Ảnh hưởng của ba nhân tố theo mô hình fama-french đối với tỷ suất sinh lời của cổ phiếu: Nghiên cứu trên thị trường cổ phiếu Việt Nam

Nội dung text: Ảnh hưởng của ba nhân tố theo mô hình fama-french đối với tỷ suất sinh lời của cổ phiếu: Nghiên cứu trên thị trường cổ phiếu Việt Nam

1. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 THE EFFECTS OF THE THREE FACTORS IN THE FAMA AND FRENCH MODEL ON STOCK RETURNS: THE CASE OF VIETNAM STOCK MARKET ẢNH HƯỞNG CỦA BA NHÂN TỐ THEO MÔ HÌNH FAMA-FRENCH ĐỐI VỚI TỶ SUẤT SINH LỜI CỦA CỔ PHIẾU: NGHIÊN CỨU TRÊN THỊ TRƯỜNG CỔ PHIẾU VIỆT NAM Le Thi Nhu Quynh, Nguyen Thuy Ha, Tran Thi Cam Hoa, Ho Thi Khanh Linh Nguyen Thi Hai Ly, Nguyen Thi Bich Ngoc National Economics University quynhle90.vt@gmail.com ABSTRACT This study focuses on the size and book to market effects and the ability of the Fama and French (1993) three- factor model to explain the cross-section of returns. Based on the return data from three stock markets in Vietnam, including HNX, HoSE and UPCoM, as well as company accounting information from January 1, 2001 to December 31, 2017, the study finds that compared to the Capital Asset Pricing Model, the Fama and French (1993) model provides increased explanatory power in explaining the cross-section of returns in Vietnam. The results show that although stock portfolios carried market risk, a market risk premium is not significant. In contrast, size and book to market effects play a role in asset pricing. From that, investors will have a certain basis of evaluation in these factors when making investment decisions in order to get a better chance of receiving an expected return. Keywords: Fama and French three-factor model, book-to-market equity (BE/ME), size, market risk, the Vietnamese stock market. TÓM TẮT Nghiên cứu này tập trung vào ảnh hưởng của yếu tố quy mô và giá trị sổ sách trên giá trị thị trường, cũng như khả năng của mô hình Fama-French ba nhân tố trong việc giải thích tỷ suất sinh lời của cổ phiếu. Dựa vào dữ liệu về lợi tức từ ba sàn chứng khoán của Vietnam, bao gồm HNX, HoSE và UpCoM, cũng như các thông tin kế toán của công ty từ 1/1/2001 đến 31/12/20017, nghiên cứu chỉ ra rằng so với mô hình CAPM, mô hình Fama- French ba nhân tố có khả năng giải thích tốt hơn cho sự khác biệt về tỷ suất sinh lời của các cổ phiếu. Các kết quả cũng cho thấy mặc dù các danh mục cổ phiếu có tiềm ẩn rủi ro thị trường, nhưng phần bù rủi ro thị trường là không đáng kể. Trái ngược, các nhân tố quy mô và giá trị sổ sách trên giá trị thị trường lại đóng vai trò trong việc định giá tài sản. Từ đó, các nhà đầu tư sẽ có cơ sở đánh giá nhất định đến các nhân tố này khi đưa ra quyết định đầu tư để có cơ hội thu được suất sinh lời kỳ vọng. Từ khóa: Mô hình Fama-French ba nhân tố, giá trị sổ sách trên giá trị thị trường, quy mô, rủi ro thị trường, thị trường chứng khoán Việt Nam. 1. Introduction Understanding the patterns of average stock returns in a world of uncertainty is the most interest of researchers in finance. They have developed models to analyze the riskiness of financial assets and associated premium. Among them, enormous empirical work has been cast in terms of the linear expected return-beta representation. The work of Sharpe (1964) with the Capital Asset Pricing Model (CAPM) laid the foundation for this approach. The CAPM is the single factor model in which expected returns of different assets are a positive linear function of their market β (market risk). In the first generation of CAPM tests such as Lintner (1965), the individual stock returns were used to test and the CAPM failed. The results show a lot of dispersion in the relationship and no significant market risk premium. Jensen et al. (1972) and Fama and MacBeth (1973) proposed grouping stocks into the portfolios by pre-ranking the market betas to reduce idiosyncratic variation in returns and improve beta estimation. Under this sort, CAPM was successfully proved in empirical work during the pre-1969 period. However, after that, the 350
2. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 power of CAPM has been questioned again as many anomalies were still unanswered. This motivates the model developed in the following decades. In the effort of searching for new explanations, a series of research has found other relationships between average returns and factors, rather than market risk. It is shown that the average return has related to the firm characteristics include the size (ME), book to market value ratio (BE/ME), earning to price ratio (E/P), leverage and so on (Banz, 1981, Basu, 1983, Rosenberg et al., 1985). Based on the information, Fama and French (1992) assessed their joint roles in the cross-section of average returns on AMEX, NYSE and NASDAQ stocks. Fama and French (1993), for the first time, introduced the three- factor model. Fama and French (1996) give an excellent summary and show how the model case solve expected return puzzles beyond the size and value effects. In addition to the firm factors, macroeconomic factors have been also considered; for example, labor income (Jagannathan and Wang, 1996), industrial production and inflation (Chen et al., 1986), investment growth (Cochrane, 1996). Under the light of the previous research, the study focuses on testing the validity of the CAMP and the Fama and French three factors in explaining return movements in the Vietnam stock markets. There is a stereotype that the returns of Vietnam‟s stocks do not follow any model but are driven by a group of “big” investors and herd mentality. Is it true? Our research is expected to contribute to the debate on the appropriate asset pricing model to be used in Vietnam. Previous studies have had limited access to accounting variables that affect the scope and their findings. This study hand collects accounting information from all stocks in three stock markets in Vietnam from annual financial reports over the period 2001 to 2017. This is a significant expansion when compared to previous studies in Vietnam and can permit a more comprehensive investigation into three factors of the model. The remainder of the research is structured into four parts. The second part is about a literature review, followed by methodology. The findings and conclusions are presented in the last part. 2. Literature Review Original CAPM and Fama and French Three-Factor Model The key point of the linear asset pricing model is that the expected return of an asset should be high if the asset has high betas or large risk exposure to factors that carry high-risk premia. In the formula, the relationship is presented as follows: In which is the excess return of asset i, is a vector of beta associated with various factors, and is a vector of factor risk premia (the price of risk exposure). In the single factor model CAPM, the amount of risk is a function of the asset‟s covariance with market return and risk premium is the excess market return . The Fama and French three-factor model adds two new factors into the original CAPM: first, the difference between the return on a portfolio of small stocks and large stocks (SMB); second, the difference between the return on a portfolio of high book-to-market stocks and low book-to-market stocks (SML). In the formula form, the expected return is specified as follows: In which are expected premiums. The factor sensitivities are slopes in the time-series regression: 351
3. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 One of the conditions for the validity of the above models is that the market portfolio of invested wealth is mean-variance efficient. Hence, the empirical test is actually the test for the efficiency of the market portfolio, given a set of test portfolios and factors. As mentioned above, the test portfolios are sorted to some criteria. For the CAPM, the test portfolio is sort by pre-ranking market beta (Jensen et al., 1972, Fama and MacBeth, 1973). For the Fama and French three-factor model, Fama and French (1993) and Fama and French (1996) used 25 size-BE/ME portfolios of value-weighted NYSE, AMEX, and NASDAQ stocks. The most popular test for the efficiency of the portfolio is GRS proposed in Gibbons et al. (1989). Under the null hypothesis, all intercepts in the time-series regression are jointly equal to zero and then the portfolio is efficient. Applications of CAPM and Fama and French Three-Factor Model Up to now, there are many studies examining the explanatory power of the CAPM and the Fama & French 3-factor model in different countries in the world. Ajili (2005) used data of 274 stocks in the French stock market in the period of 1976-2001 (300 months) to test two models. The findings suggested that the three-factor model had a stronger power in explaining the stock returns in the sample with an average coefficient of determination of up to 90.5% while in the CAPM, this coefficient was 71.4%. O'Brien (2007) used a sample of more than 98% of companies listed on ASX - Australia for 300 months from 1981 to 2005. His method of establishing portfolios imitated Fama and French (1993). The results showed that the Fama and French (1993) three-factor model explains the return of the portfolios better than CAPM with the coefficient of determination for the three factors model being 73% and that for CAPM being 44%. However, the study showed that the three-factor model had a limitation on representing the yield of big-valued portfolios and was unable to explain the yield of medium-sized portfolios. In other words, the profitability ratio had no linear relationship with the firm size. In the European market, Malin and Veeraraghavan (2004) tested the applicability of the Fama-French model to the UK market. Their study found out the evidence of size effect in the UK but no evidence of value effect (BE/ME). In brief, the applicability of the Fama and French model is varied in different markets. In addition to the studies explaining the significance of firm size and BE/ME, there was also a study to clarify which scope the model should be used in practice. Griffin (2002) raised the question of whether the Fama-French three-factor model applied to a specific country or a global scope better explained the time-series fluctuations in profits of international stocks. Regression of stock portfolios showed that the three-factor model applied to a specific country would be better. In fact, different countries had different investment practices and accounting principles and it was difficult to rank the size of enterprises among developed countries and developing countries. This led to deviations in results obtained from testing the Fama & French three-factor model among countries. Regarding applying the CAPM and the Fama and French three factors in Vietnam, the biggest problem is the shortage of time-series data and accounting information. Compared to developed countries in the world, the Vietnam stock markets began in 2001 and then has been operating for only 18 years. In addition, the stock portfolio has also changed remarkably with many newly listed stocks and delisted stocks. Therefore, the research on the Fama and French three-factor model suffers some certain limitations. Vuong Quan and Ho Hue (2008) studied the sample of 28 non-financial enterprises on HoSE with the weekly return data from January 2005 to March 2008. They found that the Fama-French three- factor model was able to explain better and more fully the changes in stock returns on HoSE compared with the traditional CAPM model. Nguyen Phong and Tran Hoang (2012) using the data from 2007 to 2011 of stocks on the HoSE stock exchange also came into similar conclusions. Vo Hong Duc and Mai Duy Tan (2014) approached many methods of classifying portfolios in Vietnam. The study uses five methods to categorize portfolios and evaluate the suitability of each of these methods. However, contrary to the initial expectations of the Fama and French three-factor model, the results of this study suggest that the market risk factor is always statistically significant without depending on how the portfolios are divided. Therefore, to determine the return rate of stocks in Vietnam, 352
4. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 the traditional CAPM model still appears to be suitable as a starting point. Then, multi-factor models can be used to provide additional evidence for any correction that investors expect. New studies in several recent years have not only stopped at testing the Fama and French three-factor model but also added two other factors, such as Huynh Ngoc Minh Tram (2017) and Nguyen Duc Minh (2016). However, these studies still selected small data samples (for a short time) in order to avoid the impact of the 2008-2009 crisis. Both studies showed that the two newly added variables (RMW and CMA) are not significant in explaining stock returns. However, these above studies suffered the same limitation of small samples, concentrating on stocks listed on only the HOSE, and the selection of a short time period, leading not to apply to the overall market. In this paper, we make an effort to approach the maximum number of securities on the Vietnamese stock market (HNX, HoSE, and UPCoM). This significant increase in the coverage is expected to give a full investigation of the influences of factors in the Vietnam stock markets. For a stock, the volatility of rate on return depends on many factors, not only the market factor, so adding other factors to the market risk can better explain the stock return rather than relying on a single factor in the CAPM. Moreover, the fact that the Fama and French model was founded on the basis of many studies such as Banz (1981), Bhandari (1988), and Rosenberg, Reid, and Lanstein (1985), created breakthroughs when building two factors SMB and HML. Vietnam's stock market is the main "playground" for individual investors who are easily affected by the "herding" factor or the asymmetric information as well as incomplete and inadequate data system. As a result, it is not yet necessary to include 2 factors (profit and investment trend) into the model at this time. Therefore, our study on Fama and French three factors are suitable for Vietnam‟s stock market because compared with CAPM, the model is more adequate and compared with FF5M, the model is more reasonable for the stock market that is young, newly developed and in the process of being more mature. 3. Methodology Dataset The research collected monthly returns data on 1611 stocks in three stock markets, HNX, HoSE, and UPCoM, from 2001 to 2017. For the accounting information related to these stocks, the data before 2008 is manually collected from financial statements while data after 2008 is collected by using the data platform Fiinpro, we have data from 2008 up to now. We use the yield on 5-year government bonds as the proxy for the risk-free rate. Methodology Calculate SMB and HML factors The most important step in testing and analyzing the Fama and French three-factor model is the construction of two SMB and HML factors. In general, the stocks in the sample were arranged in groups of 10%, 20%, 30% of stocks according to size and value orders. Then the difference in returns between the top group and bottom group was calculated to derive the historical value of factor returns. We tried different ways of classifying stock and found that 20% is the most feasible. Grouping by 10% is not feasible because in the early years of the study period there were not enough stocks to carry out classification into 100 portfolios (that were formed from the combination of 2 factors SMB and HML). Meanwhile, grouping by 30% would form 9 portfolios that do not show so significant differences in return that it is quite difficult to analyze appropriately the fluctuation and inclination among them. The SMB factor is constructed by classifying all the stocks in the sample according to the market capitalization value. For each year, based on year-ending market capitalization value, we sort stocks in decreasing order and form five stock portfolios, naming (1)-top 20% largest market capitalization; (2), (3) and (4) - three groups of 20% of stocks with the next largest market capitalization, and (5) 20% of stocks with the lowest market capitalization. 353
5. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 To calculate HML, all stocks are sorted according to the year-ending P/B ratio. Any stock with a negative P/B ratio will be removed from the sample (because companies that own those shares are often in financial difficulties, easily go into bankruptcy and become unlisted, causing the research sample to be discontinuous). Each year, there are five portfolios. The stocks with the highest 20% P/B ratio are classified into the portfolio (a) (the "growth" portfolio). Three groups of 20% of stocks with the next highest P/B ratio are the portfolios (b), (c), (d). And 20% of stocks with the lowest P/B are classified into the portfolio (e) (the "value" portfolio). Thus, there are a total of 25 portfolios created from the combination of five groups classified by the P/B ratio and five groups classified by market capitalization value. These 25 portfolios are held for 12 months and the monthly return is calculated. After 12 months, this process will be repeated. The SMB factor can be obtained by calculating the monthly return of 5 small portfolios ((5) (a), (5) (b), (5) (c), (5) (d), (5) (e)), then subtract the monthly return of 5 big portfolios ((1) (a), (1) (b), (1) (c), (1) (d), (1) (e)). Similarly, the HML factor is obtained by calculating the monthly return of the five "value" portfolios ((e) (1), (e) (2), (e) (3), (e) (4), (e) (5)), then subtract the monthly return of the five "growth" portfolios ((a) (1), (a) (2), ( a) (3), (a) (4), (a) (5)). Finally, we come up with monthly data of SMB and HML. Testing To test the model, we have to establish testing portfolios. The reason behind using portfolio returns instead of using individual stock returns is to remove idiosyncratic factors in individual stocks. In this study, 21 industry portfolios are chosen. The most common test for the efficiency of a market portfolio which consists of 21 industry portfolios is the GRS test, proposed by Gibbons (1989). Under the null hypothesis, all intercepts in the time series regression are jointly equal to zero and then the portfolio is efficient. To find out the risk premium for each risk factor, the study applies the two-step procedure as follows: - Step 1: Estimate betas from time series regression with monthly data on returns and factors: Where, is a vector of proxies for factors - Step 2: Estimate the factor risk premium λ + Traditional procedure: based on a cross-sectional regression of average monthly returns on betas with no constant: + Fama Process – MacBeth procedure: run cross-sectional regression at each month: 354
6. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 4. Empirical Results Co-movements between portfolio returns and market risk, size, and book to market factors .8 .6 .4 .2 0 -.2 2006m1 2008m1 2010m1 2012m1 2014m1 2016m1 2018m1 Market Return smb hml Figure 1: Movements of three factors Figure 1 shows different patterns of co-movements among factors. Considering the period of 132 months from January 2007 to December 2017, market returns and HML have no linear relationship at the significance level of 5%. However, for the period from January 2007 to January 2010 when the stock market experiences the large disturbances, HML, and market return exhibit a strong positive correlation (at 0.3372) (see Figure 2). In fact, the crisis in the Vietnam stock market started in the mid of 2017. Since this time, HML and market return tended to move closely until the mid of 2009. The reason behind this fact could be that during the stressed year, the prices of all stock wildly fluctuated but the companies with higher P/B experienced a greater rise and fall in the market price than those with lower P/B. Therefore the difference in return between the group of highest P/B and lowest P/B was closely linked to the market movement. In contrast, in the normal condition of the market, the relationship between the HML and market return is less remarkable. It can be explained by the fact that HML reflects the difference between groups while market return tends to offset differences through diversification effects. Meanwhile, SMB is positively and significantly correlated with HML in both normal and stressed situations. The correlation tends to be higher in the case of large fluctuation, the coefficient increasing from 0.4511 to 0.5090. This relationship is expected because both factors are heavily influenced by stock prices. One the other hand, SMB has a significantly negative correlation with market return with a correlation of 0.1981 in general but there is no remarkable relation in the stressed time. SMB reflects the difference in return due to size. At the beginning of the crisis, the stock price of a company was adjusted to its intrinsic value. During this time, the stocks whose prices were previously blown lost their position and the companies that were previously underestimated and safer gained the most. This adjustment caused a big change in the SMB factor in the beginning time that was less correlated with the market return. 355
7. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 .8 .6 .4 .2 0 -.2 2007m1 2008m1 2009m1 2010m1 Market Return smb hml Figure 2: Movements of three factors from Jan. 2007 to Jan. 2010 CAPM Estimate beta coefficient measuring market risk Table 1: Beta coefficient of 21 industry portfolios Beta coefficient measuring Standard Industry market risk (in descending t-statistic error (S.E) order) Financial services 1.490 0.085 17.568 Support, advice, design services 1.366 0.063 21.756 Insurance 1.311 0.077 17.084 Automobiles, automotive components 1.306 0.083 15.789 Mechanical engineering 1.258 0.087 14.446 Oil and gas 1.186 0.084 14.085 Construction and construction materials 1.150 0.067 17.224 Warehouse transportation 1.133 0.135 8.366 Tourism and entertainment 1.123 0.111 10.143 Infrastructure services 1.084 0.054 20.202 Technology 1.079 0.061 17.750 Banking 0.989 0.072 13.793 Chemicals 0.884 0.067 13.108 356
8. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 Electronics 0.864 0.090 9.645 Retail 0.839 0.160 5.258 Multi-function industry 0.819 0.070 11.753 Food and beverage 0.797 0.073 10.884 Personal and household consumption goods 0.784 0.109 7.206 Communications 0.783 0.078 10.058 Real estate 0.658 0.090 7.270 Pharmaceuticals and biotechnology 0.509 0.075 6.803 Source: Authors’ calculations First, the efficiency of the market portfolio is tested. The GRS test is used to evaluate the mean- variance efficiency of the market portfolio (including all stocks considered). In the one-factor CAPM model, the GRS test gives p-value = 0.1295> 0.05, hence, the null hypothesis can not be rejected. The market portfolio is efficient in the CAPM model. Table 1 presents the estimated beta coefficients of 21 industry portfolios. Beta coefficients representing the market risk of 21 economic sectors are positive, which reflects the fluctuation in the same way between the rate of return of the industry portfolio and the market portfolio. Besides, all beta coefficients are statistically significant at 1% level. Among them, there are 11 industries with a beta coefficient greater than 1, which means that their market risk is higher than the market average. These industries include financial services; support, advice, design; insurance; automobiles, automotive components; mechanical engineering; Oil and gas; construction and construction materials; warehouse transportation; tourism and entertainment, infrastructure and technology. These sectors are strongly affected by prices, income and other indicators of the market. The remaining ten industries have beta coefficients lower than 1, implying that these industries' risk is lower than the market average. The list consists of industries such as banking, chemistry, electronics, retail, communications, real estate There are some noteworthy findings from the above estimation. For instance, the banking industry and real estate industry are considered as quite sensitive to market volatility but their estimated betas are lower than 1, 0.989 and 0.658 respectively. This is a specific business sector, financial intermediaries financed from individuals, households to corporations , banks can be easily affected by only small volatility in the economy such as commodity prices, GDP growth, inflation, policy, or even fluctuations from other industries. However, with the current economic situation of Vietnam which is quite stable and growing steadily, the macro indicators are in control and in plans of the Government. Besides, the legal framework for the banking industry in recent years has been adjusted appropriately, accordingly, applying international standards in management as well as administration has made this industry operate more stably. The financing for businesses from different industries has also been standardized and better managed by the bank, so in case fluctuations happen, the bank's operations can be still in control. By comparison, the portfolio of non-bank financial service stocks has the highest beta at all (1.49). The group of insurance stocks also has a high beta at 1.311. The portfolio of construction and construction materials stocks has a beta higher than 1 (at 1.15). This result may be contrary to the normal expectation. However, it should be noted that the beta coefficient measures the relationship between the market return and the return of a portfolio. In each group categorized by industry, some stocks have higher market risk and some have lower risk, so they can offset each other. For example, in the banking industry, there are some poorly-functioning banks but if investors hold their stocks together with well-functioning banks„ stocks, returns will offset each other. Hence, the rate of return of the overall industry will less fluctuate. Another point to note is that beta only measures market risk and does not reflect specific risks. In an industry with low market risk, the industry's specific risk is not surely low. 357
9. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 Risk premium Table 2: Market risk premium (%/month) Two-pass regression (Fama - MacBeth Cross-sectional regression Procedure) Factor Risk Risk premium SE t-value SE t-value premium Market 0.12% 0.16% 0.69 0.12% 9.50% 0.01 risk Source: Authors’ calculations The market risk premium is estimated in two ways (see Table 2). The result shows that the compensated return for each unit of market risk is 0.12%/month, equivalent to 2.24%/year. In comparison, many studies conducted to conclude on market risk premium reach different values of the market risk premium such as the study of Dimson, Marsh, and Staunton (2003). In their research, the market risk premium is usually assumed to be between 3-7%, with most studies measuring the market risk premium separately for each country. The point estimates are the same for both estimation methods (two- pass regression and cross-sectional regression). Testing the statistical significance of the risk premium, both methods conclude that the null hypothesis that the risk premium is equal to zero cannot be rejected at the significance levels of 1%, 5%, and 10%. Thus, although the market risk of the portfolios can be measured, the corresponding risk premium is not statistically significant. Under the cross-sectional procedure, the market risk factor can explain only 2.32% of the variance in portfolio returns in the sample. Therefore, we can conclude that the CAPM model is very weak in explaining the volatility of the portfolios‟ returns. This result can be partly explained by the above findings of the relationship among factors: market return, SMB, HML. As presented above, market return is related to HML and SMB. The exclusion of two factors HML and SMB can lead to the problem of inefficient and biased estimators and then the insignificant of the market risk premium. We will see how the three-factor model can improve results. Fama and French Three-Factor Model The results of the GRS test for the Fama-French three-factor model give p-value = 0.1197> 0.05, implying that the null hypothesis cannot be rejected. The market portfolio is considered efficient in the French and Fama three-factor model. Estimate beta coefficient measuring market, size and value risks Table 3: Beta coefficients of 21 industry portfolios Market Risk SMB HML Industry Beta S.E t-value Beta S.E t-value Beta S.E t-value Financial services 1.58 0.09 17.54 0.25 0.08 2.94 -0.41 0.12 -3.53 Support, consulting, design services 1.55 0.11 13.97 0.81 0.1 7.82 0.04 0.14 0.31 Automobiles and automotive 1.41 0.11 12.57 0.32 0.11 3.03 -0.34 0.15 -2.34 components Insurance 1.3 0.11 11.43 -0.06 0.11 -0.61 -0.09 0.15 -0.58 Mechanical engineering 1.28 0.09 14.62 0.03 0.08 0.32 -0.19 0.11 -1.66 Construction and construction 1.25 0.07 16.67 0.32 0.07 4.64 -0.23 0.1 -2.34 materials 358
10. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 Warehouse transportation 1.17 0.07 16.94 0.11 0.06 1.73 -0.16 0.09 -1.74 Tourism and entertainment 1.15 0.17 6.9 0.15 0.16 0.97 0.1 0.22 0.46 Technology 1.11 0.08 14.33 0.04 0.07 0.58 -0.21 0.1 -2.14 Oil and gas 1.1 0.09 12.95 -0.23 0.08 -2.94 0.36 0.11 3.24 Infrastructure services 1.02 0.07 14.43 -0.18 0.07 -2.7 0.21 0.09 2.26 Banking 1.02 0.08 13.44 0.13 0.07 1.82 -0.03 0.1 -0.3 Electronics 0.96 0.09 11.04 0.25 0.08 3.04 -0.41 0.11 -3.64 Communications 0.94 0.07 13.24 0.62 0.07 9.31 -0.13 0.09 -1.45 Multi-function industry 0.93 0.08 11.72 0.37 0.07 4.91 -0.34 0.1 -3.26 Chemicals 0.92 0.06 14.49 0.17 0.06 2.77 -0.01 0.08 -0.12 Personal and household consumer 0.87 0.06 14.9 0.26 0.05 4.68 -0.3 0.08 -3.97 goods Retail 0.86 0.07 12.13 0.07 0.07 1.01 -0.03 0.09 -0.37 Food and beverage 0.8 0.06 14.25 -0.01 0.05 -0.1 -0.06 0.07 -0.86 Real estate 0.72 0.08 9.06 0.17 0.07 2.32 -0.26 0.1 -2.5 Pharmaceuticals and 0.55 0.07 7.52 0.17 0.07 2.52 -0.03 0.1 -0.35 biotechnology Source: Authors’ calculations Table 3 presents the estimation results of beta coefficients measuring three risk factors: market, size, value. First, under the French and Fama three-factor model, market risk of industry portfolios are positive and statistically significant. In terms of the magnitude, compared to those of CAPM, the betas of some industries of the Fama and French three factors have changed. Specifically, three sectors whose betas slightly decrease are oil and gas, infrastructure and insurance; the rested sectors have increasing beta. About size risk (SMB), the sign of the beta coefficient for sectors is either positive or negative. Four sectors with negative SMB beta include oil and gas; food and beverage; infrastructure services and insurance. The 17 remaining industries have positive betas. Six sectors with beta coefficients that are not statistically significant at the significance level of 5% include mechanical engineering; food and beverage; retail; tourism and entertainment; insurance; technology. The remaining sectors are influenced by size risk (SMB). Regarding value risk (HML), the sign of the beta coefficient can be either positive or negative. Four sectors with positive HML beta include oil and gas; support, consulting and design services; tourism and entertainment; infrastructure services. The 17 remaining industries have negative betas. Seven sectors with beta coefficients that are not statistically significant at the significance level of 5% include chemicals; support, consulting, design services; pharmaceuticals and biotechnology; retail; tourism and entertainment; banking; insurance. The 14 remaining sectors present the impact of value risk (HML). These above results are of significance in some aspects. First, compared to the CAPM, the Fama and French three-factor model with additional HML and SMB factors helps adjust the beta of market risk to better reflect the impact of market risk on stock returns in the sample. The market betas of portfolios are significantly adjusted as 17 over 21 show an increase. These changes reflect better co-movement between the return of portfolio and market return after these factors, SMB and HML, have captured some of the turbulence in returns. Second, that these above risk measures of HML and SMB factors are statistically significant shows the evidence of a relationship between asset returns and risks related to size 359
11. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 and value. In other words, size and value risks play a role in determining returns. Third, the evidence that some industries may have a negative or positive beta coefficient is the basis of diversifying investment portfolios to eliminate the influence of size and value factors as well as establishing the portfolios that balance between risk and required returns. Risk premium Table 4: Risk premium (%/month) Two-pass regression (Fama- Cross-sectional regression Factor Macbeth) Risk premium SE t-value Risk premium SE t-value Market 0.05% 0.82% 0.055 0.05% 0.19% 0.240 risk SMB 2.03% 1.24% 1.640 2.03% 0.73% 2.757 HML 1.55% 1.29% 1.200 1.55% 0.86% 1.802 Source: Authors’ calculations Table 4 presents the estimation of risk premiums. All risk premiums are positive. The market risk premium is negligible, about 0.05%/month or 0.6%/year. At the significance level of 10%, we cannot reject the null hypothesis that the market risk premium is equal to zero. This result is similar to the conclusion of the CAPM above. Regarding the size premium (SMB), with each increased risk unit, investors can receive an extra return of about 2.03%/month (equivalent to 24.36%/year). Under the above two estimation procedures, the size premium is statistically significant at 5%. Regarding the value premium (HML), with each increased unit of risk, investors receive an additional return of 1.55%/month, equivalent to 18.6%/year. The value premium is statistically significant at 10%. Therefore, not only do the portfolios exhibit significant exposure to value and size risks but also the premium for these risk are significant and more consistent with real returns in the market. In terms of explanatory power, the three factors can explain 33.55% of the variance of returns in the sample, a remarkable improvement from the CAPM (R-squared=2.32%). These above results also have implications for investors in pricing stocks and setting up a portfolio. One of the most obstacles for the investor is to determine the intrinsic value of a stock. To estimate it, an investor must know the movement and value the discounting return. The Fama-French model can be used to give estimates and predict the trend of returns in the market. Furthermore, we know that investors are trying to find stocks to increase their portfolio return or to adjusted their risk profile, so beta coefficients of the above industries can be a guild to investors. 5. Conclusions It is concluded that compared to the CAPM, the three-factor model has a significantly increasing power in explaining returns movement in Vietnam. Both the CAPM and the Fama and French model show that the stock portfolios expose market risk at different degrees. However, the market risk premiums in both models are statistically insignificant. In other words, the portfolios do not receive market risk premium despite their exposure to market risk. Second, the addition of HML and SMB to the model has significance in explaining the variation of stock returns at two aspects: (i) exposure to risk, (ii) premium for risk. Third, the Fama-French model with the addition of two factors, HML and SMB, can become a base for establishing or diversifying the portfolio and assessing the risk and return relationship for investors. 360
12. INTERNATIONAL CONFERENCE FOR YOUNG RESEARCHERS IN ECONOMICS & BUSINESS 2019 ICYREB 2019 REFERENCES [1] B. Bahl, "Testing the Fama and French Three-Factor Model and its Variants for the Indian Stock Returns", SSRN Electronic Journal, 2006. [2] Black, Fischer., Michael C. Jensen, and Myron Scholes, "The Capital Asset Pricing Model: Some Empirical Tests", Studies in the Theory of Capital Markets, p. 79–121, 1972. [3] Drew, M.E. and Veeraraghavan, M., "Beta, Firm Size, Book-to-Market Equity and Stock Returns: Further Evidence from Emerging Markets", Journal of the Asia Pacific Economy, pp. 8, 354-379., 2003. [4] E Dimson, P Marsh, M Staunton, "Global evidence on the equity risk premium", Journal of Applied Corporate Finance, pp. Volume 15, Issue 4, pages 27-38 , 2003. [5] Fama EF, French KR, "Common risk factors in the returns on stocks and bonds", Journal of Financial Economics, pp. Volume 33, Issue 1, Pages 3-56, 1993. [6] Fama EF, French KR, "Multifactor Explanations of Asset Pricing Anomalies," Journal of Finance, p. 51(1):55–84, 1996. [7] Fama EF, French KR, "Size and book-to-market factors in earnings and returns", The Journal of Finance, pp. Vol 50, issue 1, pages 131–155., 1995. [8] Fama EF, French KR, "The Cross ‐ Section of Expected Stock Returns", The journal of Finance, pp. Volume 47, Issue 2, Pages 427-465, 1992. [9] Fama EF, French KR, "Value versus Growth: The International Evidence", The Journal of Finance, pp. 1975-1999, 1998. [10] J. Lintner, "The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets", The Review of Economics and Statistics, pp. Vol. 47, No. 1 (Feb 1965), pp. 13-37, 1965. [11] J. M. Griffin, "Are the Fama and French Factors Global or Country Specific?" The Review of Financial Studies, vol. 15, no. 3, p. 783–803, 2002. [12] J. Mossin, "Equilibrium in a Capital Asset Market", Econometrica, pp. 34, 768-783, 1966. [13] James L. Davis, Eugene F. Fama and Kenneth R. French, "Characteristics, Covariances, and Average Returns: 1929 to 1997", The Journal of Finance, pp. Vol. 55, No. 1,389-406, 2000. [14] Lo, Andrew W., and A. Craig Mackinlay, "Data-snooping biases in test of financial asset pricing models", The Review of Financial Studies, pp. 3, 431-467., 1990. [15] M. A. O'Brien, "Fama and French Factors in Australia", 2007. [16] Michael R. Gibbons, Stephena. Ross, and Jay Shanke, "A test of the efficiency of a given portfolio", Econometrics, pp. Vol 57, 1121-1152, 1989. [17] Mirela Malin and Madhu Veeraraghavan, "On the Robustness of the Fama and French Multifactor Model: Evidence from France, Germany, and the United Kingdom", International Journal of Business and Economics, pp. vol. 3, issue 2, 155-176, 2004. [18] N. A. Phong và T. V. Hoang, "Applying Fama and French Three Factors Model and Capital Asset Pricing Model in the Stock Exchange of Vietnam", International Research Journal of Finance and Economics, pp. 95,115-120, 2012. [19] N. Billou, "Tests of the CAPM and FAMA and French Three-Factor Model", Working Paper, 2004. [20] R. W. Banz, "The relationship between return and market value of common stocks", Journal of Financial Economics, pp. Vol 9, 3-18, 1981. 361