverses in either 1999, 2000 or 2001. This seems consistent with the end of the dot-com period in the late 1990’s and early 2000’s. After a subsequent decrease in the number of stocks, a similar pattern seems to exist around the start of the financial crisis in 2007 and 2008. Besides the number of stocks, table 1 also contains the yearly median values for the P/E and M/B ratios per country. As both ratios contain some outliers in the top quartile, the median value rather than the mean value is displayed. Over the years, similar trends are visible. Stock prices and stock markets move in trends and an upward trend is usually interrupted by an unexpected shock and then reverses. The valuations of stocks up to the top of the new economical period or up to the beginning of the global financial crisis are usually relatively high. In the next chapter, the potential

influence of these two periods on the returns of the value and growth portfolios are also examined.

4.2. Value premium

P/E

Appendix B shows the annual one year buy-and-hold returns for the value-weighted value and growth portfolios per country, based on their P/E ratios. Moreover, it shows the implied value premium (i.e. the spread between the return of the value portfolio and growth

(15)

15 The sample period includes value and growth portfolio returns for 24 years. When value-weighted portfolios are formed according to the P/E ratios, Belgium yields a positive value premium in 13 out of those 24 years. Moreover, France in 12, Italy and Switzerland in 11, Germany in 10, UK in 9, Sweden in 9 and the Netherlands in 8. Based on the average annualized returns, the value premiums are negative for all countries. Moreover, a two-sample t-test is conducted to test for a significant difference in means between the two groups. Based on the t-statistics for all countries, we can state the difference is statistically insignificant. When the portfolios are formed with equal weights for the individual value and growth stocks, France yields an annual positive premium most often: 13 times. Belgium and Germany 11, Italy and the Netherlands 8, Switzerland 7, Sweden 6 and the UK the least amount of years: 5. Again, the average annualized returns for the portfolios imply negative value premiums for all countries over the full sample period. The difference is statistically insignificant according to the t-statistics.

M/B

The returns of the value-weighted value and growth portfolios, constructed according to the M/B ratios, and the corresponding value premiums are shown in appendix D. Moreover, appendix E shows the returns of the equally-weighted value and growth portfolios. Both tables also include the annualized returns per country, just as the t-statistics derived from the two sample t-test.

When value-weighted portfolios are formed based on M/B ratios, none of the countries has a positive premium for more than half of the sample period. Belgium and Switzerland

experienced a positive value premium in half of the sample period. The Netherlands and the UK have positive premiums in only 9 out of the 24 years, France, Italy and Sweden in 8, and Germany in 7. Over the full sample period only Belgium has a positive annualized value premium of 1.8%, all the other countries have negative premiums. Especially Sweden and France show large negative premiums of -8.8% and -6.9%. However, based on the t-statistics from the two sample t-test and its critical values, the difference in mean between the two groups is statistically insignificant for all countries. For the equally-weighted value and growth portfolios, Belgium, again, yields a positive premium in half of the sample years. However, the other countries experience positive value premiums in relatively few years. The average annualized returns result in negative overall value premiums for all countries.

According to the corresponding t-statistics, the differences in means between the value and growth portfolios for all countries are statistically insignificant.

(16)

16 growth stocks over the full sample period in European stock markets. However, these results do not account for shocks and trends in the stock market. It might be that the value and growth strategies perform differently depending on the market environment (i.e. markets are bullish or bearish). Lakonishok at al. (1994) showed that value stocks outperformed growth stocks during recessions and bear markets in their US sample. Chan & Lakonishok (2004) found similar results when quarterly real GNP growth was used as a proxy for good and bad times. Moreover, Athanassakos (2009) documented significantly higher value premiums during bear markets than during bull markets. Hence, it can be argued that the market environment has a significant impact on the performance of value and growth strategies. Figure 1 shows the development graph of the Euro Stoxx 50 index, the leading blue-chip index for the Eurozone, for the period 1995 to 2018. The cyclicality of the Euro Stoxx 50 index during this period includes bull markets when the investor sentiment is high and bear markets when the investor sentiment is low. Bull markets are identified as lengthy periods with rising stock prices, whereas bear markets experience the opposite (i.e. falling stock prices) for a certain period. In 2001, 2003 and 2009 more than half of the countries documented a positive value premium for both of the portfolio construction methods (P/E ratio and M/B ratio) and the portfolio weighting methods (value-weighted and equally-weighted). Based on the development of the Euro Stoxx 50 index, these years were either during a bear market (2001) or subsequent to a bear market (2003 and 2009). Moreover, for the portfolios ranked according to the P/E ratios, more than half of the countries showed a positive premium in 2002 and 2012 as well. These years fit into a similar pattern: in 2002 a severe bear market ended and recovery started, whereas 2012 was the year subsequent to the ending of a (less severe) bear market. Hence, it can be concluded that value stocks

outperform growth stocks during (the end of) a bear market and the subsequent year of recovery. However, according to the t-statistics the differences in returns are not statistically significant; therefore, one must be cautious in interpreting this result.

Figure 1: The Development of the Euro Stoxx 50 Index

(17)

17 To examine whether the performance of value and growth portfolios differs for periods of economic expansions and recessions, the annual growth rates of real GDP in the Euro area were collected (figure 2). Similar to the method of Chan and Lakonishok (2004), it serves as a proxy for good (i.e. expansions) and bad (i.e. recessions) states of the economy. For the five years of the sample period in which the annual growth rate of real GDP in the Euro area was the lowest, only two years (2009 and 2012) had a positive premium for more than half of the countries. For the other years, growth stocks outperformed value stocks. In 2002 and 2003, years in which a positive premium was also found in more than half of the countries, the real GDP growth rates were below average (although this was not the case for 2001). For the years with relatively high annual growth rates, growth stocks yield superior returns. Hence, there might be a relationship between the performance of value and growth stocks and the economic cycle; however, it is not clear-cut and it should further be examined.

Figure 2: Annual Real GDP Growth Euro Area

Source: The World Bank 4.3. Regression outcomes

Besides examining the existence of the value premium, this study also seeks to explain the return differences of the value and growth portfolios. To control for systematic risk, the annual excess returns of all the value and growth portfolios are regressed on three different (multi-) factor models, as outlined in the methodology section. If a factor model is well suited for explaining the cross-section of portfolio returns, one would expect significant coefficients for the different factors, an alpha not statistically different from zero and a high adjusted R² value. If so, this would imply that a large percentage of the portfolio returns are explained by the four factors or the returns are earned due to the exposure to these factors and earned factor premiums. However, if not (i.e. few statistically significant factor coefficients, a statistically significant non-zero alpha and/or a low adjusted R² value), the cross-section of

(18)

18 portfolio returns are largely explained by other risk factors not employed in this model and any excess return could be either due to these omitted factors or due to the superior strategy.

4.3.1. CAPM regression outcomes

Appendix Fshows the results of the CAPM regressions for all value-weighted portfolios on a country level. The upper part of the table includes the value and growth portfolios

constructed according to the P/E ratios, whereas the lower part is constructed similarly but then for the value and growth portfolios constructed based on the M/B ratios. Nearly all of the portfolios, 31 out of the 32, have a significant market beta (Mkt-RF). Moreover, all of the 16 value-weighted value portfolios have insignificant alphas. Hence, the market beta in the one-factor CAPM explains the returns of the value portfolios well. The growth portfolios, however, show only few insignificant alphas, which means that their returns are largely explained other risk factors than solely the market risk. In appendix Gthe CAPM regression outcomes for the equally-weighted portfolios are shown. Similar to the value-weighted portfolios, most of the growth portfolios have significant alphas, whereas there is only one equally-weighted value portfolio with a significant alpha. Moreover, there is only one insignificant market beta among the 32 portfolios. This is not very surprising as the market still explains the majority of returns in most models (see e.g. Adcock et al., 2019). Naturally, this is very intuitive: the individual stock performance is often related to the performance of the overall market and what is most important is the sensitivity of the firm’s returns to the market returns. Hence, it is not surprising that there is still an ongoing debate about the CAPM as the superior model. In addition, some research shows that the other factors are most important during recession or crisis periods but not in “normal” periods.

4.3.2. Carhart four factor model regression outcomes

(19)

19 could conclude that a large part of the portfolio returns are also explained by other risk

factors not employed in the model.

4.3.3. Fama French five factor model regression outcomes

Appendix Jand appendix K show the results of the regressions of the excess portfolio returns on the Fama French five factor model, for the value-weighted and equally-weighted

portfolios respectively. The model includes a profitability factor (RMW) and an investment factor (CMA) and leaves out the momentum factor (MOM). Remarkably, compared to the results of the CAPM and Carhart four factor model, the market beta is insignificant in more than half of all the value- and equally weighted portfolios. Moreover, a fairly large portion of the coefficients of other risk factors are insignificant as well. Similar to the outcomes of the other two models, the value portfolios have only one significant alpha. However, most of the growth portfolio alphas are still significant.

Overall, none of these three well-known factor models does seem to explain the cross-section of average portfolio returns very well. The only coefficient that is notably significant for almost the whole spectrum of portfolios in our sample, is the market beta. The value

portfolios have very few significant alphas for all three models. However, a large portion of the growth portfolio alphas are significantly different from zero at the 5% significance level. Therefore, the cross-section of these portfolio returns are largely explained by other risk factors not employed in this model and any excess return could be either due to these omitted factors or due to a superior growth investing strategy.

4.4. Overreaction hypothesis

As mentioned above, the results of the multiple regression analyses have showed that parts of the portfolio returns are left unexplained. Hence, it could be that the returns are driven by (irrational) investor behaviour. Similar to De Bondt and Thaler (1985), a contrarian investment strategy is employed to test the overreaction hypothesis. Based on their return performance in the prior calendar year, stocks are divided into deciles in January. The top decile is classified as the winner portfolio (W) and the bottom decile as the loser portfolio (L). The hypothesis predicts that the loser portfolios yield higher returns than the winner portfolios. Appendix L shows the annual one year buy-and-hold returns for both portfolios per country. As the first portfolio is constructed at the beginning of 1996, the year 1995 is excluded from the sample.

(20)

t-20 statistics from the two sample t-test indicate that the difference in means of the winner and loser portfolios for all countries are not significantly different from zero. There are only three years in which loser portfolios outperformed winner portfolios in more than half of the countries. In 2003 in 5 countries, in 2009 in all and in 2018 in 6. In 2003 most of the countries did show positive value premiums, whereas in 2009 all of them did. However, in the other years that more than half of the countries had positive premiums, the winner portfolios clearly outperformed the loser portfolios. Hence, the overreaction hypothesis does not seem to drive the premium in this sample. De Bondt and Thaler (1985) argue, however, that the overreaction effect is the strongest in the second and third year of the portfolio returns and that price reversals correcting the overpriced and under-priced securities do not occur within in the first twelve months. Therefore, the results could be different if the buy-and-hold period is extended.

5. Conclusion

Results from this study show that for the period from 1995 to 2018 value stocks do not persistently outperform growth stocks in the major European stock markets. Based on stock- and accounting data from eight major European stock markets, both value-weighted and equally-weighted value and growth portfolios have been constructed, based on the P/E (price-to-earnings ratio) and M/B (ratio of market-to-book value of equity) indicators. None of the countries examined yield a persistent positive premium over the full sample period. This is not in line with the literature that has established persistent value premiums in the US (see e.g., Basu, 1977; Rosenberg et al., 1985; Fama and French, 1992; Lakonishok et al., 1994) and in Europe (see e.g., Capaul et al., 1993; Bauman et al., 1998; Fama and French, 1998). In line with Lakonishok et al. (1994) and Athanassakos (2009), the value premium is the

(21)

21

5.1. Recommendations

(22)

22

References

Adcock, C., Bessler, W., Conlon, T., 2019. Fundamental factor models and macroeconomic risks – An orthogonal decomposition. Unpublished working paper.

Athanassakos, G., 2009. Value versus growth stock returns and the value premium: The Canadian experience 1985-2005. Canadian Journal of Administrative Sciences 26(2), 109-121.

Basu, S., 1977. Investment performance of common stocks in relation to their price-earnings ratios: A test of the efficient market hypothesis. The Journal of Finance 32(3), 663-682. Basu, S., 1983. The relationship between earnings' yield, market value and return for NYSE common stocks: further evidence. Journal of Financial Economics 12(1), 129-156.

Black, A., McMillan, D., 2006. Asymmetric risk premium in value and growth stocks. International Review of Financial Analysis 15(3), 237-24

Bauman, W., Conover, C, M., Miller, R., 1998. Growth versus value and large-cap versus small-cap stocks in international markets. Financial Analysts Journal 54(2), 75-89.

Bauman, W, S., Miller, R, E., 1997. Investor explanation and the performance of value stocks versus growth stocks. Journal of Portfolio Management 23(3), 57-68.

Bauman, W, S., Conover, C, M., Miller, R, E., 1998. Growth versus value and large-cap versus small-cap stocks in international markets. Financial Analysts Journal 54(2), 75-89. Bessembinder, H., 2018. Do stocks outperform treasury bills? Journal of Financial

Economics 129(3), 440-457.

Bessembinder, H., Chen, T., Choi, G., Wei, K., 2019. Do Global Stocks Outperform US Treasury Bills? Unpublished working paper. Arizona State University.

Black, F., 1972. Capital market equilibrium with restricted borrowing. The Journal of Business 45(3), 444-455.

Capaul, C., Rowley, I., Sharpe, W., 1993. International value and growth stock returns. Financial Analysts Journal 49(1), 27-36.

Carhart, M., 1997. On persistence in mutual fund performance. The Journal of Finance 52(1), 57-82.

Chan, L, K, C., Hamao, Y., Lakonishok, J., 1991. Fundamentals and stock returns in Japan. The Journal of Finance 46(5), 1739-1764.

Chan, L, K, C., Lakonishok, J., 2004. Value and growth investing: Review and update. Financial Analysts Journal 60(1), 71-86.

(23)

23 Cochrane, J, H., 2011. Presidential address: Discount rates. The Journal of Finance

66(4), 1047-1108.

De Bondt, F., Thaler, R., 1985. Does the stock market overreact? The Journal of Finance 40(3), 793-805.

De Bondt, F., Thaler, R., 1987. Further evidence on investor overreaction and stock market seasonality. The Journal of Finance 42(3), 557-581.

Fama, E., 1970. Efficient capital markets: A review of theory and empirical work. The Journal of Finance 25(2), 383-417.

Fama, E., 1991. Efficient capital markets: II. The Journal of Finance 46(5), 1575-1617. Fama, E., French, K., 1992. The cross-section of expected stock returns. The Journal of Finance 47(2), 427-465.

Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33(1), 3-56.

Fama, E., French, K., 1995. Size and book-to-market factors in earnings and returns. The Journal of Finance 50(1), 131-155.

Fama, E., French, K., 1998. Value versus growth: The international evidence. The Journal of Finance 53(6), 1975-1999.

Fama, E., French, K., 2015. A five-factor asset pricing model. Journal of Financial Economics 116(1), 1-22.

Fama, E., MacBeth, J., 1973. Risk, return, and equilibrium: Empirical tests. Journal of Political Economy 81(3), 607-636.

Graham, B., Dodd, D, L., 1934. Security Analysis. Whittlesey House, McGraw-Hill Book Co.

Hou, K., Xue, C., Zhang, L., 2015. Digesting anomalies: An investment approach. The Review of Financial Studies 28(3), 650-705.

Jaffe, J., Keim, D., Westerfield, R., 1989. Earnings yields, market values, and stock returns. The Journal of Finance 44(1), 135-148.

Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: implications for stock market efficiency. Journal of Finance 48(1), 65-91.

Jensen, M., 1968. The performance of mutual funds in the period 1945-1964. The Journal of Finance,23(2), 389-416.

(24)

24 Lintner, J., 1965. The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. The Review of Economics and Statistics 47(1), 13-37.

Mateus, I., Mateus, C., Todorovic, N., 2019. Review of new trends in the literature on factor models and mutual fund performance. International Review of Financial Analysis 63, 344-354.

Nicholson, S, F., 1960. Price-earnings ratios. Financial Analysts Journal 16(4), 43-45. Rosenberg, B., Reid, K., Lanstein, R., 1985. Persuasive evidence of market inefficiency. Journal of Portfolio Management 11(3), 9-16.

Schmidt, P., Von Arx, U., Schrimpf, A., Wagner, A., Ziegler, A., 2019. Common risk factors in international stock markets. Financial Markets and Portfolio Management 33(3), 213-241. Sharpe, W., 1964. Capital asset prices: A theory of market equilibrium under conditions of risk. The Journal of Finance 19(3), 425-442.

(25)

Appendix

A. Descriptive Summary Statistics Sample

Year _N _MedianBelgium France Germany Italy

P/E Median M/B N Median P/E Median M/B N Median P/E Median M/B N Median P/E Median M/B

(26)

Notes: N is the number of stocks included in the sample per year and per country. P/E is the price-to-earnings ratio measured as 𝑃 𝐸⁄ 𝑖,𝑡= 𝑃𝑙̇,𝑡⁄𝐸𝑃𝑆𝑖,𝑡. M/B is the market-to-book value of equity measured as 𝑀 𝐵⁄ 𝑖,𝑡= 𝑀𝑉𝑖,𝑡⁄𝐵𝐸𝑖,𝑡.

Year _N Netherlands _Median Sweden Switzerland UK

P/E Median M/B N Median P/E Median M/B N Median P/E Median M/B N Median P/E Median M/B

(27)

B. Value Premium: Value-Weighted P/E Portfolios

Year

Belgium France Germany Italy

Value Growth Premium Value Growth Premium Value Growth Premium Value Growth Premium

(28)

Notes: Value is the annual value-weighted portfolio return on the low P/E ratio portfolios. Growth is the annual value-weighted return on the high P/E ratio portfolios. The

premium is the spread in returns between value and growth. An. is the annualized average return for the portfolios.

Year

Netherlands Sweden Switzerland UK

(29)

C. Value Premium: Equally-Weighted P/E Portfolios

Year

(30)

Year

Value Growth Premium Value Growth Premium Value Growth Premium Value Growth Premium 1995 3,2% 16,3% -13,1% -5,0% 7,8% -12,8% -2,5% 17,7% -20,2% 4,2% 23,3% -19,0% 1996 24,4% 52,9% -28,5% 38,2% 71,5% -33,4% 5,0% 15,6% -10,6% 6,7% 30,1% -23,4% 1997 15,3% 36,6% -21,2% 21,1% 40,4% -19,3% 28,1% 36,9% -8,8% 0,4% 17,2% -16,8% 1998 -9,3% 23,9% -33,1% -21,6% 37,4% -59,0% -0,8% 14,6% -15,4% -16,5% 7,6% -24,0% 1999 3,4% 24,7% -21,4% 7,5% 37,8% -30,3% 17,6% 49,2% -31,5% 23,1% 60,6% -37,5% 2000 18,2% 11,2% 7,0% 13,0% 32,5% -19,5% 11,1% 46,1% -35,0% -0,8% 13,5% -14,3% 2001 -18,6% -13,7% -4,9% 6,6% -15,0% 21,6% -23,4% -20,3% -3,0% 0,7% -5,5% 6,1% 2002 -11,5% -8,0% -3,5% 1,9% -12,9% 14,9% -16,1% -12,7% -3,3% -2,6% -6,7% 4,0% 2003 26,7% 5,6% 21,0% 24,3% 13,6% 10,7% 29,1% 25,5% 3,6% 25,0% 21,5% 3,5% 2004 18,1% 14,8% 3,3% 18,1% 23,8% -5,7% 20,5% 16,3% 4,2% 7,6% 17,7% -10,1% 2005 12,2% 27,9% -15,7% 27,4% 48,4% -21,0% 33,0% 30,9% 2,0% 5,2% 13,1% -7,9% 2006 8,8% 21,4% -12,6% 14,3% 16,8% -2,5% 24,2% 24,1% 0,1% 5,9% 14,2% -8,3% 2007 -0,2% 11,7% -11,9% -1,5% 6,5% -7,9% 9,8% 29,7% -19,9% -2,1% 7,0% -9,0% 2008 -30,0% -14,2% -15,9% -41,8% -18,3% -23,5% -39,4% -20,3% -19,2% -35,3% -8,8% -26,5% 2009 26,9% 10,5% 16,4% 68,5% 32,3% 36,2% 22,5% 9,7% 12,8% 41,3% 16,1% 25,2% 2010 3,4% 6,4% -3,0% 6,0% 9,5% -3,5% 13,8% 19,4% -5,6% 7,1% 10,6% -3,6% 2011 -2,7% -4,9% 2,2% -7,9% 0,3% -8,1% -8,8% -3,3% -5,5% -3,5% 2,0% -5,5% 2012 5,3% 5,7% -0,4% 5,8% 5,0% 0,8% 16,3% 5,3% 11,1% 6,5% 6,0% 0,5% 2013 5,0% 9,5% -4,5% 36,7% 20,1% 16,6% 14,5% 14,6% -0,1% 11,8% 13,8% -2,0% 2014 1,3% 1,2% 0,1% 10,0% 11,3% -1,3% 8,2% 10,1% -1,9% -0,2% 1,8% -2,0% 2015 4,0% 12,1% -8,1% 15,4% 27,9% -12,5% 5,4% 5,5% -0,2% 0,4% 6,7% -6,3% 2016 0,9% 0,6% 0,3% 10,2% 12,3% -2,0% 1,4% 9,4% -8,0% 1,7% 7,0% -5,4% 2017 10,2% 12,5% -2,3% 5,0% 14,7% -9,7% 7,2% 22,7% -15,5% 3,0% 14,1% -11,0% 2018 -0,1% -1,0% 0,9% 2,0% 12,0% -10,0% 0,7% -2,9% 3,6% -4,1% 1,7% -5,7% An. 4,8% 11,0% -6,2% 10,6% 18,2% -7,6% 7,4% 14,3% -6,9% 3,6% 11,9% -8,3% t-statistic -1,4744 1,2505 -1,3624 -2,0450

Notes: Value is the annual equally-weighted portfolio return on the low P/E ratio portfolios. Growth is the annual equally-weighted return on the high P/E ratio portfolios.

(31)

D. Value Premium: Value-Weighted M/B Portfolios

Year

(32)

Year

1995 4,3% 22,4% -18,0% -6,7% 40,8% -47,5% 5,2% 23,2% -18,0% 16,9% 26,6% -9,7% 1996 23,9% 49,1% -25,2% 33,9% 47,6% -13,7% 16,2% 18,7% -2,5% 16,2% 15,2% 1,0% 1997 19,9% 50,8% -30,9% 22,5% 55,5% -33,0% 74,9% 49,8% 25,1% 12,9% 18,4% -5,6% 1998 _-8,7% _31,4% _-40,1% _-14,9% _46,9% _-61,8% _4,8% _20,9% _-16,1% _-0,5% _25,0% _-25,6% 1999 18,7% 14,7% 4,0% 24,8% 77,8% -53,1% 13,1% 17,0% -3,9% 49,7% 37,6% 12,2% 2000 11,1% 13,3% -2,2% -2,4% 15,8% -18,1% 21,9% 20,8% 1,1% 2,9% 7,7% -4,9% 2001 -14,1% -15,1% 1,0% 17,0% -24,7% 41,8% -22,6% -21,5% -1,1% -14,4% -5,8% -8,7% 2002 _-1,1% _-17,1% _15,9% _-12,3% _-26,7% _14,4% _-21,9% _-9,6% _-12,3% _-16,7% _-10,6% _-6,0% 2003 35,3% -2,0% 37,3% 26,5% 7,7% 18,8% 28,0% 8,5% 19,4% 17,1% 10,5% 6,6% 2004 16,7% 4,0% 12,8% 15,0% 43,7% -28,7% 5,2% 5,2% 0,0% 13,9% 6,7% 7,3% 2005 25,4% 22,8% 2,6% 25,6% 28,2% -2,6% 39,7% 37,3% 2,4% 6,2% 21,0% -14,8% 2006 17,8% 23,8% -5,9% 14,9% 23,4% -8,5% 16,3% 13,8% 2,5% 18,5% 14,9% 3,6% 2007 _13,8% _18,0% _-4,3% _0,4% _19,9% _-19,5% _4,2% _21,3% _-17,1% _0,5% _16,0% _-15,5% 2008 -32,0% -26,3% -5,7% -36,3% -23,7% -12,6% -36,2% -24,4% -11,8% -24,4% -16,9% -7,5% 2009 65,5% 19,8% 45,6% 81,2% 44,2% 36,9% 28,6% 18,4% 10,3% 42,7% 29,5% 13,2% 2010 -3,7% 7,4% -11,1% 14,5% 27,3% -12,8% 2,3% 10,2% -7,9% -0,8% 15,1% -15,9% 2011 -30,3% 1,7% -32,1% -4,7% -3,5% -1,2% -8,0% -2,6% -5,3% -1,1% 8,3% -9,4% 2012 _5,1% _17,0% _-11,8% _3,4% _10,3% _-6,9% _28,1% _21,8% _6,3% _13,2% _13,2% _0,0% 2013 24,4% 12,6% 11,8% 22,0% 20,6% 1,4% 24,6% 24,6% 0,0% 25,6% 17,4% 8,2% 2014 -7,1% 13,4% -20,5% 12,8% 14,0% -1,2% 13,6% 16,3% -2,8% 4,2% 9,9% -5,7% 2015 -7,2% 20,1% -27,3% 17,1% 32,1% -15,0% 9,4% 4,8% 4,7% -3,0% 9,7% -12,7% 2016 19,9% -5,7% 25,6% 8,2% -1,3% 9,5% 2,5% -4,5% 7,1% 39,1% 5,4% 33,8% 2017 _12,6% _30,3% _-17,7% _12,2% _18,0% _-5,8% _13,0% _25,4% _-12,4% _-0,7% _13,4% _-14,1% 2018 0,9% 2,5% -1,6% 14,6% 6,0% 8,7% 2,6% 1,7% 0,9% -0,4% 8,3% -8,6% An. 8,8% 12,9% -4,1% 12,1% 20,8% -8,8% 11,1% 12,4% -1,3% 9,1% 12,4% -3,3% t-statistic -0.7117 -1.2693 -0.2300 -0.7459

Notes: Value is the annual value-weighted portfolio return on the low M/B ratio portfolios. Growth is the annual value-weighted return on the high M/B ratio portfolios. The

(33)

E. Value Premium: Equally-Weighted M/B Portfolios

Year

(34)

Year

1995 5,6% 22,3% -16,7% 0,9% 17,6% -16,8% 6,3% 17,2% -10,9% 9,7% 25,6% -15,9% 1996 24,1% 71,3% -47,3% 43,1% 60,6% -17,6% 6,9% 16,3% -9,3% 13,2% 25,5% -12,3% 1997 13,4% 37,0% -23,6% 17,3% 39,8% -22,5% 36,0% 33,5% 2,5% 7,5% 8,7% -1,2% 1998 -3,7% 20,3% -24,0% -16,8% 33,1% -50,0% 6,8% 9,4% -2,6% -12,2% 6,0% -18,2% 1999 9,6% 14,3% -4,7% 21,7% 45,4% -23,7% 24,7% 66,3% -41,6% 26,1% 64,2% -38,1% 2000 5,2% 9,2% -4,1% 9,3% 17,4% -8,1% 17,5% 42,9% -25,3% 0,3% 10,0% -9,7% 2001 -11,6% -17,8% 6,2% -7,0% -20,4% 13,3% -21,2% -31,9% 10,8% -5,7% -10,4% 4,7% 2002 _-9,5% _-11,1% _1,6% _-16,5% _-20,3% _3,8% _-19,0% _-20,4% _1,4% _-10,1% _-8,0% _-2,1% 2003 33,0% 9,2% 23,7% 18,5% 16,5% 1,9% 24,5% 24,1% 0,4% 27,3% 26,9% 0,4% 2004 10,1% 18,7% -8,6% 24,4% 16,6% 7,8% 13,4% 11,2% 2,2% 5,9% 13,1% -7,2% 2005 8,8% 21,3% -12,6% 43,0% 47,5% -4,5% 37,1% 33,3% 3,8% -0,9% 11,8% -12,6% 2006 7,2% 18,1% -11,0% 9,4% 26,9% -17,5% 17,3% 28,2% -10,9% 2,4% 12,1% -9,8% 2007 _-3,0% _9,8% _-12,8% _-4,0% _9,5% _-13,5% _4,5% _30,6% _-26,1% _-1,8% _10,4% _-12,2% 2008 -23,2% -18,2% -4,9% -38,6% -26,2% -12,4% -23,7% -29,2% 5,5% -28,0% -12,6% -15,4% 2009 22,1% 10,7% 11,4% 53,0% 38,2% 14,8% 13,7% 20,8% -7,1% 39,0% 22,8% 16,1% 2010 4,6% 3,9% 0,7% -0,8% 6,3% -7,1% 5,3% 15,0% -9,8% 5,1% 11,4% -6,2% 2011 -6,8% -2,8% -4,0% -12,3% -3,2% -9,1% -9,2% -7,7% -1,6% -4,1% 2,3% -6,4% 2012 _1,5% _4,4% _-3,0% _-7,5% _7,6% _-15,1% _3,8% _15,1% _-11,3% _2,8% _7,2% _-4,4% 2013 9,3% 8,3% 1,0% 25,8% 46,8% -21,0% 7,3% 22,8% -15,5% 8,9% 16,2% -7,2% 2014 -0,2% 3,1% -3,3% -3,5% 8,9% -12,4% 5,8% 13,1% -7,3% -3,9% 2,3% -6,2% 2015 1,7% 8,4% -6,7% 16,8% 44,5% -27,7% -1,9% 10,3% -12,2% -3,5% 6,2% -9,7% 2016 1,3% 9,1% -7,8% 5,9% 13,5% -7,5% 2,3% 14,2% -11,8% 6,4% 5,7% 0,7% 2017 _3,1% _23,1% _-20,0% _-1,1% _17,3% _-18,4% _10,3% _22,1% _-11,8% _0,5% _13,8% _-13,3% 2018 -1,1% -3,4% 2,4% 0,6% 6,7% -6,1% -2,0% -7,3% 5,4% -1,7% -0,3% -1,5% An. 4,2% 11,2% -7,0% 7,6% 18,8% -11,2% 6,9% 14,6% -7,6% 3,5% 11,3% -7,8% t-statistic -1,5755 -1,7621 -1,3821 -1,8632

Notes: Value is the annual equally-weighted portfolio return on the low M/B ratio portfolios. Growth is the annual equally-weighted return on the high M/B ratio portfolios.

(35)

F. CAPM Regression Results: Value-Weighted Portfolios

Notes: Results from CAPM regressions (4): 𝑅𝑖,𝑡− 𝑅𝐹𝑡= 𝑎𝑖+ 𝑏𝑖(𝑅𝑀𝑡− 𝑅𝐹𝑡) + 𝑒𝑖,𝑡. The values in the parentheses are the robust standard errors of the coefficients. Statistically significant coefficients have been denoted with * for the 10 percent significance level, ** for the 5 percent significance level and

*** for the 1 percent significance level. Adj. R2 are the adjusted R2_values.

Value (low P/E) Growth (high P/E) Value (low M/B) Growth (high M/B)

Country Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2

(36)

G. CAPM Regression Results: Equally-Weighted Portfolios

Notes: Results from CAPM regressions (4): 𝑅𝑖,𝑡− 𝑅𝐹𝑡= 𝑎𝑖+ 𝑏𝑖(𝑅𝑀𝑡− 𝑅𝐹𝑡) + 𝑒𝑖,𝑡. The values in the parentheses are the robust standard errors of the coefficients. Statistically significant coefficients have been denoted with * for the 10 percent significance level, ** for the 5 percent significance level and

*** for the 1 percent significance level. Adj. R2 are the adjusted R2_values.

Value (low P/E) Growth (high P/E) Value (low M/B) Growth (high M/B)

Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2 Alpha Mkt-RF Adj. R2

(37)

H. Carhart Four Factor Model Regression Results: Value-Weighted Portfolios

Value (low M/B) Growth (high M/B)

Country Alpha Mkt-Rf SMB HML MOM Adj. R2 Alpha Mkt-Rf SMB HML MOM Adj. R2

Belgium 0.0639 (0.0469) 0.9125*** (0.1536) -0.3090 (0.3270) 0.1889 (0.2519) -0.0681 (0.1958) 0.5380 0.0837** (0.0399) -0.6920*** (0.1622) -0.9674** (0.4276) -0.1678 (0.2262) -0.0520 (0.1447) 0.3854 France 0.0217 (0.0289) 0.7832* (0.1374) 0.2097 (0.3532) 0.0080 (0.2449) -0.2574* (0.1307) 0.6704 0.0892* (0.0475) 0.5814** (0.1651) -0.2817 (0.3987) -0.6006* (0.3167) 0.1282 (0.2007) 0.4119 Germany -0.0149 (0.0291) 0.6526*** (0.1893) -0.0150 (0.5940) 0.2383 (0.2378) 0.1669 (0.1156) 0.3572 0.0575** (0.0271) 0.7163*** (0.1058) -0.6260 (0.3857) -0.3440* (0.1982) -0.0936 (0.1510) 0.6152 Italy 0.0767 (0.1480) 0.6340 (0.4633) 0.2289 (1.2743) 0.0711 (1.0902) -0.2544 (0.6158) -0.0525 0.0740 (0.0487) 0.5786*** (0.2006) -0.5421 (0.5626) -0.2499 (0.3478) -0.0389 (0.1918) 0.2270 Netherlands 0.0478 (0.0336) 0.5665*** (0.1065) 0.6221* (0.3007) 0.2473 (0.1841) -0.3523** (0.1326) 0.7109 0.0848** (0.0318) 0.6781*** (0.1467) -0.9327** (0.3421) -0.3242 (0.1898) -0.0038 (0.1008) 0.5270 Sweden 0.1204** (0.0467) 0.3447 (0.2077) 1.0118** (0.3711) -0.1730 (0.2428) -0.4739** (0.1909) 0.5474 0.1704*** (0.0481) 0.8225*** (0.1714) -0.3983 (0.4781) -0.8953** (0.3504) -0.0385 (0.1794) 0.5309 Switzerland 0.0702* (0.0364) 0.6956*** (0.2043) -0.4651 (0.6712) 0.0677 (0.2524) -0.2257 (0.1316) 0.3651 0.0742** (0.0307) 0.6216*** (0.1404) -0.6892* (0.3363) -0.1224 (0.2031) -0.0181 (0.1120) 0.4526 UK 0.0706** (0.0307) 0.4301*** (0.1149) 0.4877 (0.3337) -0.3526 (0.2392) -0.2360** (0.0990) 0.4621 0.0997*** (0.0210) 0.4144*** (0.0921) -0.1310 (0.1437) -0.4589*** (0.1319) -0.0794 (0.0715) 0.6825

Notes: Results from the Carhart four factor model regressions (5): 𝑅𝑖,𝑡− 𝑅𝐹𝑡= 𝑎𝑖+ 𝑏𝑖(𝑅𝑀𝑡− 𝑅𝐹𝑡) + 𝑠𝑖(𝑆𝑀𝐵𝑡) + ℎ𝑖(𝐻𝑀𝐿𝑡) + 𝑚𝑖(𝑀𝑂𝑀𝑡) + 𝑒𝑖,𝑡. The values

Value (low P/E) Growth (high P/E)

(38)

(39)

I. Carhart Four Factor Model Regression Results: Equally-Weighted Portfolios

Value (low M/B) Growth (high M/B)

Belgium 0.0651** (0.0311) 0.4415*** (0.1094) (0.3383) 0.3360 (0.2646) -0.0283 -0.1749* (0.0840) 0.4780 (0.0310) 0.0437 0.8355*** (0.1969) (0.6554) -0.9216 (0.1891) 0.0188 (0.1040) 0.1578 0.4370 France 0.0632 (0.0430) (0.2833) 0.0525 (0.3396) 0.5496 (0.5165) 0.6355 (0.2242) -0.1086 0.1373 0.0985** (0.0456) 0.3355** (0.1539) (0.3248) -0.0727 (0.2935) -0.0956 (0.2000) -0.1085 0.1167 Germany 0.0621 (0.0400) 0.6069*** (0.1377) (0.4322) 0.2477 (0.2512) -0.2419 -0.4451* (0.2155) 0.4833 0.0838** (0.0313) 0.4670*** (0.1182) (0.3134) 0.2327 (0.1891) -0.1532 (0.1605) -0.0958 0.3896 Italy -0.0441 (0.0379) 0.5674** (0.1552) (0.4376) -0.2193 (0.2485) 0.1468 (0.1327) -0.0568 0.3333 (0.0525) 0.0447 0.5936*** (0.1940) (0.4955) -0.1430 (0.3681) -0.2075 (0.2223) -0.0446 0.2491 Netherlands 0.0092 (0.0155) 0.3726*** (0.0782) (0.1794) 0.2704 (0.0724) 0.0040 -0.1736** (0.0644) 0.6853 0.0517** (0.0233) 0.6150*** (0.1336) -0.6364* (0.3132) (0.1429) -0.1207 (0.1108) 0.0695 0.3518 Sweden 0.0447 (0.0324) 0.4278** (0.1771) 0.7746** (0.3674) (0.1807) -0.0308 -0.2816** (0.1314) 0.4164 0.1375*** (0.0409) 0.7052*** (0.1710) (0.3965) -0.0446 (0.2876) -0.4484 (0.1639) -0.0499 0.3823 Switzerland 0.0286 (0.0304) 0.4866*** (0.1126) (0.3508) -0.0681 (0.2232) 0.0234 (0.1524) -0.1227 0.3837 (0.0627) 0.1025 0.6182*** (0.1985) (0.4880) 0.0752 (0.4303) -0.3479 (0.2806) -0.1027 0.3142 UK 0.0178 (0.0197) 0.3244*** (0.1036) 0.5618** (0.2411) (0.1313) -0.1734 -0.2636*** (0.0600) 0.6448 0.0754** (0.0279) 0.3821*** (0.0873) (0.3212) 0.4393 -0.4749* (0.2637) (0.1259) -0.0192 0.5349

Notes: Results from the Carhart four factor model regressions (5): 𝑅𝑖,𝑡− 𝑅𝐹𝑡= 𝑎𝑖+ 𝑏𝑖(𝑅𝑀𝑡− 𝑅𝐹𝑡) + 𝑠𝑖(𝑆𝑀𝐵𝑡) + ℎ𝑖(𝐻𝑀𝐿𝑡) + 𝑚𝑖(𝑀𝑂𝑀𝑡) + 𝑒𝑖,𝑡. The values

(40)

(41)

J. Fama French Five Factor Model Regression Results: Value-Weighted Portfolios

Value (low M/B) Growth (low M/B)

Country Alpha Mkt-RF SMB HML RMW CMA Adj.R2 Alpha Mkt-RF SMB HML RMW CMA Adj.R2

Belgium 0.0277 (0.0474) 0.8499*** (0.2640) -0.2761 (0.3617) 0.6428 (0.6293) 0.4928 (0.5427) -0.4911 (1.0061) 0.5268 0.0089 (0.0445) 0.6312*** (0.1818) -1.0063* (0.4797) 0.6706* (0.3603) 1.2389** (0.5853) -0.7068 (0.5682) 0.4921 France 0.0180 (0.0346) 0.5694** (0.2004) 0.2806 (0.3203) 0.6381 (0.4980) -0.5067 (0.5602) -1.1889 (0.7482) 0.6571 0.1168** (0.0469) 0.2746 (0.2072) -0.3029 (0.4052) 0.2574 (0.5088) -0.0214 (0.7097) -1.4829* (0.7206) 0.4778 Germany 0.0282 (0.0457) 0.2887 (0.1818) 0.0348 (0.5715) 1.0948* (0.5251) -0.1947 (0.4709) -1.6401** (0.7043) 0.4220 0.1472*** (0.0308) 0.5150*** (0.1056) -0.5676* (0.3009) -0.2922 (0.2753) -1.1803*** (0.2829) -0.6004 (0.4151) 0.7545 Italy 0.1190 (0.1419) -0.3048 (0.8073) 0.3150 (0.9849) 2.7216 (1.8858) -1.0148 (1.0616) -4.8247* (2.7737) 0.0960 0.0973** (0.0430) 0.2058 (0.2383) -0.4825 (0.4793) 0.8097 (0.6319) -0.3247 (0.4972) -1.8457** (0.7033) 0.3585 Netherlands -0.0260 (0.0415) 0.4845** (0.1794) 0.6605* (0.3390) 0.9138* (0.4847) 0.3769 (0.5950) -0.8549 (0.6939) 0.6206 0.0687* (0.0332) 0.4462*** (0.1376) -0.9324** (0.3386) 0.6388 (0.3700) 0.4116 (0.4789) -1.3073** (0.4542) 0.6295 Sweden 0.0017 (0.0541) 0.1497 (0.1847) 1.0762** (0.4285) 1.1092** (0.4917) 0.8503 (0.8215) -1.6704** (0.7756) 0.5150 0.1738*** (0.0404) 0.3019 (0.1959) -0.3578 (0.3119) 0.7770 (0.5055) 0.0661 (0.4172) -2.7354*** (0.7593) 0.7368 Switzerland 0.0998* (0.0546) 0.4061* (0.2018) -0.3243 (0.6393) 0.7337 (0.6113) -0.9312** (0.4380) -1.3765 (0.8081) 0.4013 0.0982** (0.0403) 0.4370** (0.2015) -0.6474* (0.3371) 0.3477 (0.5082) -0.3387 (0.3844) -0.8515 (0.7221) 0.4838 UK 0.0425 (0.0461) 0.3432* (0.1607) 0.4906 (0.3241) 0.0612 (0.4367) -0.0993 (0.4523) -0.6698 (0.7474) 0.3777 0.0966*** (0.0214) 0.2589** (0.0997) -0.1189 (0.1087) 0.0659 (0.2620) -0.0740 (0.2190) -0.8612** (0.3792) 0.7527

Notes: Results from the Fama French Five Factor Model regressions (6): 𝑅𝑖,𝑡− 𝑅𝐹𝑡= 𝑎𝑖+ 𝑏𝑖(𝑅𝑀𝑡− 𝑅𝐹𝑡) + 𝑠𝑖(𝑆𝑀𝐵𝑡) + ℎ𝑖(𝐻𝑀𝐿𝑡) + 𝑟𝑖(𝑅𝑀𝑊𝑡) + 𝑐𝑖(𝐶𝑀𝐴𝑡) + 𝑒𝑖,𝑡 . The

Country Alpha Mkt-RF SMB HML RMW CMA Adj. R2 Alpha Mkt-RF SMB HML RMW CMA Adj.R2

(42)