The Effect of Credit Risk on Stock Returns:

1 The Effect of Credit Risk on Stock Returns: An Empirical Research on the Dutch, German and French Stock Market from Dec to Dec Abstract This researc...

Author:  Tracey Harrell
0 downloads 0 Views 1MB Size

The Effect of Credit Risk on Stock Returns: An Empirical Research on the Dutch, German and French Stock Market from Dec. 2004 to Dec. 2012.

Abstract This research examines whether credit spread captures systematic risk which cannot be fully explained by CAPM or FF3. The dataset is based on Dutch, French and German firms between December 2004 and December 2012. The Merton (1974) model is used to create credit spread as characteristic for credit risk. In this way, it is possible to include firms without credit spread or ratings information in the analysis, consequently, this research is free of sample selection biases. This research shows no significant relationship with excess return for the market premium, size premium and credit risk premium. However, there is a significant relationship between excess return and value premium from December 2004 to December 2012 for the FF3 model and the FF3 model in combination with HSMLS.

Student name: Lieke Janssen Administration number: 666045

Date: 23 October, 2014

1. Introduction It is important for managers and investors to know why asset values are changing in the way they do. Asset pricing theories are giving an understanding about the values of assets that generate uncertain future cash flows as the relationship between equity prices and their expected returns. Managers use asset pricing theories to calculate the appropriate cost of capital for making investment decisions and investors are using asset pricing theories to benefit on the revelation of mispriced assets. Furthermore, asset pricing theories are used to provide expected returns in respect of portfolio optimization and benchmarking. In general, asset pricing theories prescribe that riskier assets should command higher returns. Existing theories, however, leave unexplained a host of empirically documented crosssectional patterns in stock returns, classified as anomalies. Before going into detail about asset pricing theories, a distinction is made between two kinds of risks: systematic risk and idiosyncratic risk. Idiosyncratic risk is firm- or industry-specific risk that is inherent in each investment. This kind of risk can be diversified away by holding many different stocks in several industries. Examples of idiosyncratic risk are a new competitor, a management change or a product recall. In contrast, systematic risk is the risk related to the entire market, not just a particular firm or industry. Systematic risk is unpredictable and cannot be avoided completely since it cannot be diminished through diversification. Some examples of systematic risk are interest rate changes, inflation and economic downturns since they affect the entire market. Investors receive a premium for this systematic risk exposure on excess of the risk-free rate which is known as excess return. The most common-used model to predict excess return is the Capital Asset Pricing Model (CAPM) of Sharpe (1964), Lintner (1965) and Black (1972). The CAPM argues that a part of the risk, systematic risk, is unavoidable. However the other part of risk, idiosyncratic risk, can be diversified away. The CAPM explains excess returns based on the stocks price’s covariation with an overall market portfolio that pays a market premium. The beta-coefficient hereby measures the sensitivity of the stock price fluctuations with respect to changes in the market portfolio and is therefore the benchmark for the systematic risk involved. Other academics (Merton, 1973; Basu, 1977, 1983; Banz, 1981; Bhandari, 1988) indicate that systematic risk is a multifactor model. Systematic risk should take state variables and firm-specific effects into consideration as well to improve the model. Fama and French (1992) developed the Fama and French three factor model (FF3) by extending the CAPM with two complementary factors: size and value. Size is the difference in asset returns between small firms and lager firms (SMB), value is the difference in asset returns between firms with high Book-to-Market-equity (B/M-equity) ratios called value stocks and firms with low B/M-equity ratios called growth stocks (HML). The FF3 model shows that value and small cap stocks outperform markets on a regular basis. Fama and French proved to better measure market returns by adding those two additional factors to the CAPM. 1

Another influence on the development of stock prices and returns was discovered by Jegadeesh and Titman (1993) who revealed the Momentum-effect (MOM). Carhart (1997) added MOM as an additional factor to the model, MOM is the difference in asset returns between previous ‘winners’ and previous ‘losers’ in the way that securities that performed well over the past three till twelve months will perform well in the next three till twelve months and the other way around. Contrary to SMB and HML, MOM is an additional factor which finds its fundamentals in irrational pricing and investment behavior. However, the focus of this research is on credit spread based on rational asset pricing, the reason for this decision is explained in section two. A measurement of alpha, R2 and betas are used to measure the performance of an asset pricing model. An asset pricing model is perfect when: their alpha is equal to zero, R2 is equal to one and their betas are significant unequal from zero. In this case, the variation in the stock price returns are perfectly explained by the independent variables in the way that systematic risk is completely captured by the model and no space is left for random errors. Soentjens (2012) argues that the performance of FF3 worsens during economic downturns which might indicate that the effect of SMB and HML lessens during economic downturns or the data might contain a high degree of noise what lowers the explanatory value of FF3. Another explanation is that other explanatory variables than SMB and HML are becoming more relevant during economic downturns and therefore the relevance of SMB and HML decreases during those periods. C. O. Kang and H. G. Kang (2009) argue that credit risk is related to the business cycle since credit risk is increasing during economic downturns. A previous study (Avramov, Chordian, Jostova & Philipov, 2012) showed that there are implications of financial distress for the profitability of anomaly-based trading strategies. However, the strategies of price momentum, earnings momentum, credit risk, dispersion, idiosyncratic volatility, and capital investments are only profitable during economic downturns. In contrast, the value anomaly is profitable during stable or improving credit conditions and the accruals anomaly is profitable during all periods of the business cycle. In this paper is analyzed if credit risk has a premium after controlling for CAPM and FF3 in The Netherlands, Germany and France, and if so, if this premium strengthened during economic downturns. The geographical choice is established by the fact that Germany and France are two major trading partners of The Netherlands with respectively 25% and 9% of the Dutch export in 2012, besides, Germany and France are the biggest economies in the Eurozone. Also Belgium is an important country for the Dutch export with a share of 12% but it is decided to exclude Belgium from this research since its economic value within the Eurozone is negligible (CBS StatLine, 2012). This paper has a strong connection with the paper of Soentjens (2012) who tested CAPM, FF3 and FF4 and concluded that the asset pricing models deteriorated during economic downturns. This paper has also a strong connection to C. O. Kang and H. G. Kang (2009) who concluded that credit risk exhibits a positive premiums after controlling for FF3 and FF4 in the Korean stock market.

2

The paper proceeds as follows. The next sections describes the current literature, section three explains the methodology used to estimate credit risk. Section 4 describes the data, section 5 presents the empirical results, and section 6 concludes with some recommendations for further research and the limitations to this research.

2. Literature The beginning of modern asset pricing models is founded by Markowitz (1952), he presented a new perspective on portfolio selection called Modern Portfolio Theory. Markowitz argues that it is possible to reduce the total portfolio risk by adding more securities to the portfolio and as a result diversifying away idiosyncratic risk of the assets in the portfolio (Appendix 1). A fundamental item of his diversification strategy is adding assets with a low or even negative correlation across the portfolio. Consequently, the strategy in portfolio selection changed from a focus on the individual risks and returns of assets to a mean-variance optimization model. The key concept of the mean-variance optimization model is to build a portfolio with the same or even higher expected return against a lower volatility by eliminating idiosyncratic risk as far as possible using the diversification strategy. As a result, every individual asset and every combination of individual assets can be plotted in a riskexpected return region, the upper edge of this region is called the efficient frontier (Appendix 2). Tobin (1958) added his Separation Theorem to the Modern Portfolio Theory by incorporating an asset which pays-off a risk-free rate. The risk-free rate has zero volatility in its returns and is uncorrelated with the other assets in the portfolio as well. A tangent line called the capital market line is drawn through the risk-free rate and touches the efficient frontier, the point where the tangent line touches the efficient frontier is called the tangency portfolio (Appendix 2). The combination of the risk-free asset with the tangency portfolio has a superior risk-expected return compared to the other portfolios on the efficient frontier. Using the risk-free asset, investors should make a trade-off between the risk-free asset and the tangency portfolio. The more risk-averse an investor is, the more he invests in the risk-free asset, the more risk-seeking an investor is, the more he invests in the tangency portfolio or even goes short on the risk-free asset to use the proceeds for an additional investment in the tangency portfolio. Sharpe (1964), Lintner (1965) and Black (1972) used the models of Markowitz (1952) and Tobin (1958) as starting point to create a capital asset-pricing model (CAPM). This model is based on two essential key assumptions. First, all investors should have the same expectations about returns, risks and distributions of each asset in the market and second, as well lending as borrowing should be available for each investor against the risk-free rate. From this view, all investors should hold the tangency portfolio disregarded their desired risk level. According to the CAPM, asset returns contain two components which is the risk-free rate and a component regarding the systematic risk the portfolio is exposed to. A component for idiosyncratic risk is excluded from the CAPM since this can be diversified away as shown in Appendix 1. The general idea of the CAPM is that investors have to be 3

compensated for the time value of money, which is the risk-free rate (rf), plus a risk premium as compensation for systematic risk. This risk premium is expressed by the market sensitivity (β) times the market risk premium (Rm-rf). The market risk premium is the difference between the market return and the risk-free rate. Hence, if beta is equal to one, the expected return is equal to the expected return of the market. Or in algebraic terms: ( )

(

)

Where, β is the covariance between the return of the asset and the return of the market divided by the variance of the return of the market: (

) (

)

In conclusion, the expected return of the asset increases when the sensitivity of the asset to the market increases as well. A graph can be drawn with a security market line (SML) which presents all betas and their corresponding expected returns. A security is undervalued when it is positioned above the SML and overvalued when the security is under the SML. The theory of the CAPM and its associated SML is graphically presented in Appendix 3. However, the CAPM neglects to show its strength in practice. To test the strength, academics base their researches on the efficient market hypothesis (EMH) of Fama (1970). The EMH includes a market where prices always fully reflect available information. If a market is efficient, it provides informative signals for investors about the value of assets. The EMH contains three degrees of relevant information subsets: (1) the weak form when prices are based on historical prices; (2) the semi-strong form when prices reflect all publicly available information; and (3) the strong form when prices contain all information also from investors or groups of investors who have monopolistic access to relevant information. The theoretical justification of the efficient market hypothesis is based on three principle ideas: (1) all investors are fully rational; (2) some investors are less than fully rational, but their effect cancels out in the aggregate; and (3) some investors are non-rational in similar, correlated ways, however, rational arbitrageurs eliminate their influences on prices. In other words, market prices are always right and the strength of a model depends on the explanatory power of the model regarding to the prices of the securities in the market. Most important criticism about the CAPM is that the underlying assumptions of the model, based on the EMH, are too simplified and unrealistic (Black, 1972). Furthermore, other academics (Merton, 1973; Basu, 1977, 1983; Banz, 1981; Bhandari, 1988) argue that the CAPM should be a multifactor model containing state variables and firm specific characteristics. Similarly, the results of Fama and MacBeth (1973), Fama and French (1992), Black, Jensen and Scholes (1972), Chan, Hamao and Lakonishok (1990) are confirming those empirical findings. Merton (1973) was the first academic who questioned the single-dimension approach of the CAPM and argued that the CAPM is a multifactor linear model with wealth and state variables, called

4

the ICAPM. The ICAPM takes into account that investors are hedging against shortfalls in consumption or against changes in future investment opportunity set. Banz (1981) found that, on average, smaller firms have higher returns than larger firms which indicates a negative relationship between expected return and firm size. With this size effect, Banz confirms that the CAPM is a multifactor model. However, the effect is non-linear since the size effect is strongest for the smallest firms and fades for average and large firms. Furthermore, it is also not clear if the size effect is a proxy for systematic risk or more true unknown factors correlated with size. In addition, also other academics (Reinganum, 1981; Blume & Stambaugh, 1983; Brown, Kleidon & Marsh, 1983; Chan et al., 1991; Fama & French, 1992) are confirming the size effect. Nowadays, no conclusively explanation for the size effect is provided, Amihud and Mendelson (1986) and Liu (2006) devote the size effect to an illiquidity premium which means that smaller stocks are more illiquid and so require a higher expected return for investors. Other investors (Banz, 1981; Zhang, 2006) argue the future performance of smaller firms are harder to predict coherent with a lower supply of corporate information. Basu (1977) finds that Price-to-Earnings ratio, due to exaggerated investor expectations, are indicators of future investment performance. The low Price-to-Earnings portfolios have, on average, higher returns than the high Price-to-Earnings portfolios. As a consequence, Basu argues that publicly available Price-to-Earnings ratios seem to have an information content since according to the efficient market hypothesis all asset prices fully reflect available information in a rapid and unbiased way. Stattman (1980) built further on the findings of Basu a found evidence for a value effect as well, however, his theory was based on the B/M-equity ratio of the firm. He concludes that high B/M-equity firms (value stocks) are realizing a higher expected return than low B/M-equity firms (growth stocks). Rosenberg, Reid and Lanstein (1985) and Chan, Hamao and Lakonishok (1991) showed similar evidence of the persistence of the value effect on respectively the US and Japanese stock markets. Other academics state that the value effect finds its origin in exogenous macroeconomic factors since value stocks are dealing worse with economic downturns or negative external shocks. As a result, including value stocks in a portfolio increases the risk of the portfolio since the performance is poorer during economic downturns in contrast to growth stocks. Because of this additional risk, the investor requires a higher expected return, the difference in expected return between value stocks and growth stocks is the value premium. (Petkova & Zhang, 2003). Bhandari (1988) states that expected returns have a positive relation with the Debt-to-Equity ratio, also after controlling for market sensitivity and firm size. Therefore, Debt-to-Equity ratio is an additional variable to explain expected returns and no proxy for systematic risk. Until the early 1990’s, the value and size effect was only used to indicate that the market beta was not a proper benchmark to explain systematic risk. However, Fama and French (1992) combined the CAPM, size and value effect in a new model (FF3). They showed that market sensitivity seems to have no explanatory value to the average returns, while size and value capture the cross-sectional 5

variation in average stock returns that is related to leverage. The underlying formula for the FF3 model is shown below: ( )

(

)

(

)

(

)

Where SMB is small minus big to incorporate the size premium for the additional risk investors are taking when adding smaller firms to their portfolio over larger firms. HML is high minus low to incorporate the value premium which value stocks are receiving since they are more sensitive to economic downturns compared to growth stocks. The relevance of the SMB and HML factors are subject of discussion. Fergusson and Shockley (2003) claim that SMB and HML serve as a proxy for default risk. According to O. Spalt (personal communication, February, 2014) who suggests that the FF3 model explains a large part of the anomalies, however it just simply defines anomalies away. In other words, it is questionable if SMB and HML are a good proxy for leverage but there is good reason to assume that small and high B/M-equity stocks defaults are correlated. If this cannot be diversified away, then it is systematic risk and commands a premium. Vassalou (2003) designed a model that includes a factor for news related to future Gross Domestic Product (GDP) growth along with the market factor. In his study he shows that he can explain the cross-section of equity returns about as well as the FF3 model can. Furthermore, SMB and HML contain mainly news related items to future GDP, when this factor is included in the FF3 model, SMB and HML lose considerable explanatory power. A fourth factor which is added frequently to asset pricing models is the momentum-factor of Carhart (1997). Until Carhart used the momentum effect as an additional factor in the FF3 model, the momentum effect was an individual theory (Jegadeesh and Titman, 1993) to explain stock price anomalies regarding the CAPM and FF3. By adding the momentum effect in the FF3 model, a new model called Carhart’s 4-factor model (FF4) is created. In this model a factor for market sensitivity, size, value and momentum was included were momentum implies the difference between the average return on the high prior return portfolios minus the average return on the low prior return portfolios. However, since this research is examining the properties of credit spread, and the momentum effect does not include relevant information about credit spread, this factor is ignored. Over the years, there has been a lot of criticism by other academics and analysts on both models mentioned. Kothari, Shanken and Sloan (1995) mention that past B/M-equity results using COMPUSTAT data are affected by a selection bias and provide indirect evidence. As a consequence, the relation between B/M-equity ratios and returns is weaker and less consistent as shown in Fama and French (1992). In addition, Soentjens (2012) argues that the performance of FF3 worsens during economic downturns which might indicate that the effect of SMB and HML lessens during economic downturns or might contain a high degree of noise what lowers the explanatory value of FF3. Another explanation is that other explanatory variables than SMB and HML are becoming more relevant during economic downturns and therefore the relevance of SMB and HML decreases during those periods. As a consequence, researchers are focusing on new properties for default risk to investigate if 6

there is a link between default risk and stock returns during economic downturns. However, those researches are providing conflicting results. Avramov et al. (2012) show that the profitability of anomaly-based trading strategies like price momentum, earnings momentum, credit risk, dispersion, idiosyncratic volatility and capital investment anomalies derives exclusively from periods of financial distress. The dynamics of anomalies can be related to a sharp fall of asset prices during times of financial distress (Hand, Holthausen & Leftwich, 1992; Dichev & Piotroski, 2001). The motivation of Avramov et al. to examine financial distress is well-founded by Fama and French (1993) who argue that the size and value factors proxy for a priced distress factor. Conversely, Campbell, Hilscher, and Szilagyi (2008) argue that distressed firms have high loadings on SMB and HML factors but generate lower returns instead of higher returns as expected. However, Daniel and Titman (1997) argue that the impact on stock returns are dedicated to the size and value characteristics, not SMB and HML factor loadings. As a consequence, Avramov et al. considered credit ratings to be a characteristic of financial distress and focused on this characteristic since it has direct consequences for a firm’s future performance. For instance, financial distress may cause into loss of customers, suppliers, and key employees. Moreover, managerial time is spend on dealing with financial distress instead of focusing on value-enhancing projects. In addition, investment institutions are dealing with regulatory restrictions on the minimum ratings of firms in which they can invest in. Accordingly, financial distress is an ex ante indicator of firm’s future performance. Where credit ratings were used as proxy for financial distress (Avramov et al., 2012), credit spread of individual firms measured from the Merton (1974) model is used by C. O. Kang and H. G. Kang (2009), Vassalou and Xing (2004) and Gharghori, Howard and Robert (2009). C. O. Kang and H. G. Kang examined the effect of credit risk on the return of stocks and find that credit spread captures a systematic risk in the Korean stock market which FF3 and FF4 cannot explain completely. More specifically, they defined the credit factor as the return difference between the portfolios of stocks with high and low credit spreads. Then they tested if this factor is fully explained by FF3 and FF4 and showed that the credit factor generates a statistically significant alpha when it is regressed on FF3 and FF4 which implies that it captures a systematic risk that FF3 and FF4 cannot explain. Vassalou and Xing studied the U.S. equity market and they claim that default risk is priced in equity returns and that the FF3 model is an appropriate alternative for default risk. However, Gharghori et al. are showing contradicting results for the Australian market. And to make it even more confusing, Anginer and Yildizhan (2010) are showing unusually low returns for distressed stocks in the U.S. corporate bond market, in other words, default risk is not priced in equity returns, although distressed stock performed abnormally based on leverage, volatility and profitability. Those conflicting results are arising a new question; whether the pricing of default risk differs across equity markets. Several studies with a different geographical focus are conducted to test the performance of CAPM and FF3. Bauer, Cosemans and Schotman (2010) argues that the explanatory power of FF3 is higher in Europe compared to the US. In addition, they confirm that the size effect which vanished in 7

the US after its discovery, is still present in Europe. Another European oriented study is conducted by Akgul (2013), in this study is the difference between the FF3 model before and after the formation of the EMU investigated. Akgul shows that the FF3 model is significant in eleven out of thirteen countries before the formation of the EMU and seven out of ten countries get even better results after the formation of the EMU. So far, several studies about credit risk are conducted, as well studies which tested assetpricing models with a European dataset. However, no study can be found which used the Merton (1974) model to investigate if credit spread captures a systematic risk in the Dutch, German and French stock market which CAPM and FF3 cannot explain completely.

3. Methodologies The price of a stock reflects the sum of all future dividends payments, discounted back to their present value (Gordon, 1962). A stock price drops to nearly zero when a firm defaults since no more dividends are paid to equity holders, just perhaps an amount which is left after all junior holders are paid. In other words, equity can be interpreted as debt with the last seniority that pays regularly dividends as coupons, hence, equities are subject to credit risk as corporate debts are. There exist various methodologies to obtain data for credit risk related to equities. Avramov et al. (2012) used issuer credit ratings as measurement of credit risk, however, for this research there does not exist enough Dutch, German and French credit ratings to obtain reliable results. As a consequence, in this paper is chosen for the methodology of Merton (1974) which is based on the Black-Scholes-Merton model (Black & Scholes, 1973; Merton, 1973) to measure credit risk at an individual firm level. The Black-Scholes-Merton model argues that derivatives on the firm’s assets is the basic approach for the valuation of stocks and corporate bonds, viewing the firm’s equity as a call option on its assets because equity holders are entitled to the residual value of the firm after all its obligations are paid. Moreover, by using the credit risk of individual firms measured by the Merton model enables to include firms without credit risk or rating information in the analysis so that the analysis is free of sample selection bias. C. O. Kang and H. G. Kang (2009) used this method as well, and their interpretation of the Merton’s model is used to obtain a variable for credit risk called credit spread. Credit spread is the difference in yield between securities that are comparable to each other except for quality rating. At denotes the firm’s asset value at time t. The firm’s asset value is financed by equity (E) and zero-coupon bonds with face value DT maturing at time T. When the firm’s total asset value at maturity AT falls below the amount of debt it has to repay (DT), it is falling into default. Assume the asset value follows a Geometric Brownian motion: (1)

8

Where µA is the expected continuous-compounded return, σA is the annualized asset volatility, and Wt is a Brownian motion. The firm’s asset value and its volatility are not observable. Because equity has limited liability, the value of equity at time T can be denoted as: [

]

(2)

As a consequence, equity is interpreted as a call option on the firm’s asset value with the exercise price equal to the face value of debt maturing at time T. Following Black and Scholes (1973) and Merton (1973), the solution of the current value of equity is: ( (

Where

)

(

)

(3)

)

√ . N is the cumulative standard normal distribution

and



function, and rf is the risk-free interest rate. Because Et is a function of At, it follows from the Itô formula that: (

)

(4)

Then, suppose that the value of equity also follows a Geometric Brownian motion: (5) Where σE is the equity volatility. Jones, Mason and Rosenfeld (1984) show that matching volatility terms in the above two equations gives: (6) Because the hedge ratio is

(

), the solution for equity volatility is: (

)

(7)

By simultaneously solving equations (3) and (7), the current asset value A0 and the asset volatility σA can be obtained from the observable variables E0, σE, DT and T. The current value of debt is (

)

where s is the credit spread of the firm. Consequently, the implied credit

spread is expressed by: (

)

(8)

Where s is denoted by credit spread in this paper. Since credit spread is a function of observable variables such as stock price, equity volatility and risk-free interest rate, firm-specific credit spreads can be calculated. In fact, credit spread should equal the probability of default times the expected loss (Pu, Wang & Wu., 2011). In other words, if the probability of default increases, the credit spread will increase ceteris paribus. The probability of default can be defined as the ability of a firm to repay its debt obligations which depends on both systematic as idiosyncratic risk factors. An example of a systematic risk factor is the overall state of the economy, in contrast to idiosyncratic risk like capital structure which is firm-specific.

9

4. Data Starting from January 4th, 1999, all share prices within the Eurozone are quoted in Euro and this is one of the main reasons that the number of shares and share prices from the Dutch, German and French market are all downloaded from Compustat Global – Security Daily from that day on. Another important reason is the necessity to diminish the amount of firms in the sample because of the timeconsuming creation of volatility over the last 252 trading days. However, at least 336 firms (48 portfolios times an average portfolio size of 7 firms) for each time-period is maintained, consequently, the final dataset starts at December 2004, the end of the dataset is set on the end of the Sovereign Debt Crises in 2012. Regarding to Appendix 4, the total number of firms included in the dataset is 1030 with an average of 532 firms per month. In the beginning of the sample, December 2004, there are 435 firms included in the sample. In December 2012, at the end of the sample, 630 firms are included. The minimum number of firms is 368, recorded in February 2009 and the maximum number of firms is 648, recorded in December 2011. For each time-period, The Netherlands are represented with a share varying between 1.1%-1.9%, France with a share varying between 42.3%-49.8% and Germany with a share varying between 49.0%-56.0%. Consequently, the contribution of The Netherlands is small and the contribution of France and Germany are approximately equal to each other. Furthermore, the sample contains only non-financial firms since high leverage is normal for financial firms. As a consequent, the degree of leverage between financial firms and non-financial firms cannot be compared with each other since high leverage for non-financials firms normally indicates financial distress (Fama and French, 1992). Before performing the FF3 analysis, two main variables are generated which is the Market Value of Equity (ME) and the Book Value of Equity (BE). ME is generated by multiplying the share prices with the corresponding number of stocks outstanding and BE is generated by Total Book Value of the Firm (BF) minus Total Book Value of Liabilities (BL) plus Balance Sheet Deferred Taxes and Investment Tax Credit (BTX) minus the Book Value of Preferred Stock (BPS) (Fama and French, 1993). All missing values of the variables BTX and BPS are replaced by zero since the data is probably not missing but equal to zero. BF, BL, BTX and BPS are gathered from Compustat Global – Fundamentals Quarterly which only contains quarterly data. Since monthly data is used in this research, the assumption is made that for the variables BE and BL one third of the difference between the two quarters can be added to the next month and two third of the difference can be added to the second next month. (C. O. Kang and H. G. Kang, 2009). To generate the variable Market Value of the Firm (MF), some important assumptions are made. The time horizon is considered to be equal to one year. The risk-free rate used is equal to the one month offer rate of EURIBOR expressed in an annual rate. The equity volatility is the daily volatility over the returns of the last 252 trading days (based on at least 100 daily returns) expressed in an annual rate. Furthermore, as an approximation, the asset volatility is assumed to be the same as the equity volatility. The current MF can be obtained by solving equation 3 or 7, according to C. O. Kang 10

and H. G. Kang (2009), the results of the two equations are similar and confirming the validity of our approximation about volatility. To perform the factor analysis, 18 mimicking portfolios are created by separating ME on its 50% percentile and both BE/ME and credit spread by their 30% and 70% percentiles (Fama and French, 1993). Since BE/ME has a stronger explanatory value for expected returns compared to size, the decision is made to separate BE/ME in three groups and size in two groups (Fama and French, 1992). Following another research of Avramov et al., (2012), it is decided to separate credit spread in three groups as well. In this research is chosen to create percentiles based on the dataset in contrast to Fama and French which created their percentiles based on the NYSE. The portfolios are constructed every year in July of year t to June of t+1. Then, the monthly value-weighted returns of the 18 portfolios are calculated. Afterwards, SMB is estimated for each month by subtracting the simple average return of the nine big ME portfolios form the simple average return of the nine small ME portfolios. HML and HSMLS are estimated in almost the same way. HML is constructed by subtracting the simple average return of the six lowest BE/ME portfolios from the simple average return of the six highest BE/ME portfolios. HSMLS is estimated by subtracting the simple average of the six lowest credit spread portfolios from the simple average of the six highest credit spread portfolios. Finally, the dependent variable called excess market return is calculated by market return minus the one month offer rate of EURIBOR. For each period, the market return is based on the valueweighted contribution of three stock market returns: the Dutch, French and German stock market returns. The market return of the Netherlands is represented by the AEX All Share Index including 117 firms, France is represented by the CAC All Tradable which includes 326 firms and the CDAX General Index represents Germany with 482 firms. Excess returns on 48 portfolios formed on ME, BE/ME and credit spread are used to explain the returns of stocks to determine if the mimicking portfolios SMB, HML and HSMLS capture a risk premium in stock returns related to size, B/M-equity and credit spread. Size, B/M-equity and credit spread are sorted on June of each year t. To sort size, ME is measured at the end of June. For the B/Mequity sort, ME is calculated at the end of December of year t-1 and BE is taken form the fiscal year ending in year t-1. Credit spread is also taken from the end of December of year t-1. From July of year t to June of year t+1, 48 portfolios are constructed on three equal percentiles for size and B/M-equity and on two percentiles for credit spread, 30% and 70%. To analyze if the HSMLS premium strengthened during economic downturns, two timespans are formed which are: (1) December 2004 to December 2012 and (2) July 2008 to December 2012. Where the first timespan is the whole final dataset, the second timespan is based on the start of the Subprime Mortgage Crises in July 2008 till the end of the Sovereign Debt Crises in December 2012. Officially, the Subprime Mortgage Crises started a bit earlier but portfolios are always reallocated in June. 11

5. Empirical Results In this section, the statistics and empirical tests regarding to the extension of the FF3 model with credit spread is described. As pronounced before, the purpose of this research is to investigate whether credit spread has a premium and if this premium becomes more relevant during economic downturns. To investigate this, the Fama and MacBeth (1973) regression procedure is used. However, at first the data between December 2004 and December 2012 is described in Table 1, 2 and 3. Table 1: Descriptive statistics of the magnitude of the dataset for 48 stock portfolios formed on size, B/M-equity ratio and credit spread between December 2004 and December 2012. Average market value per portfolio Average number of firms per (in percent) portfolio 40.88 Low credit spread 30.67 Low credit spread B/M B/M 1 2 3 4 1 2 3 4 1 0.01 0.01 0.01 0.01 1 6.3 7.3 7.4 3.6 Size 2 0.11 0.11 0.05 0.02 Size 2 21.8 17.7 9.8 4.4 3 0.69 0.56 0.40 0.08 3 29.4 21.1 14.5 3.6 4 19.17 13.53 6.31 0.57 4 37.0 25.0 11.9 1.7 Medium credit spread B/M 1 2 1 0.02 0.02 Size 2 0.09 0.10 3 0.32 0.65 4 6.93 14.90 High credit spread B/M 1 2 1 0.03 0.02 Size 2 0.04 0.05 3 0.11 0.12 4 1.93 2.41

3 0.04 0.17 0.66 13.74

46.16 Medium credit spread B/M 4 1 2 0.03 1 11.8 11.8 0.12 Size 2 14.4 15.8 0.36 3 12.9 22.3 8.87 4 14.2 29.0

3 0.03 0.07 0.25 4.23

12.97 High credit spread B/M 4 1 2 0.07 1 18.2 13.2 0.17 Size 2 8.0 7.4 0.44 3 5.0 5.7 3.24 4 2.0 4.7

39.92 3 19.4 23.4 26.7 21.6

4 15.5 19.2 15.2 16.4 29.41

3 19.1 13.0 8.6 10.6

4 42.4 27.9 17.3 10.3

Table 1 shows that most firms with a lower credit spread have a big size and a low B/M-equity ratio. Firms with a medium credit spread have a medium size and B/M-equity ratio while most firms with a high credit spread are allocated to small size percentile and high B/M-equity percentile. Those results are in line with the expectations since it is proven before that bigger (smaller) firms with a higher (lower) B/M-equity ratio are more (less) risky. And if size and B/M-equity ratio are indeed an approximation for default risk, than most firms should be allocated to the big size and low B/M-equity ratio for low credit spread and to the small size and high B/M-equity ratio for high credit spread. Table 1 also shows that the average market value per portfolio is lowest for the small size portfolio and highest for the big size portfolios. Which is remarkable is that the portfolios with the highest credit spread account for just 12.97% of the dataset while the total number of firms for those 12

portfolios is 29.41% of the dataset. This confirms again the expectation that if size is an approximation for default risk, size has to be small for high credit spread. In addition, also the value-weighted share of high B/M-equity ratios is increasing when credit spread becomes higher. Table 2 shows the descriptive statics of the independent variables constructed by size, B/Mequity ratio and credit spread. Since size is divided in four percentiles, the first percentile has the lowest average firm size and the fourth percentile has the highest firm size. The same works for B/Mequity ratio. Credit spread is divided into three percentiles, consequently, the first percentile contains the lowest credit spread and the third percentiles contains the highest credit spreads. Looking to this summary, it can be concluded that the higher credit spread, the smaller the size except for the lowest B/M-equity percentile. In addition, the higher credit spread, the higher B/M-equity. And especially for the highest credit spreads: the higher B/M-equity, the higher credit spread. Table 2: Descriptive statistics of the independent variables (size, B/M-equity ratio and credit spread) for 48 stock portfolios formed on size, B/M-equity ratio and credit spread between December 2004 and December 2012. Average firm size per portfolio Low credit spread B/M 1 2 1 20 23 Size 2 62 62 3 231 244 4 4535 4808 Medium credit spread B/M 1 2 1 18 22 Size 2 62 62 3 248 255 4 4033 4751 High credit spread B/M 1 2 1 18 18 Size 2 63 59 3 230 272 4 6041 5189

3 20 56 263 4920

3 20 63 231 5333

3 19 57 274 4202

4 22 52 245 1643

Average B/M-equity ratio per Average credit spread per portfolio portfolio Low credit spread Low credit spread B/M B/M 1 2 3 4 1 2 3 4 1 0.33 0.69 0.95 1.24 1 0.006 0.011 0.007 0.007 Size 2 0.36 0.61 0.81 1.33 Size 2 0.007 0.006 0.002 0.008 3 0.32 0.58 0.82 1.07 3 0.002 0.003 0.008 0.002 4 0.32 0.54 0.77 1.07 4 0.003 0.003 0.005 0.002

4 17 56 209 4500

Medium credit spread B/M 1 2 1 0.34 0.61 Size 2 0.32 0.57 3 0.37 0.62 4 0.37 0.62

4 18 56 246 2962

High credit spread B/M 1 2 1 0.25 0.53 Size 2 0.31 0.63 3 0.28 0.62 4 0.32 0.62

3 0.86 0.89 0.92 0.84

3 0.90 0.93 0.79 1.04

4 1.53 1.40 1.31 1.27

Medium credit spread B/M 1 2 1 0.036 0.028 Size 2 0.019 0.018 3 0.019 0.018 4 0.016 0.024

3 0.033 0.024 0.026 0.022

4 0.042 0.026 0.026 0.020

4 1.58 1.66 1.68 1.89

High credit spread B/M 1 2 1 0.166 0.127 Size 2 0.066 0.094 3 0.050 0.098 4 0.056 0.074

3 0.122 0.107 0.094 0.137

4 0.168 0.122 0.107 0.113

Table 3 presents the means, volatilities and t-statistics for means of the dependent variable, which are the excess returns of the portfolios between December 2004 and December 2012. Regarding to the t-statistics, the means are not convincingly confirming the relationship as seen in Table 1 and 2. In summary, the means of portfolios 6, 8, 13, 14, 16, 23, 24, 25, 26, 28, 29, 35, 37, 44, 45 and 48 are positively and significantly different from zero at the 10% level. The means of portfolios 11, 12, 19, 13

20, 27, 32, 38, 40, 42 and 46 are positive and significantly different from zero at the 5%. At the 1% level are the means of portfolios 22 and 36 positively and significantly different from zero. Overall, the data is Table 3 are not convincing enough to draw any conclusions about the relationship between size, B/M-equity ratio, credit spread and excess returns. Table 3: Descriptive statistics of the dependent variable (excess return) for 48 stock portfolios formed on size, B/M-equity ratio and credit spread between December 2004 and December 2012. T-statistics with a *, ** and *** are respectively significant at the 10%, 5% and 1% level. The t-tests are based on 96 degrees of freedom. Means (in percent) Low credit spread B/M 1 2 1 -0.18 0.37 Size 2 0.61 0.87 3 0.34 0.44 4 0.62 0.60 Medium credit spread B/M 1 2 1 -0.59 0.64 Size 2 0.52 1.48 3 0.93 1.03 4 0.76 0.70 High credit spread B/M 1 2 1 0.69 0.73 Size 2 1.36 1.20 3 0.43 2.25 4 1.67 2.09

3 0.41 0.75 0.95 0.42

3 1.05 0.79 1.04 0.60

3 0.77 0.50 0.86 0.19

4 0.59 1.03 1.49 1.22

Volatilities (in percent) Low credit spread B/M 1 2 1 5.3 6.8 Size 2 5.0 5.3 3 5.1 5.1 4 4.3 4.5

4 1.44 0.86 0.75 1.27

Medium credit spread B/M 1 2 1 7.7 6.2 Size 2 5.8 5.6 3 6.4 6.1 4 5.7 5.9

4 1.60 1.34 1.11 1.09

High credit spread B/M 1 2 1 5.9 5.8 Size 2 8.7 6.3 3 8.8 10.3 4 8.7 11.1

T-statistics for means

3 6.7 5.9 5.2 6.1

3 5.6 4.8 5.6 6.1

3 5.8 6.4 8.1 9.4

4 8.8 6.5 7.9 8.2

Low credit spread B/M 1 2 1 -0.33 0.53 Size 2 1.20 1.62* 3 0.67 0.83 * 4 1.40 1.32*

3 0.60 1.26 1.81** 0.68

4 0.61 1.56* 1.85** 1.43*

4 6.2 5.6 5.7 5.5

Medium credit spread B/M 1 2 1 -0.75 1.02 Size 2 0.89 2.58*** 3 1.43* 1.64* 4 1.30* 1.17

3 1.84** 1.63* 1.83** 0.97

4 2.27** 1.52* 1.30* 2.29**

4 5.1 5.7 6.8 7.9

High credit spread B/M 1 2 1 1.17 1.24 Size 2 1.51* 1.87** 3 0.46 2.16** 4 1.30* 1.73**

3 4 1.31* 3.09*** 0.77 2.33** 1.04 1.59* 0.19 1.35*

However, when comparing the means of big size/low credit spread to small size/high credit spread, it can be seen that the means for small size/high credit spread are higher. In addition the means of high credit spread/high B/M-equity ratio are higher compared to the ones of low credit spread/low B/M-equity ratio. The methodology of Fama and Macbeth (1973) is used to measure the impact of the market premium, SMB, HML and HSMLS against the excess portfolio returns. In the first pass, each portfolio receives a beta value for those factors by using the following time series regression: (

)

Where ri-rf is the excess return of portfolio i in month t, rm-rf is the market risk premium. Then, β1 is the sensitivity of the portfolios excess return to the market risk premium, β s is the sensitivity of the portfolio’s excess return to the factor SMB, βv is the sensitivity of the portfolio’s excess return to the

14

factor HML and βc is the sensitivity of the portfolio’s excess return to the factor HSMLS. α is a constant term and ε is an error term. In the second pass, the beta values gathered are used to run t cross sectional regressions: ̂

̂

̂

Then the average of each lambda is calculated by ̂



̂ ̂ . To test whether the lambdas are

significant, the respective t-statistics for each lambda is conducted with a t-test (one-sample meancomparison test with a hypothesized mean of 0). Table 4: Fama and MacBeth (1973) analysis of the dependent variable (excess return) for 48 stock portfolios formed on size, B/M-equity ratio and credit spread. Three models are chosen to examine if adding SMB, HML and HSMLS improves calculation of excess returns. Three datasets are chosen to investigate if the results differ in time; (1) December 2004 to December 2012, and (2) July 2008 to December 2012. T-statistics with a * and ** are respectively significant at the 10% and 5% level. The t-tests are based on 96 degrees of freedom for the first dataset and 53 degrees of freedom for the second dataset.

Alpha (in percent)

Market Premium (in percent)

Dec. 2004-Dec. 2012 CAPM

0.67 1.49*

0.24 0.37

Dec. 2004-Dec. 2012 FF3

0.32 0.82

Dec. 2004-Dec. 2012 FF3+HSMLS

SMB (in percent)

HML (in percent)

0.53 0.79

0.12 0.41

0.48 1.92**

0.70 1.67**

0.00 0.00

0.05 0.17

0.53 2.06**

July 2008-Dec. 2012 CAPM

0.79 1.73**

-0.04 -0.04

July 2008-Dec. 2012 FF3

0.54 1.45*

0.17 0.19

0.18 0.41

0.36 1.01

July 2008-Dec. 2012 FF3+HSMLS

0.73 1.74**

-0.14 -0.16

0.14 0.33

0.45 1.25

HSMLS (in percent)

0.15 0.42

-0.09 -0.18

Regarding to Table 4, three models are investigated, the CAPM model, FF3 and FF3 in combinations with a credit spread premium to investigate if adding SMB, HML and HSMLS improves the models. Besides, two time periods are used to examine if models differ across time. Looking to the premiums, it can be concluded that there is no significant effect for the market premium, size premium and the credit risk premium. The value premiums have a positive and significant effect at the 5% level during December 2004 to December 2012 for the FF3 model and FF3 in combination with HSMLS. The value premiums are respectively equal to 0.48% and 0.53%. In other words, firms with high B/M equity ratios are more risky (positive premium) and hence their discount rates will be higher which results in a lower value (price) today. Investors who are bearing higher risks by holding high B/M equity ratios, want to be compensated for it and require higher expected returns. Alpha is a good measurement to investigate the performance of the three models and time periods used. The closer alpha is to zero, the better the performance of the model. The alphas of the 15

CAPM, FF3 and FF3 in combination with the credit spread premium are respectively equal to 0.67%, 0.32% and 0.70% for December 2004 to December 2012. Consequently, the FF3 model has the best performance. Furthermore, the alpha of the FF3 model is not significantly different from zero. For the period July 2008 to December 2012, the alphas of the CAPM, FF3 and FF3 in combination with the credit spread premium are respectively equal to 0.79%, 0.54% and 0.73%. In other words, the FF3 model has again the best performance. Looking to the time-periods, the alphas of the models based on December 2004 to December 2012 are outperforming the alphas of the same models in the period July 2008 to December 2012. In conclusion, the FF3 model in the period December 2004 to December 2012 has the best performance.

6. Conclusion In this research is investigated whether credit spread captures systematic risk which cannot be fully explained by CAPM or FF3. In addition, it is investigated if this premium changes in time. All Dutch, French and German firms listed on respectively the AEX All Share Index, CAC All Tradable and CDAX General Index between December 2004 and December 2012 are included in the dataset. The Fama and Macbeth (1973) analysis is used for the regressions, based on 48 portfolios formed by size, book-to-market-equity ratio and credit spread. The Merton (1974) model is used to create credit spread as characteristic for credit risk. In this research is not proven that credit spread captures systematic risk which cannot be explained by CAPM or FF3. This is not in line with the findings of Avramov et al. (2012), C. O. Kang and H. G. Kang (2009) and Vassalou and Xing (2004) who are showing a positive relationship between excess returns and credit risk. Even more, this research did not prove a stronger credit risk premium during economic downturns. In conclusion, in this research is no significant relationship between credit risk and excess returns. Looking to the other premiums, it can be concluded that the market premium and size premium have also no significant explanatory value to the excess returns which is contrary to the research of Fama and French (1992). However, this result is in line with Cosemans and Schotman (2010) who argue that the size effect vanished in the US after its discovery although the showed that the size effect is still present in Europe. The value premium is present in the FF3 model and the FF3 model in combination with the credit spread premium from December 2004 to December 2012 and is significant at the 5% level. At last has to be mentioned that this research is based on some limitations and assumptions. One important limitation is the necessity to diminish the amount of firms in the sample because of the time-consuming creation of volatility over the lasts 252 trading days. As a consequence, too few firms are allocated to the portfolios, and hence, the diversification is not enough to avoid a high degree of noise. As a consequence, the results in Table 3 and 4 have low explanatory value. Second, The Netherlands is represented with a share varying between 1.1%-1.9% each period which is very low 16

compared to the shares of France (42.3%-48.9%) and Germany (49.8%-56.0%). However, the results of this research is still representative for The Netherlands since the economy of the Netherland depends heavily on the French and German economies. And third, some important assumptions are made to generate the market value of the firm since the time horizon is set equal to one year, the riskfree rate used is equal to the one month offer rate of EURIBOR and the equity volatility is assumed to be equal to the asset volatility.

17

References Akgul, B. (2013). Fama-French three factor regression on European stock markets: Before and after EMU. Working paper, University of Tilburg. Amihud, Y., & Mendelson, H. (1986). Asset pricing and the bid-ask spread. Journal of Financial Economics. 17, 223-249. Anginer, D., & Yildizhan, C. (2010). Is there a distress risk anomaly? Corporate bond spread as a proxy for default risk. Working paper, The World Bank Development Research Group. Avramov, D., Chordian, T., Jostova, G., & Philipov, A. (2012). Anomalies and financial distress. Journal of Financial Economics. 108, 139-159. Banz, R.W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics. 9, 3-18. Basu, S. (1977). Investment performance of common stocks in relation to their price-earnings ratios: A test of the efficient market hypothesis. The Journal of Finance. 32, 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, 129-156. Bauer, R., Cosemans, M., & Schotman, P.C. (2010). Conditional asset pricing and stock market anomalies in Europe. European Financial management. 16, 165-190. Bhandari, L.C. (1988). Debt/equity ratio and expected common stock returns: Empirical evidence. Journal of Finance. 43, 507-28. Black, F. (1972). Capital market equilibrium with restricted borrowing. The Journal of Business. 45, 444-455. Black, F., Jenssen, M.C., & Scholes, M. (1972). The capital asset pricing model: Some empirical tests. Studies in the Theory of Capital Markets, edited by Michael Jensen. New York: Praeger. Black, F., & Scholes, M. (1973). The pricing of options and corporate liabilities. Journal of Political Economy. Blume, M.E., & Stambaugh, R.F. (1983). Biases in computed returns: An application to the size effect. Journal of Financial Economics. 12, 387-404. Brown, P., Kleidon, A.W., & Marsh, T.A. (1983) New evidence on the nature of size-related anomalies in stock prices. Journal of Financial Economics. 12, 33-56. Campbell, J.Y., Hilscher, J., & Szilagyi, J. (2008). In search of distress risk. Journal of Finance. 63, 2899-2939. Carhart, M.M. (1997). On persistence of mutual fund performance. Journal of Finance. 52, 57-82. Centraal Bureau voor de Statistiek – StatLine (2012). Internationale handel; in- en uitvoer volgens Geharmoniseerd Systeem. Retrieved from: http://statline.cbs.nl/StatWeb/publication/?DM=S LNL&PA= 37952& D1=1&D2=0&D3=3,5,13&D4=l&HDR=T,G2&STB=G1,G3&VW=T Chan, L., Hamao, Y., & Lakonishok, J. (1991). Fundamentals and stock returns in Japan. Journal of Finance. 46, 1739-1789. 18

Daniel, K.D., & Titman, S. (1997). Evidence on the characteristics of cross sectional variation in stock returns. Journal of Finance. 52, 1-33. Dichev, I.D., & Piotroski, J.D. (2001). The long-run stock returns following bond rating changes. Journal of Finance. 56, 55-84. Fama, E.F. (1970). Efficient capital markets: A review of theory and empirical work. The Journal of Finance. 25, 383-417. Fama, E.F., & French, K.R. (1992). The cross-section of expected stock returns. Journal of Finance. 47, 427-465. Fama, E.F., & French, K.R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics. 33, 3-56. Fama, E.F., & MacBeth, J.D. (1973). Risk, return, and equilibrium: Empirical tests. Journal of Political Economy. 81, 607-636. Fergusson, M., & Shockley, R. (2003). Equilibrium Anomalies. Journal of Finance. 85, 2549-2580. Gharghori, L., Howard, C., & Robert, F. (2009). Default risk and equity returns: Australian evidence. Pacific-Basin Finance Journal. 17, 580-593. Gordon, M.J. (1962). The investment, financing, and valuation of the corporation. Hand, J.R.M., Holthausen, R.W., & Leftwich, R.W. (1992). The effect of bond rating agency announcements on bond and stock prices. The Journal of Finance. 47, 733-752. Jegadeesh, N. & Titman, S. (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance. 48, 65-91. Jones, P.E., Mason, S.P., & Rosenfeld, E. (1984). Contingent claims analysis of corporate capital structures: An empirical analysis. Journal of Finance. 39, 611–625. Kang, C.O., & Kang, H.G. (2009). The effect of credit risk on stock returns. Journal of Economic Research. 14, 49-67. Kothari, S.P., Shanken, J., & Sloan, R.G. (1995). Another look at the cross-section of expected stock returns. Journal of Finance. 50, 185-224. 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, 13-37. Liu, W. (2006). A liquidity-augmented capital asset pricing model. Journal of Economics. 82, 631-671. Markowitz, H. (1952). Portfolio selection. The Journal of Finance. 7, 77-91. Merton, R.C. (1973). An intertemporal capital asset pricing model. Econometrica. 41, 867-887. Merton, R.C. (1974). On the pricing of corporate debt: The risk structure of interest rates. The Journal of Finance. 29, 449-470. Petkova, R., & Zhang, L. (2003). Is value riskier than growth? Journal of Financial Economics. 78, 182-202.

19

Pu, W., Wang, J., & Wu, C. (2011). Are liquidity and counterparty risk priced in the credit default swap market? The Journal of Fixed Income. 20, 59-79. Reinganum, M.R. (1981). A new empirical perspective on the CAPM. The Journal of Financial and Quantitative Analysis. 16, 439-462. Rosenberg, B., Reid, K., & Lanstein., R. (1985). Persuasive evidence of market inefficiency. The Journal of Portfolio Management. 11, 9-16. Sharpe, W.F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance. 19, 425-442. Soentjens, T. (2012). Explaining stock returns: CAPM, Fama & French 3-factor model and Carhart’s 4-factor model. University of Tilburg. Stattman, D. (1980). Book values and stock returns. The Chicago MBA: A Journal of Selected Papers. 4, 25-45. Tobin, J. (1958). Liquidity preference as behavior towards risk. Review of Economic Studies. 25, 65-86. Vassalou, M. (2003). New related to future GDP growth as a risk factor in equity returns. Journal of Financial Economics. 68, 47-73. Vassalou, M., & Yuhang, X. (2004). Default risk in equity returns. The Journal of Finance. 59, 831-868. Zhang, X.F. (2006). Information uncertainty and stock returns. The Journal of Finance. 61, 105-137.

20

Appendix 1: Systematic vs. Idiosyncratic Risk

Appendix 2: Modern Portfolio Theory and Separation Theorem

21

Appendix 3: Capital Asset Pricing Model

22

Appendix 4: Number of firms in each time-period Month December, 2004

The Netherlands 8 1,8%

France 198 45,5%

Germany Total 229 52,6% 435

January, 2005

7

1,8%

172

43,9%

213

54,3%

392

February, 2005

7

1,8%

176

44,0%

217

54,3%

400

March, 2005

7

1,6%

199

46,6%

221

51,8%

427

April, 2005

7

1,9%

180

47,9%

189

50,3%

376

May, 2005

7

1,9%

173

46,8%

190

51,4%

370

June, 2005

8

1,8%

211

47,2%

228

51,0%

447

July, 2005

8

1,6%

211

43,1%

271

55,3%

490

August, 2005

8

1,7%

209

43,4%

265

55,0%

482

September, 2005

8

1,7%

206

43,0%

265

55,3%

479

October, 2005

8

1,7%

210

43,8%

261

54,5%

479

November, 2005

8

1,6%

214

43,9%

266

54,5%

488

December, 2005

8

1,6%

230

45,5%

268

53,0%

506

January, 2006

8

1,7%

207

43,3%

263

55,0%

478

February, 2006

8

1,7%

205

43,4%

259

54,9%

472

March, 2006

8

1,7%

200

42,3%

265

56,0%

473

April, 2006

8

1,8%

191

44,0%

235

54,1%

434

May, 2006

7

1,6%

192

43,6%

241

54,8%

440

June, 2006

8

1,7%

206

43,6%

258

54,7%

472

July, 2006

8

1,5%

255

47,0%

279

51,5%

542

August, 2006

8

1,5%

258

47,5%

277

51,0%

543

September, 2006

7

1,4%

235

48,5%

243

50,1%

485

October, 2006

7

1,5%

230

47,8%

244

50,7%

481

November, 2006

8

1,5%

255

47,5%

274

51,0%

537

December, 2006

8

1,5%

259

47,8%

275

50,7%

542

January, 2007

8

1,5%

250

47,1%

273

51,4%

531

February, 2007

8

1,5%

256

47,9%

270

50,6%

534

March, 2007

8

1,7%

228

47,7%

242

50,6%

478

April, 2007

7

1,5%

224

48,5%

231

50,0%

462

May, 2007

8

1,6%

244

47,7%

259

50,7%

511

June, 2007

7

1,4%

238

48,6%

245

50,0%

490

July, 2007

7

1,2%

278

48,9%

283

49,8%

568

August, 2007

7

1,2%

276

48,5%

286

50,3%

569

September, 2007

7

1,3%

266

48,4%

277

50,4%

550

October, 2007

7

1,3%

267

48,3%

279

50,5%

553

November, 2007

7

1,3%

267

48,3%

279

50,5%

553

December, 2007

7

1,3%

271

48,5%

281

50,3%

559

23

Month January, 2008

The Netherlands 7 1,3%

France 249 47,8%

Germany Total 265 50,9% 521

February, 2008

7

1,3%

254

48,4%

264

50,3%

525

March, 2008

7

1,3%

254

48,8%

260

49,9%

521

April, 2008

7

1,4%

239

47,2%

260

51,4%

506

May, 2008

6

1,3%

203

45,4%

238

53,2%

447

June, 2008

7

1,5%

206

45,1%

244

53,4%

457

July, 2008

7

1,3%

248

45,0%

296

53,7%

551

August, 2008

7

1,3%

237

44,1%

293

54,6%

537

September, 2008

7

1,4%

225

44,9%

269

53,7%

501

October, 2008

5

1,1%

210

46,6%

236

52,3%

451

November, 2008

7

1,4%

222

45,0%

264

53,5%

493

December, 2008

7

1,4%

234

46,1%

267

52,6%

508

January, 2009

7

1,7%

183

43,8%

228

54,5%

418

February, 2009

7

1,9%

157

42,7%

204

55,4%

368

March, 2009

7

1,7%

181

43,4%

229

54,9%

417

April, 2009

6

1,4%

199

45,4%

233

53,2%

438

May, 2009

6

1,2%

220

45,2%

261

53,6%

487

June, 2009

7

1,4%

219

44,2%

270

54,4%

496

July, 2009

8

1,4%

264

47,4%

285

51,2%

557

August, 2009

8

1,5%

257

47,5%

276

51,0%

541

September, 2009

8

1,4%

261

46,4%

293

52,1%

562

October, 2009

8

1,6%

239

46,5%

267

51,9%

514

November, 2009

8

1,6%

240

47,0%

263

51,5%

511

December, 2009

7

1,2%

276

47,6%

297

51,2%

580

January, 2010

7

1,3%

254

46,5%

285

52,2%

546

February, 2010

8

1,5%

247

46,3%

279

52,2%

534

March, 2010

7

1,3%

257

47,1%

282

51,6%

546

April, 2010

6

1,1%

254

46,6%

285

52,3%

545

May, 2010

7

1,3%

252

45,9%

290

52,8%

549

June, 2010

7

1,2%

259

46,1%

296

52,7%

562

July, 2010

7

1,3%

257

46,8%

285

51,9%

549

August, 2010

7

1,3%

256

47,2%

279

51,5%

542

September, 2010

7

1,1%

299

48,1%

316

50,8%

622

October, 2010

7

1,2%

282

47,1%

310

51,8%

599

November, 2010

7

1,2%

278

47,3%

303

51,5%

588

December, 2010

7

1,1%

303

48,2%

319

50,7%

629

24

Month January, 2011

The Netherlands 7 1,2%

France 293 48,3%

Germany Total 307 50,6% 607

February, 2011

7

1,1%

293

48,0%

310

50,8%

610

March, 2011

7

1,1%

294

48,3%

308

50,6%

609

April, 2011

7

1,2%

288

48,6%

298

50,3%

593

May, 2011

7

1,2%

288

49,0%

293

49,8%

588

June, 2011

7

1,1%

298

48,2%

313

50,6%

618

July, 2011

7

1,1%

318

49,1%

322

49,8%

647

August, 2011

7

1,1%

318

49,4%

319

49,5%

644

September, 2011

7

1,1%

310

49,0%

316

49,9%

633

October, 2011

7

1,1%

310

48,8%

318

50,1%

635

November, 2011

7

1,1%

303

48,2%

318

50,6%

628

December, 2011

7

1,1%

314

48,5%

327

50,5%

648

January, 2012

7

1,1%

297

48,1%

314

50,8%

618

February, 2012

7

1,1%

301

48,5%

312

50,3%

620

March, 2012

7

1,1%

299

48,9%

306

50,0%

612

April, 2012

7

1,2%

289

48,7%

297

50,1%

593

May, 2012

7

1,2%

292

48,0%

309

50,8%

608

June, 2012

7

1,1%

297

48,6%

307

50,2%

611

July, 2012

7

1,1%

313

48,6%

324

50,3%

644

August, 2012

7

1,1%

307

48,9%

314

50,0%

628

September, 2012

7

1,1%

309

48,6%

320

50,3%

636

October, 2012

7

1,1%

310

49,8%

305

49,0%

622

November, 2012

7

1,1%

307

49,4%

307

49,4%

621

December, 2012

7

1,1%

314

49,8%

309

49,0%

630

25

Life Enjoy

" Life is not a problem to be solved but a reality to be experienced! "

Get in touch

Social

© Copyright 2013 - 2019 DOKUMENTIS.COM - All rights reserved.