Master's Thesis Business Administration Finance
--- Student name: V. (Victor) IGNATIUC Student number: S2223856
E-mail: ignatiuc@student.rug.nl Phone: + 31 (0)65 342 68 91
Supervisor: Konstatinos Sklavos, PhD Student E-mail: k.sklavos@rug.nl
---
The impact of size on the performance of BRICS mutual funds
Abstract
In this dissertation is examined the relationship between size and book to market ratio on the returns of mutual funds from BRICS countries. This relationship is analyzed using Fama French Three Factor Model and Carhart’s Four Factor Model. In addition, a cross sectional analysis is used to analyze the impact of four characteristics on the performance. Consistent with the theory which analyzed the US and European mutual funds, the results suggest that there no common pattern for BRICS countries and the results are split. The analyzed size, value, and momentum risk factors cannot explain the excess return in the case of all five analyzed countries. The main contribution of this dissertation is the fact that the analysis of the funds from BRICS countries is not analyzed previously in the literature.
JEL classification: G15, G23
Key words: Mutual funds, BRICS funds, Performance evaluation, CAPM, Momentum
2 Table of Contents Abstract ... 1 Table of Contents ... 2 1 Introduction ... 3 2 Literature review ... 5
2.1 Early studies of mutual funds’ performance... 5
2.2 Characteristics of mutual funds ... 6
2.3 Empirical studies on mutual fund performances ... 8
2.4 BRICS mutual fund industry ... 9
3 Methodology ... 12
3.1 Risk-Adjusted Performance Measures ... 12
3.2 Fama French Model ... 13
3.3 Carhart’s Four-Factor Model ... 14
3.4 Cross-sectional analysis... 15
4 Data and descriptive statistics ... 16
4.1 Survivorship bias ... 16
4.2 Selection of funds ... 16
4.3 Selection of Index ... 17
4.4 Risk-Free rate selection process ... 18
4.5 Mutual funds data selection ... 18
5 Results ... 20
5.1 Risk Adjustments Measures ... 20
5.2 CAPM Based Models ... 21
5.3 Cross sectional model ... 25
6 Conclusion ... 27
Refferences ... 30
3 1 Introduction
The research on mutual funds has served as a topic for numerous academics with interest for market efficiency. With more than forty years of research on mutual funds’ performance, there is a general agreement on some aspects like the failure of mutual funds to outperform market indices. Over the last two decades, a large number of research papers in finance explored the performance of mutual funds and tried to find out what are the factors which might be used to predict it. Most of the studies were based on Fama French Three Factor Model and were focused on the mutual funds from developed economies. The literature has yet to find a consensus regarding the impact of mutual fund characteristic on the performance of mutual funds. Additionally, there are mixed results from the papers that study the US or other individual developed economies. Lately, empirical studies by Otten and Bams (2002) and Ferreira et al. (2012) focused on larger samples of countries and these studies showed a significant difference in the performance of US and non-US funds. The present study builds on the motivation that there is a limited number of research papers which are focused on the emerging markets. In addition, to my knowledge, there is no other study that looks into the group of five big emerging markets: Brazil, China, India, Russia, and South Africa (BRICS).
There are several contributions of this dissertation to the current literature. The analysis of the BRICS mutual funds can provide a worthwhile comparison that might find if there is a similar pattern that can be applied for other emerging economies. It is considered that the group of these five countries are at a comparable stage of the economic development with many common characteristics. The aim of this research paper is to look into how similar are the mutual funds from these emerging markets. Also, it will make possible to study how did the high economic grow impact on the returns of mutual funds. Likewise, the research looks at how are analyzed mutual funds related to the ones from US and other non-US developed economies. In the instance of significant differences between the countries, the study will suggest possible further research directions into what might be the cause.
4 using Carhart’s Four Factor Model (1993). The research objective is to find out if the size, value, or past performance of mutual funds can contribute to the predictability of their performance. The outcomes might give potential investors an insight into how the mutual funds react to the changes in emerging markets and the possibility to adjust their investing strategies accordingly. The motivation for this research question is the lack of research papers that look into the topic of mutual funds’ performance from BRICS.
This dissertation studies the performance of BRICS mutual funds starting from 2004 and until 2010, using a dataset of 1070 mutual funds. I find that there is an asymmetry of the results and that a single pattern cannot be attributed to these emerging markets. To a larger extent, the outcomes of this research are consistent with the studies which considers large samples of developed economies. The results are relevant to researchers and can be used for further study of the characteristics that affect the performance of mutual funds, for example strength of legal institutions, transaction fees, how developed the stock market is etc. Investors might also find a real-world use for the outcomes of this empirical study in their goal to increasing the returns of the investments.
5 2 Literature review
In the literature review are presented the main results from earlier research papers that are highly related to the research questions of this dissertation. In Section 2.1 are presented studies regarding the relationship between size, value and performance of mutual funds. Moreover, here are analyzed other factors that were considered by relevant studies on mutual fund performance. The following part, section 2.2, discusses factors that have been considered to be related to the performance of mutual funds. Sections 2.3 reviews the results from previous studies on the performance of mutual funds. The last part, Section 2.4, describes the mutual funds industry from BRICS countries and takes a look at past studies related to the topic of this dissertation.
2.1 Early studies of mutual funds’ performance
The mutual funds industry has known a substantial grow throughout last years and this contributed to the interest of many academics in the field of finance. An increased interest is focused on ways in which they can optimize the performance of these investing structures. A plethora of attributes have been considered as determinants of the performance of mutual funds, namely size, flows, past performance, fees, turnover, age, expenses, loads, etc. At the forefront of this new trend were two research papers by Sharpe (1966) as well as Treynor and Mazuy (1966). Both studies demonstrated that most of the mutual fund in the U.S. were underperforming when compared to the market return. Later, Jensen (1969) analyzed the performance of mutual funds based on the Capital Asset Pricing Model (CAPM), which is centered on the assumption that changes in the returns of the funds are driven by the market beta. He determined that the returns of mutual funds after fees/commissions were very poor. According to this study, managers did not have good stock picking skills based on a 19 year time period. These three studies used a risk-adjusted performance, which demonstrated an overall incapability of mutual fund to surpass the market portfolio.
6 important factor that was found to explain the performance is value factor, which is described by Rosenberg et al. (1985) on the example of stocks and not mutual funds.
The literature gained additional support with the study by Fama and French (1992), which showed that the CAPM market beta is not as strongly related to the equity performance as it was considered before. Besides the initial study on the US market, Fama and French (1998) provided additional evidence for their claims by means of an investigation grounded in international markets. The study by Fama and French added to the original CAPM model two additional factors: one size factor and one value factor. As a result of their study, the new multifactor model was able to be a decent measure of macroeconomic tendencies (Liew and Vassalou, 2000), with explanatory power of additional factors for the performance of assets. The first factor in Fama French Three Factor Model, Small minus Big (SMB), is the average return on the small capitalization portfolio from which is subtracted the average return on the large capitalization portfolio. The second factor, High minus Low (HML), is the difference in return between the portfolio with high book-to-market stocks and the portfolio with low book-to-market stocks.
A further development of the model is considered to be the contribution made by Carhart (1997). He uses Fama French Three Factor Model and an additional momentum factor with the purpose of testing if there is a persistence in the performance of mutual funds’ portfolios. While a positive persistence in the performance of mutual funds can be attributed to the momentum effect, it is more difficult to interpret a negative persistence. The momentum factor was created by analyzing the previous monthly performance in order to detect any cross-sectional return patterns. The original model used the previous one year monthly data, from which is excluded the last month. Carhart’s Four Factor Model is considered to be an improved version of the previous model because it shows that only the best mutual funds are able to outperform the market index by covering fund fees/commissions and that there is a negative correlation between the returns and mutual fund fees.
2.2 Characteristics of mutual funds
7 The central factor in the analysis of mutual funds’ performance is considered to be the size of mutual funds. There are no consistent results in the literature regarding the influence of size on the performance of funds. It is still not clear what could be the reason for which larger funds have a lower performance in comparison to smaller funds, or vice versa.
According to Williamson (1988) the bureaucratic and other organizational attributes play here a key role. They contribute to higher costs and have a negative effect on the performance. Another cause of lower returns for funds with high amount of assets under management is considered to be the agency cost issue and is discussed by Nanda and Wang (2008). According to them, CEO’s of the very large funds do not manage properly the assets under their control in order to maximize the performance for investors. Also, funds with relative low assets level have fewer management levels, which contributes to shorter decision taking times compared to larger funds (Stein, 2002). This means that large funds lack the flexibility and this drives their performance down. Large funds are also linked to higher transaction costs (Perold and Salomon, 1991) which drive returns down.
Chen et al. (2004) consider that the returns of funds decrease with its dimension, even taking into consideration the transaction costs. They argue as well that this is produced by bureaucratic factors and low flexibility. The empirical research by Yan (2008) supports the invers relation between size and performance by analyzing the funds from USA. According to the researcher, high grow and high turnover funds are more in line with this hypothesis. Also, in the case of small funds, managers are considered to have higher flexibility with regards to adjusting fund holdings (Pollet and Wilson, 2008). Gil-Bazo and Martinez (2003) found that there are significant differences between US and Spanish markets in terms of transaction costs which are an advantage for large US funds, but is not significant for Spanish funds.
When considering the size of mutual funds, the level of expenses seems to have a significant role of in contributing to overall fund performance (Carhart, 1997; Wermers, 1997). There is also a discussion regarding the brokerage commissions, R&D, and marketing costs which are considered to be higher in the case of small funds and this drives their performance down comparing to larger funds.
8 the returns. According to them, the persistence in performance is a result of the existing lag in relation to the changes in the market index. Carhart (1997) also analyzed the persistence in returns. In order to analyze this occurrence, he introduced a new momentum factor as an additional risk factor to Fama French Three Factor model. More recently, Bollen and Busse (2005) determined that persistence in the return is observable mostly just on short periods of time.
Elton et al. (1995) suggest that investors that typically invest in a market portfolio outperform the ones that invest in mutual funds, even if the fees are not considered. Sharpe model and Treynor model were developed as a performance measures ratios and contributed to the understanding of how the ability of the managers can be determined. With the help of these tools it can be showed that the ability of managers does not contribute to a significant increase of the performance of mutual funds.
According to Dellva and Olson (1998), the quality of information plays an important role for the final return of mutual funds. So, managers with better sources of information managed to achieve a generally superior performance than the others. Dellva and Olson (1998) reached this conclusion when they tried to find what the impact of size on the performance of mutual funds is. One possible explanation is considered to be the idea that the funds with better sources of information might also have access to less expensive sources of capital for future investments.
2.3 Empirical studies on mutual fund performances
9 The research paper by Chen et al. (2004) shows the presence of a negative impact of funds size on the returns in the US market. Their analysis was confirmed later by and Yan (2008). Otten and Bums (2002) performed an ample research of the European countries with a developed mutual funds industry and included six countries. They found that performance for the funds increases with their size, which is opposite to the US market. Contrary to the above studies, Dahlquist et al. (2000) show that for Swedish funds, performance is not noticeable affected by the size of the mutual funds. Grinblatt and Keloharju (2000) look at the performance of mutual funds from Finland and find that foreign funds use momentum strategies and are likely to have higher returns than local investors which do not use these strategies.
There is still no extensive coverage of the mutual funds’ performance in the economies outside US and European. Nevertheless, there are some exceptions, but mostly for the developed economies. Chan and colleagues (2005) show that for funds from Australia the return of the funds decreases with size as a consequence of the increase in costs. Brown et al. (2008) found that on the markets of Hong Kong, Korea, Singapore and Taiwan there is persistence in the returns and that the value factor plays an important role on the performance.
For the BRICS countries, there is a limited number of empirical studies. Tang et al. (2011) find an inverted U-shape relationship between size and performance of mutual funds from China. Su, Zhao, Yi, and Dutta (2012) found no evidence of long term persistence in the returns of Chinese mutual funds. Cao and He (2011) found that that Chinese open-ended funds provide positive abnormal returns and market timing ability. Connor and Sehgal (2001) focus on Indian mutual funds, but did not find a strong link between risk factors and the return of the funds. Roy and Deb (2003) determined that Indian funds were able to outperform the benchmark portfolio. Except for China and India, there is a gap in the literature regarding the performance of mutual fund from BRICS.
2.4 BRICS mutual fund industry
10 Global mutual fund industry has seen a significant ascending trend in the recent years, despite of the negative effects of global financial crises which had a profound impact on the economic development of the world. Recent growth of the exchange-traded fund industry had also a considerable impact on the mutual funds. Still, they managed to achieve a considerable 3.9% in 2012.
The initial BRIC acronym denoted Brazil, Russia, India, and China countries and started to be known since 2001. It indicated that these countries are at a comparable stage of economic development and it was used as a sign of the expected changes in the economic power. In this paper I will refer to the acronym BRICS which takes into account South Africa, as in 2010 it was formally admitted into this group of countries. Together, these five economies, are considered to have a great potential with a total population of around 3 billion and a GDP of $13.7 trillion which is around 20% of the global world product. According to the data from Investment Company 2012, the number of mutual funds has increased from 2005 until 2011 for Brazil by 142.6%, for India by 52.8%, for Russia by 83.7%, and for South Africa has increased by 53.5%. For China the number increased by 144% from 2007 until 2011. From Total Net Assets point of view, Brazil had the biggest increase of around 229.4% for the last six years. At the opposite side, China had a decrease of the total net assets by 21.9%, being the only country which had a negative growth.
The studies on the performance of mutual funds contribute to the understanding of how are the fund managers able to compare with the benchmark, with other funds, and with mutual funds from other markets. As a result, it is possible for investors to distribute more efficiently their holdings with higher potential returns. Except for the management skills, these studies contribute to the understanding of how other characteristics (size, value, etc.) contribute to the final result of the funds from individual financial markets and offer an insight into what might be the cause for a superior or inferior performance.
11 • BRICS mutual fund are getting on average a superior performance to that of the benchmark index regarding the risk and if they are better from the point of view of portfolio optimization to their investors.
• There is a significant spread in returns between large and small sized mutual funds. • There is a significant spread in performance of value and growth mutual funds.
• There is an impact of the past performance on the current returns of BRICS mutual funds.
12 3 Methodology
This part of the paper explains the research methodology which is going to be used for testing the research hypothesis of the dissertation. More specific, here are described the methods that are used and the type of statistical relationships need to be in order to analyze given data. I also discuss what are the essential control variables and what are my anticipations regarding them. I further analyze what are the role of the sign or of the significance and their impact on the more general hypothesis that are going to be tested.
The methodology for this dissertation use a similar approach as key research papers presented in the literature review and is developed based on the theory of risk-adjusted performance measurement and CAPM. CAPM is a single-factor model with a presumed linear relationship between the returns of the analyzed asset and of the market portfolio. While CAPM relies on just one risk-factor, it can be developed by the further introduction of new risk factors. For example, Fama and French three factor model adds two other risk factors, which reflect the size and book to market effects. Carhart’s Four Factor Model expands further the previous model by adding a momentum risk factor. In addition to the previous methods, I start my analysis with the risk-adjusted performance measures with Sharpe ratio, Treynor ratio, and Jensen’s alpha.
3.1 Risk-Adjusted Performance Measures
As mentioned above, this dissertation will analyze the risk-adjusted performance of BRICS mutual funds. The analysis will include three measurements: Sharpe Ratio, Jensen’s Alpha, and Treynor Ratio. Sharpe Ratio is calculate by dividing the difference between mutual funds’ return and risk free rate by the standard deviation of these returns. It is an assessment of how efficient is the relationship among annualized risk-free return and standard deviation. A large value for this ratio means that mutual funds are more efficient on average.
𝑆ℎ𝑎𝑟𝑝𝑒 𝑅𝑎𝑡𝑖𝑜 = 𝑅𝑝−𝑟𝑓
σ (1)
, where Rp is the expected return, σ is the standard deviation of returns, and rf is the risk free rate
13 𝑇𝑟𝑒𝑦𝑛𝑜𝑟 𝑅𝑎𝑡𝑖𝑜 =𝑅𝑝−𝑟𝑓
𝛽 (2)
, where 𝛽 is the beta of the portfolio measured against the market portfolio.
Jensen’s alpha, is considered to be an estimation of the amount of returns which are a result of the manager’s skill. It is considered that any portfolio is expected to get at least the risk free rate and the beta return, which is attributed to the market factors. The remaining part of the return is considered to be a result of the manager’s skill. Jensen’s alpha is an easy method to determine the performance of the mutual fund, even if it might not reflect every fluctuation in the returns of the mutual funds. The excess market return is used as the explanatory variable. The model is as follows:
Rpt − Rft =α + β(Rmt − Rft) + et (3) By rearranging, we get Alpha:
α = (Rpt − Rft) −β(Rmt − Rft) + et (4)
, where Rpt is the return of the portfolio of the mutual fund in period t; Rmt is the return of the benchmark in period t; Rft is the risk free rate in period t; et is the error term in period t.
3.2 Fama French Model
The single factor model used by CAPM has the assumption that it is possible to estimate investment activity of mutual funds using just one market index. This model, however, does not take into considerations the means which are not listed into this market index, smaller companies can serve as an example. For this reason, Fama and French (1992) offer solid arguments for the introduction of extra risk-factors. Fama and French propose a three-factor model that reduces regular CAPM pricing errors by size and book to market factors. Their model is:
𝑅𝑖𝑡 = α𝑖 + β0𝑖 𝑅𝑀𝑡 + β1𝑖 𝑆𝑀𝐵𝑡 + β2𝑖 𝐻𝑀𝐿𝑡 + ε𝑖𝑡 (5)
, where 𝑅𝑖𝑡 is the return of fund in month t; β0𝑖 is the excess return in US dollars on the market; 𝑆𝑀𝐵𝑡 is the average return on the small capitalization portfolio minus the average return on the large capitalization portfolio; 𝐻𝑀𝐿𝑡 is the difference in return between the portfolio with high
14 Other papers from the literature on this topic divided the data set into groups by the size variable. The approach in this dissertation will be alike, meaning that the funds for emerging markets will be sorted by total assets and by book to market value. As a result, this allows to compare the performance of big and small funds to see if there is a relationship between size and their performance.
The approach by Indro et al. (1999) was to split their database into 10 groups with a similar number of stocks. In contrast, Gorman (1991) selected specific limits for the size assets and grouped them according to these limits into the big and small ones without paying attention to the number of funds in each group. The method from this paper is more similar to the one used by Fama and French (1992). They constructed 6 value-weighted portfolios based on their size and book to market values. After ranking by the market capitalization, the initial portfolio is divided into 2 groups. Large stocks are considered the stocks which have a market capitalization above the median value from the portfolio. Small stocks are respectively the ones with a market capitalization below the median value.
Following Fama and French (1992) methodology, the large and small portfolios are used for the construction of the final six portfolios. In order to do this, the large and small portfolios are going to be sorted by book to market values and divided into three portfolios each, covering top 30%, median 40%, and bottom 30% by book to market values. The resulting 6 portfolios will be: Big Value, Neutral Value, Big Growth, Small Value, Small Neutral, and Small Growth.
3.3 Carhart’s Four-Factor Model
For the reason that several funds have a tendency to use more active investment strategies, the returns of specific mutual funds are likely to depend on various economic risk factors. For this reason, I carry on with the analysis by using the Carhart’s four-factor model, where momentum is introduced as an additional risk factor to the Fama and French three-factor model. The outcomes from these models are going to be compared with the purpose of determining if it delivers superior explanation power for the performance of mutual funds. The regression for Four Factor Model is given by:
15 , where MOMt (Momentum) is the difference in return between the portfolio with the past
12-month winners and the portfolio with the past 12-12-month losers.
The results offered by SMB, HML, and MOM risk factors are considered to be a deviation from the normal because, according to CAPM, beta is the only factor that can explain the returns of the funds.
3.4 Cross-sectional analysis
In addition to the previous analysis, this study also uses a two-stage cross sectional methodology. This approach looks at which variables may explain the cross-section of BRICS mutual fund performance. In contrast, according to CAPM, beta is only variable able to do this.
𝛼𝑖 = c0 + c1 𝛽𝑖 + c2 𝐶𝐻𝐴𝑅𝑖 + ε𝑖 (7)
, where CHARi is a variable of the mutual fund i, which is not linked to CAPM.
16 4 Data and descriptive statistics
In this section provide information about the provenience and a general summary of the data used in this study. Based on the methodology from the literature, I constructed the dataset using monthly values in order to analyze the performance of BRICS mutual funds’ performance. The data collection is based on the information imported from the Bloomberg database and contains the returns of mutual funds with additional information for funds characteristics. Overall, the dataset includes information about 1070 funds from BRICS countries for the time interval January 2004 through December 2010. All the returns are in USD in order to make the comparison between BRICS countries easier.
One disadvantage of the analysis of BRICS mutual funds is that the amount of available information is limited. Since the information available in Bloomberg database does not contain sufficient data for all mutual funds, it is not possible to analyze the performance of funds based on additional characteristics such as: fees and expenses, front-end and back-end loads, investment strategies etc.
4.1 Survivorship bias
Despite of how comprehensive is the analysis for this dissertation, the obstacles of the dataset generate inevitable biases and imprecision in the outcomes of the study. To begin with, one of the main constraint of the dissertation consisted in the insufficient data in the Bloomberg database. More exactly, some funds were left out from the sample because of the lack of information for large period of time in 2004-2010 interval. Therefore the database is not survivorship bias free. There is a high probability that a number of the excluded mutual funds were poor performers and this could have been the reason for missing data. Thus, the expect results might be skewed slightly higher as other papers showed (Carhart et. al. 2002). Also, Brown and colleagues (1992) demonstrated that the effects of survivorship bias are found not just in the estimated performance.
4.2 Selection of funds
17 ended funds and index tracking funds. Due to these limits, the initial sample was reduced to the final number of 1070 funds. In the Table 4 from Appendix can be seen the summary statistics for the sample of mutual funds.
4.3 Selection of Index
As for market indices, this paper will use Bovespa Index as a proxy for Brazilian equity market. Bovespa Index is the most important indicator of average share prices of shares traded on the São Paulo Stock Exchange and it accounts for around 70% of the all the stock value traded. As a proxy for Chinese market, I use Shanghai Stock Exchange Composite Index which takes into consideration all share classes and has more than 850 of listed companies. SENSEX Index was used in the case of Indian market. It consists of 30 biggest and most traded stocks from different sectors of Indian economy; MICEX Index accounts for more than 90% of the entire turnover of the biggest Russian stock market; finally, for South Africa I use FTSE/JSE AFRICA TOP40 Index which is a measure of South African top 40 stock market performers. All these indices were chosen as benchmarks for the market returns because they represent a wide spectrum of domestic equity and they offer a very good coverage of the domestic equity market. For the common portfolio of mutual funds is used the MSCI Emerging Markets Index as it is broadly used among researchers and market players.
The total average return for the period 2004-2010 was 202.1% for the entire BRICS portfolio of mutual funds, while the total return for the same period in the case of market indices was 251.7%. Assuming equally weighted portfolios, Brazil had the highest return for the seven year period with 329.6%, followed by China with 320.7%. India’s mutual funds had an average 115.9% return; funds from Russia had a return of 89.9%; South Africa’s mutual funds grew by 154.6%.
18 that of mutual funds, with just 89.92%. FTSE/JSE AFRICA TOP40 Index increased by 251.7% over the seven years period and outperformed the performance of mutual fund on average, which had a 154.6%. Finally, MSCI Emerging Markets Index increased by 155.85%.
4.4 Risk-Free rate selection process
The methodology for this dissertation also requires a risk free rate in order to generate excess return variables with the aim of applying the models. The risk free rate has to satisfy the condition that there can be no default risk associated to it (Damodaran, 2008). In essence, this excludes any securities emitted by a private company. The single assets type which have the possibility to be risk free are government securities, due to instrument through which they can manage money printing process. But, even these instruments do not make governmental securities risk free as can be seen in the example of Russia, which faced a default on its own debt in 1998. In the case of other BRICS countries it is not easy to find a good proxy for the risk free rate. This is mainly because the governmental bond yields for these countries are not considered to be risk free. This dissertation uses the yield of 3-Month US Treasury Billsas a proxy for the risk free rate because it is generally regarded as such.
4.5 Mutual funds data selection
The final sample includes 1070 mutual funds from Brazil, China, India, Russia, and South Africa. The number of mutual funds available in the database was lower in 2004 for Russia and China compared to the other three countries. As a consequence, the weight of Russia and China in the total sample is significantly lower than the other weights for Brazil, India and, South Africa, but can be still considered in the acceptable limits.
The overall value for the mutual funds included in the dataset is estimated to be around $4,575 billion from which: 711 funds for Brazil with total assets under management (AUM) of more than $810 billion; 18 funds for China with AUM of $66 billion; 228 funds for India with more than $2,717 billion AUM; 20 funds for Russia with more than $541 billion AUM; and 93 funds for South Africa with around $439 billion AUM.
19 for each country. Following the methodology, the large and small portfolios were used to construct the final six portfolios. Next, the large and small portfolios were sorted by book to market values and divided into three portfolios each, covering top 30%, median 40%, and bottom 30% by book to market values. The resulting 6 portfolios are: Big Value, Neutral Value, Big Growth, Small Value, Small Neutral, and Small Growth.
20 5 Results
This chapter shows the empirical results of the dissertation, which were obtained based on the methodology and the collected data. This paper applies the additional risk adjusted performance measures Sharpe Ratio, Treynor Index, and Jensen’s Alpha with the purpose of testing the hypotheses. In addition, three regression models are applied to study the mutual fund data, CAPM and the Fama/French Three Factor Model.
5.1 Risk Adjustments Measures
Table 11 from Appendix presents the results obtained from application of the above mentioned risk adjusted performance measures for the portfolio of mutual funds from each individual country. As is shown in the table, Sharpe Ratios are higher than Treynor Ratios for all the countries from my sample. This suggests that for all of the BRICS countries there is a higher return obtained for each unit of fund’s risk than when reported to beta. Taking into consideration the level of risk for the returns of the mutual funds, it seems that mutual funds from Brazil and China have the highest performance among the BRICS countries. It is also true for the results of the Sharpe Ratio where Brazil has the highest ratio with 0.266 and China follows behind with a ratio of 0.259. Sharpe Ratio is the lowest for Russia with a value of 0.098, or 9.8% and it means that investors from Russia had on average a considerable lower reward per unit of risk than the other BRICS countries. The other two countries, South Africa and India, have a slightly higher Sharpe Ratios than Russia of 0.156 and 0.136, which still is considerably lower that Brazil and China.
21 Jensen’s Alpha is considered to be a better measure than previous measures for the reason that it is able to quantify the added value of the mutual funds. Jensen’s Alpha is comparable for China and Brazil with values that are around 0.003. On the other hand, alphas for India and Russia has negative values of -0.002 and -0.004. This indicates that fund managers were not able on average to outperform the market benchmark. The value for alpha in the case of South Africa is 0.001 and it suggest that the performance of mutual funds was not significantly different from that of market index.
5.2 CAPM Based Models
As described in the methodology section, the first OLS regression will test the impact of monthly market returns on the returns of mutual funds from the portfolios generated for each BRICS country. Table 1 shows the results for CAPM model. The estimation results for market premium (Mkt.) coefficient is significant in all of the portfolios with p-values significant at 1% level. All beta coefficients are positive with the highest value for Russia with 0.92. On the other hand, from the estimated beta coefficient of the mutual funds portfolio from China is 0.49 and this is the lowest correlated volatility between the return of mutual funds and the volatility of market benchmark. This suggests that Chinese funds are not as susceptible to the systematic risk as are the funds from the other countries. For the other three countries, the estimated beta has a value of 0.60 for India, 0.71 for South Africa, and 0.67 for Brazil.
Table 1
CAPM estimation results for BRICS countries.
Estimation of equation for CAPM for the sample period 2004-2010. Estimates in the case of each BRICS country are obtained using OLS regressions. Presented is the CAPM regression coefficients that are alpha (risk-adjusted performance) and Mkt. (beta coefficient) and adjusted R-squared for the goodness of fit. All alphas are annualized. The mutual fund data are given by the equally weighted average returns.
Period 2004-2010
Country Alpha Mkt. Adj. R-squared
Brazil 0.13** 0.67*** 0.59
China 0.12* 0.49*** 0.31
India 0.03 0.60*** 0.66
Russia -0.01 0.92*** 0.63
South Africa 0.05 0.71*** 0.59
22 The adjusted R-squared values show that the CAPM model does a relatively good job in explaining this relationship for Brazil, India, Russia, and South Africa, with values ranging from 0.59 to 0.66. On the other hand, the lowest value belongs for adjusted R-squared belongs to China, with 0.31, which does not give a very strong explanatory power of the model on the relationship between Chinese market return and mutual funds portfolio.
23 Table 2
Three Factor Model estimation results for BRICS countries.
The table below presents the outcome for the Fama and French 3 Factor Model in the case of each BRICS country, with regression coefficients (Alpha, Mkt. representing Beta coefficient, SMB representing size risk factor, and HML representing value risk factor) and adjusted R-squared for the goodness of fit. All alphas are annualized. The mutual fund data are given by the equally weighted average returns.
Period 2004-2010
Alpha Mkt. SMB HML Adj. R-squared
Brazil 0.1** 0.62*** -0.25** 0.23* 0.63
China 0.18*** 0.59*** 0.44** -0.47*** 0.48
India 0.03 0.60*** 0.01 0.04 0.65
Russia -0.01*** 0.88*** -0.33* 0.07 0.64
South Africa 0.05 0.67*** -0.29* 0.06 0.63
***Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level
As it can be seen in the Table 2, these additional HML factor is significant only for Brazil at a 0.1 significance level and for China at 1% level. A negative sign for HML coefficient suggests that low book to market growth mutual funds have outperformed high book to market value funds for China. In Brazil, a 0.23 HML suggests that high book to market value mutual funds had a better performance than low value funds. SMB factor, as well as HML risk factors for India are not significant due to a high p-value. This means that size and value factors cannot be used to explain the relationship with the returns of mutual funds. Furthermore, the addition of SMB and HML risk factors cuts the goodness of fit for the regression and it means that these two factors are not improving the initial CAPM model. This result is supported by the study of Connor and Sehgal (2001), which found that the risk factors are not significant.
24 Table 3
Four Factor Model estimation for BRICS countries
The table below represents the outcome for the Carhart’s Four Factor Model in the case of each BRICS country, with regression coefficients (Alpha, Mkt. representing Beta coefficient, SMB representing size risk factor, HML representing value risk factor, and MOM representing momentum risk factor), their corresponding probabilities, and adjusted R-squared for the goodness of fit. All alphas are annualized. The mutual fund data are given by the equally weighted average returns.
Period 2004-2010
Alpha Mkt. SMB HML MOM Adj. R-squared
Brazil 0.10** 0.62*** -0.25* 0.22* 0.01** 0.63
China 0.17 0.60*** 0.44*** -0.54*** 0.21*** 0.53
India 0.03 0.59*** 0.01 0.04 -0.03 0.65
Russia 0.002 0.86*** -0.33* 0.15 -0.25*** 0.69
South Africa 0.05 0.67*** -0.29* 0.07 -0.03 0.61
***Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level
25 Table 4
Four Factor Model estimation for Top and Bottom portfolios
The table below represents the outcome for the Carhart’s Four Factor Model in the case of each BRICS country, with regression coefficients (Alpha, Mkt. representing Beta coefficient, SMB representing size risk factor, HML representing value risk factor, and MOM representing momentum risk factor), their corresponding probabilities, and adjusted R-squared for the goodness of fit. All alphas are annualized. The table analyzes the top and bottom deciles for a common sample containing funds from Brazil, India, and South Africa. China and Russia are included into another sample due to a lower number of mutual funds.
Period 2004-2010
Alpha Mkt SMB HML MOM Adj. R-squared
Top 10% 0.05 0.78*** -0.24 0.23 0.01 0.70
Bottom 10% 0.06 0.53*** -0.04 0.06 -0.03 0.74
Top 30% -0.03 0.75*** -0.17 0.25* -0.23*** 0.71
Bottom 30% 0.14** 0.75*** 0.36** -0.45*** 0.15** 0.64
***Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level 5.3 Cross sectional model
Except for the value variable of mutual funds, the selected characteristics demonstrated significant values at 1% level. The total net assets variable, confirms the outcomes from the Three and Four Factor models discussed above, which showed that size can have a significant impact on the performance. Model 2 indicates a slight negative impact of the size variable. In the Model 3, which analyzes the impact of all four characteristics of mutual funds, there is no evidence of significant value impact on the alphas of the funds. It also shows a significant value of -0.01 impact of the fund age on the performance.
Table 5
Cross-section estimations for BRICS common sample
The table below represents the results for the cross-section regression of the average monthly returns for the common sample of BRICS mutual funds with four characteristics: beta for the mutual funds, the log of mutual fund total assets (LN(TNA)), log of the mutual funds’ value (LN(Be/ME)), log of the mutual funds’ age represented in the number of months from inception, and the volatility of monthly returns. The mutual fund data are given by the equally weighted average returns.
Model BETA LN(TNA) LN(Be/Me) LN(AGE)
1 0.01***
2 -0.003*** -0.001***
3 0.02*** -0.001*** 0.00 -0.01***
27 6 Conclusion
In this dissertation, I have analyzed the risk-adjusted performance and the relationship between size and book to market ratio on the returns of mutual funds from Brazil, China, India, Russia, and South Africa (BRICS) during a time length of seven years, between 2004 and 2010. To my knowledge, there are no papers that have applied the Fama-French Three Factor Model and the Carhart Four Factor Model to study the risk-adjusted performance and a cross sectional analysis for the mutual funds from these five countries. The motivation for this dissertation was the opportunity to study if there is a consensus in the way in which performance of mutual funds from BRICS countries is affected by factors studied in the literature. Increasing importance of the BRICS economies and the high grow rates in the mutual funds industry from these countries contributes to the interest in analyzing this topic.
The risk adjusted performance analysis suggests that mutual funds from China and Brazil considerably outperformed on average mutual funds from the India, Russia, and South Africa. Also, Chinese mutual funds have on average a superior performance than that of the market benchmark. In addition, this dissertation shows that for BRICS countries there is not a common pattern regarding the relationship between size and value of mutual funds on their performance. This is also valid for the persistence of the mutual fund previous one year performance.
28 The momentum risk factor, introduced by Carhart’s Four Factor Model can explain to some extent the performance in the case of China, Brazil, and Russia. The performance from previous year of Chinese and Brazilian mutual funds is likely to be persistent, which means that returns are likely to be positive if the returns from previous one year were positive, and the same is for past negative returns.
The cross sectional analysis on the impact made by funds’ characteristics on the performance suggest that the results are in line with the literature on the topic, as showed by Ferreira and others (2012). The size and age have a slight negative impact on the performance of the common sample of mutual funds from BRICS.
The finding of this dissertation are not consistent across these analyzed group of countries as I expected. However, is demonstrated that similarly to other samples from developed economies, there is no symmetry in the size-performance relationship between BRICS funds results. South Africa and India funds showed results similar to the ones from US, where it is found a direct relationship between size and performance.
In spite of the comprehensive analysis led within the present dissertation, the limits of the dataset distorted the outcomes to some degree. In the beginning, a significant constraint for this analysis was generated by the absence of sufficient data regarding BRICS mutual funds in the Bloomberg database. To be more precise, because of the lack of data for long period of time in the interval 2004-2010, some mutual funds were left out from the initial sample. Hence, the dataset is not survivorship bias free and there is a high possibility that the excluded funds were poor performers with expected results which might be skewed slightly higher.
Another limit concerning the dataset is the limited number of mutual funds data available starting with 2003 for Russia and China. Even if then number of mutual funds used for these countries was similar to the one used by other studies, I consider that a larger number of mutual funds might offer an increased confidence for the outcomes.
30 Refferences
Basu, S., "The relationship between earnings' yield, market value and return for NYSE common stocks: Further evidence." Journal of financial economics 12, no. 1 (1983): 129-156.
Bollen, N.P., Busse, J.A., Short-term persistence in mutual fund performance. Review of Financial Studies 18, (2005):569–597
Brown, P., Kleidon A.W., Marsh T. A., "New evidence on the nature of size-related anomalies in stock prices." Journal of Financial Economics 12, no. 1 (1983): 33-56.
Brown. S., Goetzmann. W., Ibbotsson. R., Ross. S., “Survivorship bias in performance studies.” The review of financial studies. Vol. 5, No. 4, (1992):553-580.
Brown, S. J., Goetzmann, W. N., “Performance Persistence”, Journal of Finance, Vol. 50, No. 2, (1995):679-698.
Brown, S., Yan Du D., Rhee S. G., Zhang L, "The returns to value and momentum in Asian markets." Emerging Markets Review, no. 2 (2008):79-88.
Cao, B., He, W., "The Performance of Chinese Open End Stock Mutual Funds: A First Look." In Midwest Finance Association 2012 Annual Meetings Paper. 2011.
Capaul, C., Rowley I., Sharpe W., "International value and growth stock returns." Financial
Analysts Journal (1993): 27-36.
Carhart, M. “On Persistence in Mutual Fund Performance” Journal of Finance 52, (1997): 57-82.
Chen, J., Hong, H., Huang, M., Kubik, J. “Does Fund Size Erode Mutual Fund Performance? The Role of Liquidity and Organization.” American Economic Review, Vol. 94, No. 5, (2004):1276-1302.
31 Christensen, M.,”Evaluating Danish mutual fund performance” Working paper, The Aarhus School of Business, (2003).
Connor, G., Sehgal, S., "Tests of the Fama Model in India," FMG Discussion Papers dp379, Financial Markets Group. (2001).
Cremers, K. J. M., Petajisto A., “How Active Is Your Fund Manager? A New Measure That Predicts Performance” Review of Financial Studies, 22(9), (2009):3329-3365.
Dahlquist, M., Engström S., Söderlind P., "Performance and characteristics of Swedish mutual funds." Journal of Financial and quantitative Analysis (2000): 409-423.
Damodaran, A., "What is the risk free rate? A Search for the Basic Building Block." A Search for
the Basic Building Block, (2008).
Dellva, W., Olson, G., “The Relationship between Mutual Fund Fees and Expenses and Their Effects on Performance.” The Financial Review 33, (1998):85-104.
Elton, E., Gruber M., Blake C., "The persistence of risk-adjusted mutual fund performance." NYU
Working Paper No. FIN-95-018, (1995).
Engstrom, S., “Costly information, diversification, international mutual fund performance” Pacific-Basin Finance Journal 11(4), (2003):463-482.
Fama, E., French K., "The cross-section of expected stock returns" Journal of Finance 47, (1992):427-465.
Fama, E., French K., “Common risk factors in the returns on stocks and bonds” Journal of Financial Economics 33, (1993):3-56.
Fama, E., French K., “Multifactor explanations of asset pricing anomalies.” Journal of Finance 51, (1996):55-84.
32 Fama, E., MacBeth, J., "Risk, Return, and Equilibrium: Empirical Tests." Journal of Political Economy. (1973).
Ferreira, M., Keswani A., Miguel A., Ramos S., "The Determinants of Mutual Fund Performance: A Cross-Country Study." Review of Finance (2012).
Gil-Bazo, J., Martinez, M.A., “The Black Box of Mutual Fund Fees.” Revista de Economia Financiera, Vol4, December, (2004):54-82.
Grinblatt, M., Keloharju M., "The investment behavior and performance of various investor types: a study of Finland's unique data set." Journal of Financial Economics 55, no. 1 (2000): 43-67.
Grossman, S., Stiglitz, J., “On the Impossibility of Informationally Efficient Markets”, American Economic Review 70, (1980):393-408.
Liew, J., Vassalou M., "Can book-to-market, size and momentum be risk factors that predict economic growth?" Journal of Financial Economics 57, no. 2 (2000): 221-245.
Meyer, K., "Complete Financial Product: Alternative Energy Fund." Available at SSRN
1606503 (2010).
Nanda, V., Wang, Z. “Why do aggressive payout policies reduce fund discounts–is it signaling, agency costs, or dividend preferences?” Working Paper, University of Illinois, (2008).
Otten, R., Bams D., "European mutual fund performance." European Financial Management 8, no. 1 (2002): 75-101.
Perold, A., Salomon, R. Jr. “The right amount of assets under management”, Financial Analysts Journal, Vol. 47, No. 3, (1991):31-39.
Pollet, J., Mungo W. “How does size affect mutual fund behavior?” Working Paper, Hong Kong University of Science and Technology, (2007).
Rosenberg, B., "Extra-market components of covariance in security returns.” Journal of Financial
33 Roy, B., Deb, S. S., The Conditional Performance of Indian Mutual Funds: An Empirical Study (2003).
Sharpe, W., "Mutual fund performance." Journal of business (1966): 119-138.
Stein, J., “Information Production and Capital Allocation: Decentralized versus Hierarchical Firms.” Journal of Finance, 57(5), (2002):1891–921.
Su, R., Zhao, Y., Yi, R., Dutta, A., "Persistence in Mutual Fund Returns: Evidence from China."
International Journal of Business and Social Science, 3(13), (2012)
Treynor, J., Mazuy K., "Can mutual funds outguess the market?" Harvard Business Review 44, no. 4 (1966): 131-136.
Tang, K., Wang W., Xu R., "Size and performance of Chinese mutual funds: the role of economy of scale and liquidity." Pacific-Basin Finance Journal (2011).
Williamson, O., "Logic of Economic Organization, The." JL Econ. & Org.4 (1988): 65.
Wermers, R., "Mutual fund performance: an empirical decomposition into stock-picking talent, style transaction costs and expenses", Journal of Finance 60, (2000):1655-1695.
34
Su, Roger, Ying Zhao, Ronghua Yi, and Amitabh Dutta. "Persistence in Mutual Fund Returns: Evidence from China."
Appendices
Appendix 1. Summary statistic of mutual funds
The returns of the sample of mutual funds are calculated as an average for the monthly returns of the funds included in each country. The results are presented below and include average, median, max, min, standard deviation over the sample period 2004-2010, which represents 84 observations. Also, the table presents the number of available funds per country, the total assets under management of the funds from each country and also the average age for each national portfolio of funds.
Brazil China India Russia South Africa
Mean -0.005 0.018 0.01 0.011 0.013 Median -0.002 0.022 0.015 0.02 0.016 Maximum 0.052 0.150 0.201 0.195 0.142 Minimum -0.381 -0.15 -0.15 -0.316 -0.196 Std. Dev. 0.044 0.059 0.05 0.0739 0.062 Observations 84 84 84 84 84 Number of funds 711 18 228 20 93 Total assets 810,764 66,601 2,717,923 541,203 439,033
Avg. age (yy-mm) 14/03 10/09 14/04 12/08 14/03
Appendix 2. The performance of BRICS mutual funds during 2004-2010
In the table below are represented the yearly and total raw returns of the samples of mutual funds for each BRICS country during the seven year period. The data is not adjusted for mutual funds fees.
Year Brazil China India Russia South Africa
35 Appendix 3. Summary statistics for Indices return.
The table reports summary statistics of the indices used for the regressions. In the summary below are listed: Bovespa Index, Shanghai Stock Exchange Composite Index, SENSEX Index, MICEX Index, FTSE/JSE AFRICA TOP40 Index during the sample period 2004-2010.
Brazil China India Russia South Africa
Mean 0.03 0.03 0.02 0.02 0.02 Median 0.03 0.03 0.02 0.03 0.02 Maximum 0.24 0.23 0.28 0.31 0.18 Minimum -0.33 -0.14 -0.24 -0.33 -0.25 Std. Dev. 0.1 0.07 0.08 0.11 0.08 Observations 84 84 84 84 84
Appendix 4. The performance of BRICS market indices during 2004-2010
In the table are reported the yearly and total raw returns of the market indices selected as benchmarks for each BRICS country during the seven year period.
Year Brazil China India Russia South Africa
2004 28.8% -15.4% 18.7% 13.0% 42.0% 2005 45.9% -6.0% 37.3% 76.9% 28.8% 2006 44.6% 138.3% 49.3% 82.9% 24.7% 2007 72.9% 110.1% 65.2% 20.1% 18.3% 2008 -55.3% -63.0% -61.5% -72.7% -46.0% 2009 144.9% 79.9% 89.4% 116.5% 64.0% 2010 5.9% -11.2% 22.2% 21.1% 28.4% 2004-2010 444.2% 135.6% 258.5% 213.8% 206.5%
Appendix 4. HML risk factor summary statistics
The table reports the summary statistics for HML (high minus low) risk factor or value factor used for regressions in Fama French and Carhart models.
Sample: 2004M01 2010M12
36 Appendix 5. MOM risk factor summary statistics
The table reports the summary statistics for momentum risk factor or value factor used for regressions in Carhart model.
Sample: 2004M01- 2010M12
Brazil China India Russia South Africa Mean 0.007 0.0007 0.11 0.001 0.008 Median 0.004 -0.0001 0.15 -0.0009 0.01 Maximum 0.04 0.01 0.44 0.01 0.02 Minimum -0.009 -0.009 -0.77 -0.005 -0.02 Std. Dev. 0.01 0.004 0.26 0.005 0.01 Observations 82 82 82 82 82
Appendix 6. SMB risk factor summary statistics
The table reports the summary statistics for SMB (small minus big) risk factor or value factor used for regressions in Fama French and Carhart models.
Sample: 2004M01- 2010M12
Brazil China India Russia South Africa
Mean 0.003 -0.006 -0.003 -0.004 0.001 Median 0.0004 -0.006 -0.003 -0.005 0.004 Maximum 0.25 0.08 0.008 0.03 0.03 Minimum -0.03 -0.07 -0.02 -0.04 -0.05 Std. Dev. 0.03 0.03 0.005 0.02 0.02 Observations 84 84 84 84 84
Appendix 7. Risk-adjusted performance measures of BRICS mutual funds Sharpe Ratio Treynor Ratio Jensen's Alpha
Brazil 0.27 0.03 0.003
China 0.26 0.04 0.004
India 0.14 0.01 -0.002
Russia 0.1 0.01 -0.004
37 Appendix 8. Correlation between the portfolios of mutual funds from BRICS
Sample: 2004M01 2010M12
Correlation Brazil China India Russia S. Africa
Brazil 1.000
China 0.308 1.000
India 0.711 0.395 1.000
Russia 0.920 0.241 0.684 1.000
S. Africa 0.653 0.385 0.715 0.525 1.000
Appendix 9. Correlation between BRICS market indices
Sample: 2004M01 2010M12
Correlation Brazil China India Russia S. Africa
Brazil 1.000
China -0.237 1.000
India 0.757 -0.245 1.000
Russia 0.739 -0.211 0.591 1.000
S. Africa 0.808 -0.321 0.695 0.749 1.000
Appendix 10. Mutual fund characteristics
The table contains the average values for the characteristics used in the cross sectional analysis in the 2004-2010 sample period.
Return (monthly) AGE (months) TNA ($ million) Beta 𝞼 (monthly) Observations
