Академический Документы
Профессиональный Документы
Культура Документы
Introduction 5 1.1 Research Background .........................................................................6 1.2 Research Aim.......................................................................................6 1.3 Research Objective .............................................................................7 1.4 Research Questions ............................................................................8 Literature Review 9 2.1 Prior Evidence of the Four-Factor Model in the UK.......................13 2.2 Hypotheses....................................................................................15 Data and Methodology 17 3.1 Research Data ..............................................................................17 3.2 Research Methodology .................................................................18 Empirical Results and Discussion 21 4.1 Summary Statistics...............................................................21 4.2 Data Analysis........................................................................27 4.2.1 Full Sample Regression ......................................................27 4.2.1.1 Full Sample Analysis ...........................................................28 4.2.1.2 Graph of the Full Sample Analysis ......................................30 4.2.2 Bull Market Regression........................................................34 4.2.2.1 Bull Market Analysis ............................................................35 4.2.3 Bear Market Regression ......................................................38 4.2.3.1 Bear Market Analysis ...........................................................39 4.2.4 Behavior Finance Arguments ..............................................40 Summary and Conclusion 42 5.1 Recommendations.........................................................................43 References Appendices 44 49
Page 1
Table 1
Table 2(a): Excess Returns on the six portfolios (Full Sample) ...............................24 Table 2(b): Excess Returns on the six portfolios (Bear Market) ..............................25 Table 2(c): Excess Returns on the six portfolios (Bull Market) ................................26 Table 3 : Regression on the six portfolios (Full Sample) .......................................27 Table 4 : Regression on the six portfolios (Bull Market) ........................................34 Table 5 : Regression on the six portfolios (Bear Market) ......................................38 Figure 1 : Market Factor ........................................................................................... 30 Figure 2 : Size Factor ................................................................................................ 31 Figure 3 : Book-to-market Factor.............................................................................. 32 Figure 4 : Momentum Factor.................................................................................... 33 Picture 1 : Market Return ...........................................................................................49 Picture 2 : SMB Return
...........................................................................................49
Acknowledgement
MSC Investment Management Coventry University 2009 Page 2
Abstract
This paper, we study the significance of the four-factor asset pricing model (market factor, size factor, book-to-market factor and momentum factor) in explaining the cross-sectional
MSC Investment Management Coventry University 2009 Page 3
1.0 Introduction
In 1970s, (Sharpe 1964), (Lintner 1965) and (Black 1972) introduce The Capital Asset Pricing Model (CAPM) which is the first asset pricing model in finance. This model has becomes popular in explaining the relationship between return and market beta on risky
MSC Investment Management Coventry University 2009 Page 4
Page 7
Momentum factor) show a significant value in explaining the average stock returns in the UK stock market? (Fama and French 1995) state that if stocks are priced rationally, the multi-factor model should able to give explanation the cross sectional variation in stock returns. For example, size and book-to-market become common risk factors which sensitive in returns.
2) Does the Four-Factor model works well in different capital markets (UK) other
market conditions (Bull and Bear market)? In this paper, we would like to organize the structure of contents as follows: Chapter 2 presents a brief of literature review. Chapter 3 shows the data collection and details of methodology employed. Chapter 4 presents the data analysis and discusses research findings. Chapter 5 contains summary and conclusion about the implications of the research and recommendations for future research.
E(Ri) = Rf + i (E(Rm)-Rf)
(1)
E(Ri) is the expected return of security i, Rm is the rate of return on the market portfolio, Rf is the risk free rate, i is the beta coefficient of the i security. However, recent empirical research has brought into question the robustness of the CAPM in explaining the cross-sectional variation in average stock returns. (Fama and French 1992) argue that the critical evidence with regard to the insufficiencies of CAPM, they tested CAPM on the basis of return of assets and observed a non-linear relationship between average return and beta coefficient. The expected rate of return from some investment portfolios which are based on firm characteristics (size, earnings/price, bookto-market ratio, past performance on sales growth, long term and short term stock returns) cannot be explained by the CAPM beta. For example, (Banz 1981) indicates that there is a strong average returns on small firms compare to big firms. (Rosenberg, Reid and Lanstein 1985) finds that average returns on firms with high book-to-market ratio (value stocks) outperform than those with low book-to-market ratio (growth stocks). Their findings are consistent with (Fama and French 1992) empirical works who argue that the portfolio with size (size effect) and book-to-market ratio (book-to-market effect) capture the cross sectional variation in average security returns. (Fama and French 1993) argue that the three-factor model does well in explain the cross section of returns on US stocks. Their model indicates that the excess return on an asset portfolio is explained by: i) market premium the expected return on a market portfolio minus risk free rate; ii) size premium the return on a portfolio of small stock minus the return on a portfolio of big stock; small minus big (SMB) iii) book-to-market premium the average returns on two high book-tomarket stock portfolios minus the average returns on two low book-to-market stock portfolios ; high minus low (HML). (Fama and French 1996) point out that these patterns which are used in average stock returns are not explained by the CAPM, they are typically called anomalies. They find that the three-factor model is capable to capture the size and book-to-market effects, but not momentum effect which remains the greatest challenge to their model. Momentum effect is another variable which been tested by (Jegadeesh and Titman 1993) in empirical work and
MSC Investment Management Coventry University 2009 Page 9
provides a better explanation for the relation between book-to-market and average returns compare to the characteristic-based model; b) the robustness of Fama and French multifactor model capable captures the average returns on US portfolios formed on size, book-to-market and other variables (earnings/price, cashflow/price) which cannot explain by the CAPM; c) the characteristics-based model of Daniel and Titman need to be tested with more extensive data rather than sample period. A recent research evidence by (Stambaugh and Pastor 2000) who make the argument on the robustness of characteristicsbased model versus factor-based model pricing model. They argue that there is no difference between the Daniel and Titman characteristics-based model and the Fama and French risk-based model because both models lead to similar portfolio selections in investment. In addition, (Lakonishok, Vishny and Shleifer 1994) argue that the value premium might be genuine to explain the cross-sectional in average stock returns but irrational. They suggest that investor overreaction causes the value stocks underpriced and the growth stock overpriced. Therefore, investor overreaction makes the value premium exist in the stock market. In other words, investors overreact to the stocks which perform relatively well in the past and buy them. These stocks become overpriced. For the stocks that have performed relatively poorly in the past, investors tend to sell them. Therefore, these stocks become underpriced. (Mackinlay 1995) who debate that the explanation of Fama and French multifactor model towards value premium is caused by data snooping. They make further explanation that the best way to evaluate the data snooping hypothesis is to use different time periods and in different countries. They challenge most of the Fama and French empirical studies on multifactor model are tested in the US market. Their findings indicate that risk factors capture the cross sectional variation in average security returns in the US markets. The vast majority of empirical studies on three-factor model have been conducted using US data. They find that this model does work well in explaining the variation in average stock returns in the US portfolios. (Fama and French 1998) find that the annually book-tomarket premium is 7.68 percent in international markets for the sample period 1975-1995. In the same year, (Arshanapali, Coggin and Doukas 1998) state that the three factor model is not only applicable in the US stock market, but also efficient in most of the international stock markets. Another study which conducted by (Maroney and Protopapadakis 2002) in
MSC Investment Management Coventry University 2009 Page 12
2.2 Hypotheses
Before we start to analyse the multifactor asset pricing anomalies, we want to identify the hypothesis for each risk factor. This research paper is prepared to determine whether three risk factors (size, book-to-market and momentum) can be used to capture much of the cross-sectional variation in average stock returns. Hypothesis 1 According to (Fama and French 1992, 1993, 1995), they indicate that small firms tend to have higher expected returns compare to large firms. This is because small firms are relatively financial distress risk. Small firms tend to receive higher returns as compensation because they need to bear higher risk. If our results are consistent with size effect hypothesis, size premium should generate positive return.
MSC Investment Management Coventry University 2009 Page 14
Page 15
(1)
HML (High minus Low) is the two high book-to-market stock portfolios average returns minus the two low book-to-market stock portfolios average returns.
HML = 1/2
(2)
According to (Carhart 1997), momentum factor can explain considerable variation in returns. The four-factor model extends the Fama-French three factor model by adding fourth factor, momentum factor. We follow (L'her, Masmoudi and Suret 2004) model to construct momentum factor. In the process of construction portfolios, we rank the stocks based on size for each month from July of year t-1 to June of year t. The size is based on the values of ME at the end of June in year t-1 and prior performance is based on the previous 11-month stock return lagged one month. (Jegadeesh and Titman 2001) point that the reason excludes the most recent month return because to attenuate the continuation effect which caused by bid-ask spread. Winner (WR) contains the top 30% of the total stocks with the highest average prior performance whereas Loser (LR) consists the bottom
MSC Investment Management Coventry University 2009 Page 17
(3)
(4)
Rpt is the average return of six portfolios (S/L, S/M, S/H, B/L, B/M and B/H). Rf is the risk-free rate of AAA rated UK bonds. SMB (Small minus Big) is the three small stock portfolios average returns minus the three big stock portfolios average return. HML (High minus Low) is the two high book-to-market stock portfolios average returns minus the two low book-to-market stock portfolios average returns. The factor sensitivities bi,si,hi are the slope coefficients in the time-series regressions. The three-factor model works better in explaining for the anomalies, except for the shortterm momentum strategy presented by (Jegadeesh and Titman 1993), (Carhart 1997) construct 4-factor model by adding a momentum anomaly with (Fama and French 1993) 3factor model. Carharts fourth factor is based on long on the best-return stocks and short on the worst-return stocks over the previous year. In order to mimic such momentum factor, WML is defined as the return on a portfolio of winner-stocks minus the return on a portfolio of loser-stocks. The four-factor model is written as follow:
Page 18
Page 19
Before we take a step to do analysis on four-factor model, we would like to focus at the mean returns on the four risk factors (Rm-Rf, SMB, HML and WML). In table 1, the market premium, Rm-Rf for the full sample period (7-year) is -0.07 percent per week (tstatistic = -0.64). The market premium for the Bear market is -0.40 percent per week versus 0.13 percent per week for the Bull market. Thus, there is a surprisingly market premium in returns is not very strong. Our market premium results in Bear and Bull market are symmetrical with (Pettengill, Sundaram and Mathur 1995) who point out that the market excess return in negative (Bear) and positive (Bull). However, our results contradict with (Fletcher 1997) findings in the UK stock market over 1975-1994 periods. He shows the market premium in the Bear market is positive and higher than Bull market. He suggests that his findings are inconsistent with (Pettengill, Sundaram and Mathur 1995) and the relationship between beta and return in puzzle. (Graph of market returns can be visualized in Appendix.) Furthermore, we can see that there is a positive return in size premium (SMB) over the sample period. The mean return for SMB is large in full sample, July 2000 to June 2007 (0.10 percent per week) and (0.21 percent per week) in Bull market, July 2004 to June 2006. The annualised SMB standard deviation (8.51 percent) is lower than the market factor standard deviation (14.93 percent). Thus, we suggest that SMB factor is riskless compare to market factor. Next, our results show that book-to-market premium (HML) in positive returns. The HML mean returns is (0.06 percent per week) for full sample, July 2000 to June 2007. In Bear market, the HML mean returns is (0.08 percent per week) which slightly higher than SMB average returns (0.05 percent per week). In Bull market, it gains 0.15 percent per week. Our SMB and HML mean returns results are approximately similar with (Liew and Vassalou 2000) findings in Bear and Bull market. They report annually returns for HML and SMB are 4.37 percent, 5.96 percent respectively in Bull
Page 21
Based on the table 2 (a), we observe that the three small size portfolios (S/L, S/M, S/H) tend to outperform than the three big size portfolios (B/L, B/M, B/H). Refer to the statistical results in Table 1, the total excess returns on the three small stocks portfolios gain 21.84 percent per annum. For the three big stocks portfolios only generate total of return at 6.76 percent per annum. There is a clear inverse relationship between size and average return. We suggest that small stocks generate higher returns than big stocks. This is because small stocks are more risky assets and need higher returns as compensation to bear higher risk. Apart of size factor findings, we also notice that there is a direct relationship between book-to-market and average returns. In other words, the high book-tomarket ratio stocks (value stocks) tend to perform relative well (higher returns) compare to low book-to-market ratio stocks (growth stocks). We look into the high book-to-market portfolios (S/H, B/H), the excess returns is 17.16 percent per annum. For the low book-tomarket portfolios (S/L, B/L), we calculate the excess return is 10.40 percent per annum. The reason value stocks outperform than growth stocks because value stocks are more risky and need higher returns in order to bear higher risk. Our statistical results are consistent with (Fama and French 1996) empirical results. They state that small stocks and value stocks generate higher excess returns compare to big stocks and growth stocks. Table 2(b): Excess Returns on the six portfolios sorted by size and book-to-market ratios from July 2000 to June 2002. (Bear market)
Page 23
Table 2(c): Excess Returns on the six portfolios sorted by size and book-to-market ratios from July 2004 to June 2006. (Bull market)
In Bull market, we examine that all risk factors premium are positive excess returns. We look into the three small stock portfolios which perform well and generate 39 percent per annum. For the three big stock portfolios, they only generate 28.08 percent per annum. There is a clear inverse relationship between size and average return. We can suggest that size factor does exist in the Bull market because small stocks tend to achieve higher returns than big stocks. Moreover, we also notice that there is a clear direct relationship between book-to-market and average returns. The table2 (c) shows that the value stocks (28.60 percent per annum) have higher excess return than growth stocks (17.68 percent per annum). We suggest that firms carry risk premium. In summary on the statistical results, we find that size factor and book-to-market factor does work well in full sample period, bear market and bull market. The results are consistent with Fama and French hypotheses.
MSC Investment Management Coventry University 2009 Page 24
Significant different from zero at the 5% level. a 0.01 (1.82) S/M S/H B/L 0.01 (1.80) 0.01 (1.88) 0.001 *(1.99) b 0.675 *(24.84) 0.703 *(21.71) 0.742 *(27.10) 0.831 *(23.03) s 0.353 *(13.39) 0.272 *(8.65) 0.304 *(11.45) -0.402 *(-11.48) h -0.235 *(-8.81) 0.084 *(2.65) 0.301 *(11.21) -0.304 *(-8.57) w 0.054 (1.90) -0.094 *(-2.78) 0.214 *(7.49) 0.284 *(7.54)
Page 25
Portfolio S/L
In Figure 1, it shows that there is positive relationship between small stock portfolios with market excess. It means that the small stock portfolios tend to perform well (poor) when the market outperform (underperform). According to CAPM, small stocks more risky than big stocks. Therefore, small stocks tend to have high beta and receive higher return as compensation. However, our empirical results show that the market beta does not have any relationship with size factor. Our summary statistics in Table 1, it shows higher beta in big stock portfolios rather than small stock portfolios. The scatter points focus mostly in minus 0.05 percent and 0.05 percent range. The pattern of the graph shows in upward trend. Figure 2: Size Factor
In Figure 2, we observe that the relationship between big stock portfolios and size factor is negative relationship. In other words, holding the big stock portfolios tend to receive lower
MSC Investment Management Coventry University 2009 Page 28
In Figure 3, there is positive relationship between high book-to-market (value) stocks with book-to-market factor. We notice that higher the book-to-market, they can receive higher returns. The reason is value stocks have higher risk than growth stocks due to financial distress in the company. Investors only want to hold this risky assets if higher returns as compensation to them. Our relationship is consistent with Fama and French (1992) findings. There is positive slope to high book-to-market stock portfolios. The scatter points mostly concentrate within the range of minus 0.04 percent and 0.04 percent.
In Figure 4, we observe that the relationship between low book-to-market (growth) stock portfolios with momentum factor is negative relationship. It means that low book-tomarket stock portfolios tend to perform poorer than high book-to-market stock portfolios. (Jegadeesh and Titman 2001) state that both winners and losers are more likely to be small firms because small firms have more volatile and high extreme return. Smaller firms tend to have high book-to-market and large firms more likely to have low book-to-market. Therefore, we suggest that one of the possible reasons is growth stocks are less risky than value stocks because they do not have financial distress problems. Growth stocks have
MSC Investment Management Coventry University 2009 Page 29
Portfoli o S/L
a 0 (0.2)
R2 0.656
S/M
0.001 (0.62)
0.686
S/H
0.658
Page 30
(0.48) *(8.21) *(3.76) *(9.26) (1.03) After we examine the four factor model in full sample period (July 2000 to June 2007), we continue to study this asset pricing model in different market conditions (Bull and Bear market). The reason is check whether the robustness of four factor model has any explanatory power of those risk factors against returns in different market conditions. According to (Pettengill, Sundaram and Mathur 1995), they measure market beta against returns in two different sampling periods (up and down markets). Their findings show that market beta is high against low returns during down market because market returns fall below the risk-free rate. In up market, the market beta is high against high returns because market returns rise above the risk-free rate. Therefore, we can notice that risk-free rate has significant relationship between portfolios returns and market beta in different market conditions. However, (Fama and French 1996) argue that there is weak relationship between beta and return in the US market. Other empirical study done by (Chan, Chen and Hsieh 1985) emphasize that the risk differences between small and large stocks arise from the differences in their market conditions to changes in the underlying risk factors.
Page 33
Significant different from zero at the 5% level. a 0.01 (0.30) 0 (0.06) 0.01 (0.48) 0 (0.30) 0 (0.32) 0 (0.19) b 0.67 *(10.22) 0.712 *(11.52) 0.985 *(11.21) 0.82 *(9.66) 0.774 *(13.87) 0.84 *(9.57) s 0.275 *(6.26) 0.198 *(4.78) 0.3 *(5.09) -0.336 *(-5.90) -0.099 *(-2.65) -0.148 *(-2.52) h -0.217 *(-4.52) 0.191 *(4.22) 0.278 *(4.33) -0.213 *(-3.42) 0.032 (0.80) 0.328 *(5.12) W -0.036 (-0.56) -0.308 *(-5.07) 0.228 *(2.64) 0.125 (1.49) -0.226 *(-4.12) -0.109 *(2.00)
Page 34
Page 37
The Four-Factor Model in the UK Stock Market 5.0 Summary and Conclusion
In our study, we examine the four-factor model in the UK stock market from July 2000 to June 2007. The four-factor model includes market factor (excess market return), size factor (SMB), book-to-market factor (HML) and momentum factor (WML). Based on our research results, we find that the four-factor model does well in explaining the average returns in the UK stock market. According to (Fama and French 1992), they state that the expected returns may not depend wholly on the beta values. In other words, they point out that this signal the death of beta in Fama and French three factor model. In our study, we observe that our market beta shows highly significant at 5 percent level in the four-factor model. This finding is consistent with (Carhart 1997) who shows that beta becomes statistically significant after added momentum factor in the multifactor model. Apart from the market beta is found highly significant, we also find that there is significant in size and book-to-market factor. We find that small stock portfolios seem to have performed better than big stock portfolios and high book-to-market stock portfolios (value) seem to have performed relatively well than low book-to-market stock portfolios (growth). Based on our findings, small firms and value firms tend to perform better in terms of realized returns because these firms carry a risk premium. In other words, small firms and value firms are more likely to have financial distress problem and poor earnings in future. They need to obtain higher returns as compensation for bearing higher risk. Our results are consistent with the (Fama and French 1996) findings. They state that small stocks and value stocks generate higher excess returns compare to big stocks and growth stocks. If the stock price is priced rationally, then other risk factors related to size and book-to-market capable to explain the average returns. Our results show that size premium and book-to-market premium gain average returns of 5.20 percent per annum and 3.12 percent per annum, respectively. However, market premium does not generate a positive return in our full sample period. The annual returns for the market portfolio are minus 3.64 percent per annum. The figures clearly show that size portfolio and book-to-market portfolio perform better than market portfolio. According to (Fama and French 1992), they state that the relationship between market beta and stock returns is flat and stock returns are more likely to depend on size and book-to-market factor. In our view, the evidence that market beta does not suffice to explain expected return in the UK stock market. For the momentum factor, the average annual return is 26.52 percent per annum. Our findings are consistent with (Jegadeesh and Titman 2001) empirical results. They show that
MSC Investment Management Coventry University 2009 Page 38
5.1 Recommendations
In this paper, research findings should give new insights to our understanding of the fourfactor model. The results also make contribution to different fields in terms of academic and investment. For example, portfolio managers can build their investment portfolios based on size factor and book-to-market factor because allocate their investment money in small stock portfolios and value stock portfolios tend to achieve higher returns in the UK stock market. However, our study on the four-factor model still have the limitation to answer numerous questions such as (1) Is the four-factor model pervasive in explaining the risk of industry-sorted portfolios. (2)Is the four-factor model in returns due to common factors in shocks to expected earnings. We intend to pursue these issues in our further research.
6.0 References:
Arshanapali, B., Coggin, D. and Doukas, J. (1998) 'Multifactor Asset Pricing Analysis of International Value Investment Strategies.' Journal of Portfolio Management 24, (4) 10-23
MSC Investment Management Coventry University 2009 Page 39
Page 40
Page 41
Page 43
Appendix:
Picture 1: Market Return in Full Sample (2000-2007)
It demonstrates the market returns are more volatile during the Bull market (20002002).We observe that volatility of the market becomes lower when the market is rising. The market returns trend looks quite stable after Bear market. Picture 2: SMB Return in Full Sample (2000-2007)
In picture 2, we can notice that high volatility appear in Bear market (1-104 week) and Bull market (253-323 week). During that period, the small stock portfolios tend to perform better than big stock portfolios. Picture 3: HML Returns in Full Sample (2000-2007)
MSC Investment Management Coventry University 2009 Page 44
In picture 3, we observe that the volatility of HML looks quite strong during the whole sample period. The HML premium appears largely in the range of 0.02 percent and above. In other words, high book-to-market firms generate higher returns than low book-to-market firms. Picture 4: WML Returns in Full Sample (2000-2007)
In picture 4, the momentum returns have very large volatility in Bear market (53-131 week) compare to Bull market (261-339 week). It shows that winner stocks tend to achieve higher returns than loser market during Bear market. We can notice the WML risks are high in Bear compare to Bull market. Therefore, higher returns appear in Bear market as well.
Page 45