Performance of mutual funds in developed and
emerging countries: using DAPM model.
Tessa van Olphen 10002405
University of Amsterdam
Msc Business Economics, Finance Track Master Thesis
30-‐09-‐2014
Supervisor: Prof. Zou
Abstract
This paper provides empirical results on whether managers of mutual funds do outperform the market in both upper and lower market conditions by having selective abilities and/or market timing abilities, and whether this outperformance is persistent on the short-term. The dataset consists of mutual funds for the four most developed countries, measured by the HDI index, and the BRIC countries, in the period 1998-2013. The results suggest that mutual fund managers do not have any selective abilities, but do have market timing abilities. This outperformance is persistent and positive, except in periods of crises.
1. Introduction
In 1774 the first mutual fund was created by Adriaan van Ketwich in the Netherlands. It was called ‘Eendragt Maakt Magt’, which means “unity creates power”. The motivation to create a mutual fund was to increase diversification for investors with limited means (ICI, 2013). Since the first mutual fund, the mutual fund industry has experienced extreme increase in popularity. The industry has grown from $6,895,127 millions of total net assets in 1997 to $24,131,866 millions of dollars in 2012 (ICI, 2003; ICI, 2013).
In order to add value for the investor, the manager of a mutual fund needs to build a portfolio that is expected to generate a return that
outperforms the market. Therefore, the investor expects that the mutual fund managers have some selective abilities (stock picking abilities) and market timing abilities. These abilities should lead to persistent outperformance of mutual funds compared to their benchmarks. If mutual funds are able to outperform the market, which explains the increasing popularity of mutual funds, this will be a violation of the efficient market hypothesis. The efficient market hypothesis states that stock prices incorporate and reflect all relevant information (Fama, 1965). In other words nobody can beat the market.
Because the outperformance of the market would violate economic theory the abilities of mutual fund managers and the persistency of the performance of mutual funds have been discussed by many researchers already. However, there is still no unambiguous explanation of the still increasing popularity of mutual funds. This paper will contribute to the existing literature in the following ways. First, this paper will test for both the abilities of the mutual fund managers (selective ability and market timing ability) as well as the persistency of the performance of mutual funds. This will give an answer to why investors choose to invest in mutual funds. Surprisingly this is not yet done in previous papers. The second contribution is that this paper will separate the upper and lower market, which can lead to a better
insight. Especially, on the effects that a crisis can possibly have on the overall performance of mutual funds. The market is an upper market when the total gross return of the market is higher than the riskfree interest rate and the market is a lower market when the market’s gross return is lower than the riskfree interests rate. The aim of this thesis is to answer the following research question.
RQ: Can mutual fund outperform the market in both upper and lower
markets based on manager’s abilities? And is this outperformance persistent?
To answer this research question I have formed three hypotheses based on previous empirical papers and theories. These hypotheses will be tested in this empirical thesis, and are formulated below.
H1: The managers of mutual funds have no selective abilities in both upper and
lower markets.
H2: The managers of mutual funds have positive market-timing abilities. H3: The persistency of mutual fund performance in both upper and lower
markets is positive in the short-term.
To provide an answer to the three hypotheses the Dichotomous Asset Pricing Model (DAPM) will be used. The DAPM is developed by Zou (2006) and the model provides a theoretical foundation for separating the expected excess return (mutual fund gross return minus risk-free interest rate) for upper and lower markets conditions. Reasons for choosing the DAPM instead of one of the common performance measures that are based on the Capital Asset Pricing Model (CAPM) are the following. First, the DAPM improves the
CAPM’s pricing accuracy while retaining its theoretical insight. Second, the DAPM has a weaker set of assumptions compared to the CAPM, namely: homogenous beliefs, two fund-separation, and no approximately arbitrage opportunities. Third, the DAPM makes use of the best bèta, which gives the opportunity to separate the performance of mutual funds when the market is an upper market and when it is in a lower market. The final reason is that the DAPM model not only captures the ability to measure the selective abilities of mutual fund managers, but also the market-timing ability of managers, and the persistency in performance of mutual funds, while traditional models will only be able to look at one aspect.
The data used in this thesis is collected from two databases, namely DataStream and the Lipper database for the time period 1998 to 2013. The focus is on the four most developed countries (US, Australia, Norway and the Netherlands), measured by the Human Development Index (HDI) and on the BRIC countries (Brazil, Russia, India and China). The data sample consists of closed-end equity mutual funds. Therefore the performance of mutual funds can be compared to the MSCI index for each country.
The empirical results in this paper are presented in Table 2 and Table 3, which are listed in Appendix 1.2 and 1.3. Based on these results none of the zero-hypotheses can be rejected. Therefore, managers of mutual funds do not have selective abilities, but do have market timing abilities. Due to the market timing abilities mutual funds are able to beat the market on average. The results in Table 3 show that there is positive short-term persistency in mutual fund performance. In other words, winners (losers) stay winners (losers) over a time period of one year.
This thesis continues as follows. In section 2 the development of the mutual fund industry is presented. In section 3 a recapitalization of previous empirical papers on evaluation of mutual fund performance is presented and the hypotheses are formed. This section is split into three subsections, namely empirical papers on selective ability of fund managers, market-timing abilities, and persistency of mutual fund performance. In section 5 the DAPM model
and the research methodology is explained more in detail. Section 6 will explain data collection and the main descriptive statistics. Section 7 will present the main empirical results of this thesis. Section 8 holds an overall conclusion. At the end of the paper the reference list and the Appendix is given.
2. The development of the mutual fund industry
In 1774 the first mutual fund was created by Adriaan van Ketwich in the Netherlands. It was called ‘Eendragt Maakt Magt’, which means “unity creates power”. The motivation of the creation of the mutual fund was to add
diversification for investors with limited means (ICI 2013). Figure I shows the growth of the mutual fund industry over the period 1997 to 2012. The mutual fund industry has grown from $6,895,127 millions of total net assets in 1997 to $24,131,866 millions of dollars in 2012 (ICI 2003 and ICI 2013).
Figure I: The growth of the mutual fund industry in total net asset value. The NAV is obtained
from the sources: ICI, 2003 and ICI, 2013.
6,9 11,1 10,5 23,4 17,0 22,1 21,5 24,1 0 3 6 9 12 15 18 21 24 27 30 N A V (triillions of US dollars) Year
Of the $24,13 trillion of total net assets in 2012 only $2,28 trillion is of emerging countries, which is only 3.3%. A developed country is a country that has a sovereign state with a highly developed economy and an advanced technological infrastructure relative to other countries. Emerging markets are countries that are currently heavily developing themselves. The MSCI World Index covers 23 developed market countries and 20 emerging countries (MSCI). In figure II the top 10 of developed countries in the mutual fund industry in 2012 and their share in net asset value (NAV) is presented, while in figure III the top 10 of emerging countries is presented.
Figure II: Developed countries Figure III: Emerging countries
Figure II: The share in net asset value of
the 10 most developed countries in the developed mutual fund industry in 2013. The NAV is obtained from the sources: ICI, 2003 and ICI, 2013.
Figure III: The share in net asset value of
the 10 most emerging countries in the emerging mutual fund industry in 2013. The NAV is obtained from the sources: ICI, 2003 and ICI, 2013.
As is shown in figure II, the United States mutual fund market is the biggest among the developed countries with a share of 60 percent of net asset value (NAV). Followed by Australia, which has a share of 8 percent, which
contributed to a total net asset value of $1,667,128 millions. In figure III it is shown that Brazil is a key player in the mutual fund industry for emerging countries in 2012. It has grown from $180,606 in net asset value in 1997 to
$1,070,988 in 2012. The second biggest mutual fund market is China with a share of 19 percent.
Over the last 15 years the mutual fund industry has grown enormously, especially the emerging countries. In figure IV the change in NAV shown in percentages for developed countries and emerging countries is shown.
Figure IV: The percentage change in net asset value for developed and emerging countries. The NAV is obtained from the sources: ICI, 2003 and ICI, 2013.
Overall developed market and emerging markets are moving parallel over time. However, the line of emerging countries is showing higher peaks and downs than the developed country line. In 1998 emerging countries have grown by 58.92 percent compared to developed countries that have grown by 30.74 percent. The decline after 1989 can be caused by the Asian crisis. In 2000 developed countries even drop further. One possible reason for this is the Internet bubble that caused a mild recession in 2000. Another interesting peak is in 2006, where there is a suddenly grow in the emerging countries compared to the developed countries. The main reason for this growth is that before 2007 there was no data available for the Chinas mutual fund market compared to the
-‐40 -‐20 0 20 40 60 80 100 % c hange Year
Figure IV: Change in NAV
$434,063 in 2007. The volatile market since 2007 is most likely caused by the recent financial crisis.
3. Literature review of previous empirical papers
In this chapter previous empirical papers on the evaluation of performance of mutual funds are presented. To give a clear outlay of the previous literature this section is split into three subsections. The first subsection presents the previous literature on the selective abilities of mutual fund managers. The following subsection gives a literature review on the market timing of mutual fund
managers. In the last subsection previous empirical papers on the persistency of mutual funds are presented. After each subsection a hypothesis is formed based on the previous literature and theories.
3.1 Mutual fund Managers’ selective abilities / stock picking abilities:
When a manager has selective abilities he/she purchases securities, which in his/her opinion will have a higher value now or in the future and will sell securities, which in his/her opinion will be worth less now or in the future. However, this is not in line with the Efficient Market Hypothesis (EMH), which states that security prices incorporate all relevant information (Fama, 1965). In other words based on the EMH nobody can beat the market. Because
outperformance of the market would violate economic theory there has been a strong interest in the selective abilities of mutual fund managers. Most common models that were used to test for selective abilities in previous papers were the Sharpe ratio, Jensen’s alpha, Three-factor model of Fama and French, and the Carhart model.
Sharpe (1966) was one of the first who examined mutual fund
performance. He used a data sample of 34 open-end mutual funds in the time period 1954 to 1963. He developed the Sharpe ratio and the Dow Jones index was
used as the market benchmark. The Sharpe ratio tells us whether a portfolio’s returns are due to good investment decisions or a result of excess risk taking. Sharpe (1966) showed that the Sharpe ratio for each mutual fund portfolio was lower than the Sharpe ratio of the Dow Jones index. This indicated that the mutual funds portfolios in the period 1954 to 1963 were not able to outperform the market. And therefore, mutual fund managers might not have selective abilities, which is in line with the EMH.
Jensen (1968) used a sample of open-end mutual funds from the period 1945 to 1964. In this paper the Jensen’s alpha was used to measure whether fund managers could beat the market. The Standard & Poor’s 500 Stock Price Index was used as a benchmark for the US market. Jensen’s results indicated that the 115 mutual funds on average were not able to outperform the market. Therefore, he concluded that mutual fund managers on average do not have selective abilities.
In the paper of Ippolito (1989) an analyses is made of the performance of 143 US mutual funds for the time period 1965 to 1984. In this paper there are multiple market benchmarks selected, namely: New York Stock Exchange index (NYSE), the Standard & Poor 500 stock index, and Standard & Poor 500 stock – Salomon Brothers long-term bond market index. The study showed that
Jensen’s alpha is positive for 87 basis points based on the NYSE index, 81 basis points based on the S&P 500 stock index, and 248 basis points based on the S&P 500 stock – Salomon Brothers long term market index. Therefore, Ippolito
concludes that mutual funds outperform on average the market, which indicates that mutual fund managers have selective abilities. Thus, the EMH is violated
Grinblatt and Titman (1989) examined whether mutual fund managers have superior selection abilities to outperform the market. They made use of the performance measurement Jensen’s alpha (1968) and used a dataset of US
mutual funds for the time period 1975 to 1984. The dataset is controlled for survivorship bias. Survivorship bias is the tendency for mutual funds with poor performance to be dropped by mutual fund companies; this results in an
that survivorship bias had a relative small (0.5% per year or less) impact on the abnormal returns of mutual funds. Therefore, they concluded that mutual fund managers have enough selecting ability to outperform the market.
Malkiel (1995) made use of the Jensen’s alpha (1968) to measure the performance of mutual funds in the US for the sample period 1971 to 1991. Malkiel study finds that the survivorship bias is considerably more important than Grinblatt and Titman (1989) suggested. The results of the paper shows a negative and significant Jensen’s alpha, which confirms that the mutual funds in general were not able to outperform the market during the sample period. They also studied the persistency in performance during this sample period and found that during the 1970s there existed persistency in performance, but did not
found consistency in fund returns during the 1980s.
In the paper of Otten and Bams (2000) an overview of the European mutual fund industry is presented. A survivorship bias controlled sample of 506 open-end equity mutual funds from France, Italy, Germany, the Netherlands and the UK for the period 1991 to 1998 is collected. They made use of the Carhart four-factor asset pricing model (Carhart 1997). They found that
European mutual funds, especially small cap funds, are able to generate positive alphas after the costs are deducted. When management fees are not deducted in four out of five countries mutual funds have outperformed the market.
Therefore, they conclude that European mutual fund managers have selection ability.
Wemers (2000) also finds that most stocks held by mutual funds
outperform the market benchmark during the period 1975 to 1994. The author uses the value weighted CRSP index as the market benchmark. Wemers makes just as Otten and Bams (2000) use of the four-factor model of Carhart (1997). They found that the mutual funds outperform the index by 130 basis points, where 71 basis points are due to stock picking talent of managers.
In the paper of Sapar and Ravindran (2003) the mutual funds
performance in India is examined for the period 1998 to 2002. During this time period the market was a bear market. They made use of the following measures:
relative risk performance index, risk-return analysis, Treynor’s ratio, Sharpe ratio, Jensen’s alpha and Fama and French three-factor model. The total
number of Indian mutual funds was 433 out of which 311 were open-end mutual funds and 87 closed-end mutual funds. They concluded that 58 of the open-end mutual funds were outperforming the market during the four-year bear period.
Xiu-juan and Shou-yang (2007) did an empirical paper on the Chinese mutual fund performance in the years 2004 and 2005. They included 24 open-end mutual funds and 54 closed-open-end mutual funds. It was found that the
Chinese mutual funds are closely related to the Chinese equity market and that most of the Chinese market benchmarks are inefficient. And therefore, mutual funds were able to outperform the market.
Eling and Faust (2010) have investigated the performance of 243 hedge funds and 629 mutual funds that focus on emerging markets in 1995 to 2008. They find that mutual funds due not outperform or underperform relative to the market benchmark.
Bialkowski and Otten (2010) have investigated the Polish mutual fund performance using a survivorship bias controlled sample of 140 funds. They made use of the Carhart four-factor model and concluded that on average Polish mutual funds are not able outperform the market.
Because previous papers do not give an unambiguous answer on the question whether mutual fund managers have selective abilities. This paper expects that the efficient market hypothesis is not violated and therefore the first hypothesis is formulated.
H1: The managers of mutual funds have no selective abilities in both upper and
3.2 Mutual fund manager’s timing abilities
Market timing can be defined by the ability of the mutual fund manager to
control the market exposure. The goal of the manager is to have a large exposure when the market premium is positive (upper market) and a smaller exposure in lower markets. Therefore, the managers of mutual funds wants to invest when the market is moving up and divest when the market is going down. This section will give an outlay of papers that studied market timing in previous years.
In the paper of Hendriksson (1984) market-timing abilities of mutual fund managers are studied using a data sample that consisted 116 open-end mutual funds over the period 1968 to 1980. They concluded that mutual fund managers are not able to follow an investment strategy that is successful in timing the return on the market portfolio.
Kon (1983) proposes an empirical methodology to measure market-timing ability of mutual fund managers. A data sample of 37 mutual funds was collected for the time period 1960 to 1976. He finds that in that time period there is
evidence of significant market-timing abilities of mutual fund managers, which leads to higher mutual fund performance. However, the results were not
inconsistent with the efficient market hypothesis (no abnormal returns). Chang and Lewellen (1984) test for superior market-timing ability or selection-ability in managed portfolios. They used a sample that consists of 67 mutual funds for the time period 1971 to 1979. The value-weighted stock index of CRSP was used as the market benchmark. Their results indicate that mutual fund managers have superior market-timing abilities, but that managers do not have selective abilities.
Lee (1990) empirically examines the market-timing ability and the selective ability of mutual fund managers. They find that the market-timing ability of the mutual fund managers is positive. However, his results also
indicated that the mutual funds have the ability to outperform the market, which is in contrast of the efficient market hypothesis.
In the empirical paper of Bollen and Busse (2001) they researched market-timing abilities of mutual fund managers by using daily tests instead of using monthly returns like previous papers did. They show that using daily tests
instead of monthly tests results in stronger evidence of timing ability. Therefore, they suggest that mutual funds may possess more market timing ability than is previously documented in previous empirical papers.
All the articles suggest that managers have the ability to time the market, which indicates that the bèta in upper markets is significantly higher than the bèta in lower markets. Therefore, we have formed the second hypothesis.
H2: The managers of mutual funds have positive market-timing abilities.
3.3 Mutual fund performance persistency
Mutual fund persistency in performance can be described as the phenomenon that winners (losers) stay winners (losers) over time. If mutual fund performance is persistent this indicates that high performance is not based on
coincidence/luck, but on managers skills. Therefore, it would be profitable for investors to invest in mutual funds that have generated high returns in the past assuming the manager stays the same. This section will give a chronological outlay of previous empirical papers on mutual fund performance persistence.
Grinblatt and Titman (1992) were one of the first authors who researched mutual fund performance persistency. They used a sample of 279 mutual funds that existed from 1974 to 1984. To analyze the persistency in performance they used a three-step procedure. First they split their data sample into two five-year subperiods. Second they computed the abnormal returns for each mutual fund using Jensen’s alpha performance measure and the eight-portfolio market benchmark. The final step is to estimate the bèta of absolute returns computed
from the last five-year subperiod on the first five-year subperiod. Their results indicate that persistency in mutual fund performance is positive in their sample.
The paper of Hendricks, Patel and Zeckhauser (1993) studied the
persistency in mutual fund performance by using investment strategies that are based on mutual funds with ‘hot hands’. Mutual funds have ‘hot hands’ if the fund managed to had superior performance in the year prior to the investment. Their data set contains mutual fund performance over the period 1974 to 1988. They made use of Jensen’s alpha to measure mutual fund performance. They find that investing in mutual funds that have ‘hot hands’ will lead to significant outperformance of the market benchmark. However, they conclude that the ‘hot hand’ effect was not an important explanation for persistency in mutual fund performance.
Carhart (1997) studied persistency in US mutual fund performance using a survivorship-free data sample for the time period 1962 to 1993. They made use of two performance measures, namely the three-factor Fama and French measure and the Carhart four-factor model. Using the strategy of buying securities of the top-decile mutual funds of the last year and selling last year’s bottom-decile securities yields in return of 8% per year. Carhart reports that the bottom-decile securities explain most part of the persistency in US funds. However, on the long run, this strategy will lead to lower yields. Therefore Carhart concluded that there exists positive short-run persistency, but no long-run persistency in performance.
In the empirical paper of Bollen and Busse (2005) the short-term persistency in mutual fund performance is researched. They made use of 230 mutual funds that existed during the time period 1985 to 1995. They rank funds every quarter by their risk-adjusted return. They find that mutual funds that were in the top-decile generated significant abnormal returns in the
post-ranking quarter of 0.25 to 0.39. Therefore, a positive persistency in mutual funds exists during their data sample.
In the paper Huij and Post (2010) the performance persistency of
137 equity mutual funds over the period 1993 to 2006. They rank the mutual funds by return over the past quarter. They find that mutual funds have positive performance persistency. In emerging countries the top-decile mutual funds have the largest contribution to this persistency in mutual fund performance, which is in contrast with the paper of Carhart (1997).
Most articles mentioned that suggest there exists a positive short-term persistency in mutual fund performance. Therefore, the third hypothesis is formed as follows:
H3: The persistency of mutual fund performance in both upper and lower
markets is positive in the short-term. .
4. Methodology:
In this thesis the Dichotomous Asset Pricing Model (DAPM) will be used to answer all three hypotheses. Zou (2006) developed the DAPM, which provides a theoretical foundation for separating the expected excess return of the mutual fund (mutual fund return minus risk-free interest rate) conditional for both upper and lower markets. The market is an upper market when the gross return of the market is larger than the riskfree interest rate. When the gross return of the market is lower than the riskfree interest rate the market is a lower market.
The reasons for using the DAPM model instead of one of the common performance measures, which are based on the Capital Asset Pricing Model (CAPM) are the following. First the DAPM improves the CAPM’s pricing
accuracy, while retaining its theoretical insight. Second, the DAPM has a weaker set of assumptions compared to the CAPM, namely: homogenous beliefs, two fund-separation, and no approximately arbitrage opportunities. The key
expected losses over a reference level instead of in terms of the standard deviation of returns. The two fund-separation implies that the market is both mean-variance efficient as well as gain-loss efficient (Ross 1978). The
assumptions hold when the market is gain-loss efficient as well as mean-variance efficient. Third, the DAPM makes use of the best bèta, which gives the
opportunity to separate the performance of mutual funds when the market is an upper market and when it is in a lower market. Therefore, the predicted
relationship between the expected excess return of the market and the bèta are strictly stronger than the CAPM predictions. The final reason is that the DAPM model not only captures the ability to measure the selective abilities of mutual fund managers, but also the market-timing ability of managers, and the
persistence in performance of mutual funds, while traditional models will only be able to look at one aspect. Mathematically the DAPM is stated as follows.
Upper market: E(x!,!) = a!+ β!!E(x!,!), β!! = E xi,t,xm,t E (xm,t)2
;
Where 𝑥!,! = 𝑅!,!− 𝑟!, 𝑖𝑓 𝑅!,! > 𝑟! 0 , 𝑖𝑓 𝑅!,! ≤ 𝑟! 𝑥!,! = 𝑅!,! − 𝑟!, 𝑖𝑓 𝑅!,! > 𝑟! 0 , 𝑖𝑓 𝑅!,! ≤ 𝑟! Lower market: 𝐸(𝑥!,!) = 𝑎!+𝛽!!𝐸 𝑥 !,! , 𝛽!! = 𝐸 𝑥!,!, 𝑥!,! 𝐸((𝑥!,!)!) Where 𝑥!,! = 𝑟!0 , 𝑖𝑓 𝑅− 𝑅!,!, 𝑖𝑓 𝑅!,! ≤ 𝑟! !,! > 𝑟! 𝑥!,! = 𝑟!− 𝑅!,!, 𝑖𝑓 𝑅!,! ≤ 𝑟! 0 , 𝑖𝑓 𝑅!,! > 𝑟!The market is a upper market when the market gross return of the market is larger than the riskfree interest rate and is a lower market when the market gross return is smaller than the riskfree interest rate. Alpha (𝛼!) measures the
absolute return over the expected rate of return. Therefore, alpha indicates whether mutual fund managers have selective abilities, which lead to
outperformance of the market benchmark. The upper-market bèta (𝛽!!) measures
the systematic risk in an upper market and the lower-market bèta (𝛽!!) measures
the systematic risk in a lower market.
Therefore, to answer the first hypothesis (The managers of mutual funds
have no selective abilities in both upper and lower markets.) we need to look at the
absolute return over expected rate of return (𝛼!). 𝛼! is tested using Ordinary
Least Squares (OLS) regression analyses. If 𝛼! is significantly larger than zero
the mutual fund manager have selective ability and outperforms the market, but when 𝛼! is smaller or equal t0 zero the mutual fund manager has no selective
ability. Thus, the zero-hypotheses and the alternative hypotheses can mathematically stated as follows:
𝐻!: 𝑎! ≤ 0 𝐻!: 𝑎! ≤ 0
𝐻!: 𝑎! > 0 𝐻!: 𝑎! > 0
To answer the second hypothesis (The managers of mutual funds have
positive market timing abilities.) we test the difference between the upper-market
and the lower-market bèta using again OLS. If this difference is positive it will indicate that the manager has taken higher systematic risk in upper markets and lower systematic risk in lower market. This leads to higher mutual fund
performance. Therefore, this difference in upper and lower bèta indicates that the managers have market timing skills. Therefore, the zero-hypothesis and the alternative hypothesis can mathematically stated as follows:
𝐻!: 𝛽!!− 𝛽
To answer the third hypothesis (The persistency of mutual fund performance
in both upper and lower markets is positive in the short-term.) we test whether
winners (losers) stay winners (losers) in both upper and lower markets. We define winners (W) as mutual funds that have higher or equal gross returns (𝑅!)
than the median and losers (L) as funds that have lower gross returns than the median. We will only look at a one-year horizon (short-term persistency). The first step to test the short-term persistency of mutual fund performance is to make a contingency table, which represents WW, WL, LL, and LW. The second step is to test whether winners (losers) significantly have a higher change (p) to stay winners (losers). Therefore the hypotheses are mathematically stated as follows.
𝐻!: 𝑝!,! > 0.5 𝐻!: 𝑝!,! ≤ 0.5
T0 test these hypotheses the Malkiel (1995) test is used. It follows a standard normal distribution and is calculated as follows:
𝑀 = (𝑊𝑊 − 𝑊𝑊 + 𝑊𝐿 ∗ 0.5) ( 𝑊𝑊 + 𝑊𝐿 ∗ 0.5 ∗ 0.5)
5. Data:
This section will explain which data is needed and which databases are used. At the end of this section some descriptive statistics are mentioned. The statistics are presented in Table 1, which is listed in Appendix 1.1.
The data sample used in this thesis is collected through two databases, namely DataStream and the Lipper database. The databases were connected through the Lipper codes of the mutual funds. The data sample consists of the four most developed countries (US, Australia, Norway and the Netherlands), measured by the Human Development Index (HDI) and the BRIC countries
(Brazil, Russia, India and China). The HDI index ranks levels of social and economic development of countries based on four criteria: life expectancy at birth, mean years of schooling, expected years of schooling and gross national income per capita.
The sample will only consist of closed-end equity mutual funds with a domestic investment focus. Therefore, the mutual funds can be compared to the performance of the market benchmark of the country (MSCI). The market benchmark is the MSCI benchmark and is collected through DataStream. In the data sample both ‘survivor’ mutual funds and ‘dead’ mutual funds will be
included to prevent survivorship bias. Survivorship bias is the tendency for mutual funds with poor performance to be dropped by mutual fund companies. This will result in an overestimation of the past performance of mutual funds. A list of equity mutual funds with a domestic investment focus is collected through the Lipper database for each country.
Based on these lists the total return index for each mutual fund and the market benchmark is collected through DataStream. The return index (RI) shows the growth in value of a share that is held for a specific period of time, assuming that dividends are reinvested to purchase additional units of an mutual fund at the closing bid price. The return index is mathematically formulated as follows:
𝑅𝐼! = 𝑅𝐼!!!∗𝑃!+ 𝐷! 𝑃!!!
After collecting the total return indexes the gross return of the mutual funds and the market benchmark is calculated in the following way:
𝑅!,! = ln
𝑅𝐼!,!
𝑅!,! = ln
𝑅𝐼!,!
𝑅𝐼!,!!! + 1
The riskfree interest rate is collected through DataStream in order to generate the excess returns, which is the gross return minus the riskfree interest risk. A time horizon will be set from 1998 to 2013. Therefore, enough information can be collected for as well upper markets and lower markets (Asian crisis, technology bubble, and financial crisis).
In Appendix 1.1 Table 1 presents some descriptive statistics of the gross market return and the gross mutual fund return for both upper and lower market conditions. The table shows that the minimum value of the gross return of the market is equal to zero in the upper market and the maximum value is equal to zero in the lower market. This is logical because the market is a upper market when the market gross return of the market is larger than the riskfree interest rate, otherwise the gross return is set to zero (𝑅!,! = max (𝑅!,! − 𝑟!, 0)),
and the market is a lower market when the market gross return is smaller than the riskfree interest rate, otherwise the gross return is set to zero (𝑅!,! =
max (𝑟!− 𝑅!,!, 0)). The mean of the gross market return is equal to 0.088 in the upper market and 0.016 in the lower market, while the gross return of mutual funds is equal to 0.077 in the upper market and -0.126 in the lower market.
6. Empirical Results
In this section I will present and analyze the empirical results of this paper. The section is organized to find an answer to the three hypotheses that were formed earlier in this paper. For simplicity reasons I have repeated these hypotheses below.
H1: The managers of mutual funds have no selective abilities in upper and
lower markets.
H2: The managers of mutual funds have positive market-timing abilities. H3: The persistency of mutual fund performance in both upper and lower
markets is positive in the short-term.
The empirical results are presented in Table 2 and Table 3, which are listed in Appendix 1.2 and 1.3. Table 2 presents the regression results of the excess cash return of mutual funds for three data samples, namely a sample that only include the developed countries, a sample only including the BRIC countries, and a total sample that includes all countries. The t-statistics are given in the parentheses and ***, **, * indicate if the coefficient is significant at the 1%, 5% or 10% level. This table gives the opportunity to answer the first and the second hypotheses. Table 3 shows the empirical results on short-term performance persistency of mutual funds conditional for upper and lower markets. The Malkiel t-statistics are presented and ***, **, * indicate if the coefficient is significant at a 1%, 5%m 10% level. This section will continue with the analysis of the two tables to provide an answer to the hypotheses.
To answer the first hypothesis we need to look at the 𝛼! in table 2.
Because as mentioned previous in this paper a manager who has selective
abilities will purchase securities, which in his/her opinion will have higher value now or in the future, and will sell securities which in her/his opinion will be worth less nor or in the future. Therefore, managers that have selective ability will generate abnormal returns for the mutual funds, which is represented in the table by alpha. In other words, if alpha is significantly higher than zero a
manager has selective abilities. As can be seen in Table 2 the alpha in all data samples for both upper and lower market conditions is not significant different
from zero. Therefore, the zero-hypothesis can’t be rejected and the efficient market hypothesis is not violated. And thus, mutual funds will not outperform the market based on selective abilities of mutual fund managers. This empirical result is in line with the results of the papers of Sharpe (1966), Jensen (1968), Malkiel (1995), Wemers (2000), Eling & Faust (2010) and Bialkowski & Otten (2010).
The second hypothesis can be answered based on the last column of Table 2, which represents the difference between the upper market bèta and the lower market bèta. This difference represents market timing abilities of mutual fund managers, because if the manager has this ability the fund will have larger risk exposure when the market premium is positive and a smaller exposure in lower markets. Therefore, if the difference is significantly larger than zero, we can suggest that the managers of mutual funds are capable of timing the market. The results of Table 2 show that when the data sample only consists of
developed countries the market-timing abilities of the fund managers will
increase significantly with 0.234%. However, when the data sample only consists of BRIC countries the difference is positive (0.090), but not significant. In the total sample the market-timing abilities is 0.140% and significant at a 5% level. Based on these results the zero-hypothesis will not be rejected. And thus, we assume that managers of mutual funds have some market-timing skills, which lead to an outperformance of the market.
The answer to the third hypothesis can be found in Table 3 in
Appendix 1.3. In this table short-term persistency performance in both upper and lower market is presented. Short-term performance persistency can be described as the phenomenon that winners (losers) stay winners (losers) over a one year time period. The results display that for almost all the time periods (except 2001-2002, 2002-2003, 2007-2008) there is positive performance
persistency. It is however striking that the time periods that are periods of the beginning of the burst (technology bubble 2001-2003; the financial crisis 2007-2008) have significant negative performance persistency. Based on these
results we conclude that the third hypothesis is not rejected as long as there is no unsuspected changes within the market, like a bursting bubble.
Conclusion
Over the last decade the popularity of mutual funds has increased rapidly. However, until now there is still no unambiguous answer to why this popularity in mutual funds is still rising. In order to create value for the investor, mutual fund managers need to create a diversified portfolio that is expected to generate a return, which outperforms the market. In order to beat the market the mutual fund manager needs to have some selective abilities and/or market timing abilities. These abilities should lead to persistent outperformance of mutual funds. If mutual funds were able to outperform the market on average this would be a violation of the efficient market hypothesis (EMH). The EMH states that stock prices incorporate and reflect all relevant information (Fama, 1965). Therefore, the aim of this paper is to answer the following research question.
RQ: Can mutual funds outperform the market in both upper and lower
markets based on manager’s abilities? And is this outperformance persistent?
To answer this research question we have empirically tested three hypotheses, which are listed below, with a data sample that consist of the four most
developed countries and the BRIC countries. Both survivor funds and dead mutual funds are included to avoid survivorship bias. The data sample is collected for the time period 1998-2013 through the Lipper database and Datastream.
H1: The managers of mutual funds have no selective abilities in both
upper and lower markets.
H2: The managers of mutual funds have positive market timing
abilities.
H3: The persistency of mutual fund performance in both upper and
lower markets is positive in the short-term.
In this thesis the DAPM model, developed by Zou (2006), is used. The empirical results in this paper are presented in Table 2 and Table 3, which are listed in Appendix 1.2 and 1.3. Based on these results we found that none of the zero-hypotheses can be rejected. Therefore, managers of mutual funds do not have selective abilities, but do have market timing abilities. Due to the market timing abilities mutual funds are able to beat the market on average. The results in Table 3 show that there exists positive short-term persistency in mutual fund performance, except in periods of a crisis. In other words, winners (losers) stay winners (loser) over a time period of one year.
8. References:
Białkowski, J., & Otten, R. (2011). Emerging market mutual fund
performance:Evidence for Poland. North American Journal of Economics and
Finance, Vol. 22, pp. 118–130.
Blake, D. and Timmermann, A. (1998). Mutual Fund Performance: Evidence from the UK. European Finance Review, Vol. 2, pp. 57-77.
Bolle, N.P.B. and Busse, J.A. (2001). On the timing ability of mutual fund managers. Journal of Finance, Vol. 56, No.3, pp. 1075-1094.
Bolle, N.P.B. and Busse, J.A. (2005). Short-term persistence in mutual fund performance. The review of Financial Studies, Vol. 18, No.2, pp. 569-597. Cai Jun, Chan K.C., Yamada Takeshi (Summer 1997) ‘The performance of Japanese Mutual Funds’, The Review of Financial Studies, Vol. 10, No.2, pp.237-273.
Campbell, J.Y and Voulteenaho, T. (2003). Bad Beta, Good Beta. Harvard Institue of Economic Research, Harvard University, Discussion paper No. 2016. Carhart, M.M. (1997). On Persistence in Mutual Fund Performance. The Journal
of Finance, Vol. 52, No.1, pp. 57-82.
Chang, E.C and Lewellen, W.G. (1984). Market timing and mutual fund investment performance. The Journal of Business, Vol. 10, No.1, pp. 57-72.
Droms, W.G and Walker D.A. (2001). Performance persistence of international mutual funds. Global Finance Journal, Vol. 12, pp. 237-248.
Eling, M., Faust, R. (2010). The performance of hedge funds and mutual funds in emerging markets. Journal of Banking in Finance, pp. 1993-2009
Fama, E.F. (1965). The Behavior of Stock-Market prices. Journal of Business, Vol. 38, pp. 34-105.
Grinblatt, M. and Titman, S. (1989). Mutual Fund Performance: An Analysis of Quarterly portfolio Holdings. The Journal of Business, Vol. 62, No.3, pp. 313-416. Grinblatt, M. and Titman, S. (1992). The persistence of Mutual Fund
Performance. The Journal of Finance, Vol.47, No.5, pp.1977-1984.
Hendricks, D. and Patel. J and Zeckhauser, R. (1993) Hot Hands in Mutual Funds: Short Run Persistence of Relative Performance 1974-1988. The Journal
Henriksson, R.D. (1984). Market Timing and Mutual Fund Performance: An Empirical Investigation. Journal of Business, Vol. 57.
Huij, J. and Post, T. (2010). On the performance of emerging market equity mutual funds. Emerging markets Review, Vol. 12, pp. 238-249.
ICI (2003). 2003 Investment company factbook: a review of trends and activities in the US investment company industry. 53rd edition.
ICI (2013). 2013 Investment company factbook: a review of trends and activities in the US investment company industry. 53rd edition.
Ippolito, R.A. (1989). Efficiency with costly information: a study of mutual fund performance 1965-1984. The Quarterly Journal of Economics, Vol.104, No.1, pp. 1-23. Jensen, M. (1968). The performance of mutual funds in the period 1945-1964. The
Journal of Finance, Vol. 23, No.2, pp. 389-416.
Kon, S.J. (1983) The Market Timing of Mutual Fund Managers. The Journal of
Business, Vol. 56, No.3, pp. 323-347
Lee, C. and Rahman, S. (1990). Market timing, selectivity, and mutual fund performance: an empirical investigation. Journal of Business, Vol. 63, No.2. Malkiel, B.G. (1995). Returns from investing in equity mutual funds 1971 to 1991.
Journal of Finance, Vol. 50, No.2, pp. 549-572.
Otten, R., Bams, D. (2002). European Mutual fund performance. European
Financial Management, Vol. 8, pp. 75-101.
Otten, R. and Schweitzer, M., A comparison between the European and the U.S. mutual fund industry, Managerial Finance, pp. 14–34.
S. Narayan Rao , M. Ravindran. Performance Evaluation of Indian Mutual Funds, Working paper
www.papers.ssrn.com/sol3/papers.cfm?abstract_id=433100.
Sharpe, W.(1966). Mutual fund performance. Journal of Economics, Vol. 4, pp. 129-176.
Wermers, R. (2000). Mutual Fund Performance: An Empirical Decomposition into Stock- Picking Talent, Style, Transaction Costs, and Expenses. Journal of
Finance, Vol. 55, No.4, pp.1655-1695.
Zhao, X. and Wang, S. (2007). Empirical study on Chinese mutual funds’ performance. Systems Engineering –Theory & Practice, Vol.27, No.3, pp. 1-11.
Zou, L. (2005). Dichotomous Asset Pricing Model. Annals of Economics and
Finance, No. 6, pp. 185-207.
Appendix 1.1
Table 1: This table shows some descriptive statistics of the market gross return and the mutual funds gross return for both upper
and lower markets.
Descriptive statistics
Upper market Lower market
Mean Median Min Max Variance Mean Median Min Max Variance
Sample only developed countries
Market return Mutual fund return
0.093 0.045 0.073 0,042 0 -‐1.533 0.294 1.908 0.010 0.030 -‐ 0.071 -‐0.030 0 0 -‐0.505 -‐1.694 0 0.718 0.018 0.027
Sample only BRIC countries
Market return Mutual fund return
0.114 0.121 0.103 0.132 0 -‐1.435 0.354 1.862 0.021 0.039 0.018 -‐0.235 -‐0.015 -‐0.221 -‐0.638 -‐1.761 0 0.871 0.020 0.029 Total sample Market return Mutual fund return
0.088 0.771 0.891 0.072 0 -‐1.533 0.354 1.908 0.042 0.049 0.016 -‐0.126 0 -‐0.018 -‐0.638 -‐1.761 0 0.871 0.031 0.041
Appendix 1.2
Table 2: This table presents the regression results of the excess cash return of mutual funds on the excess cash return of the
market for both upper and lower markets. The t-statistics are given in the parentheses and ***, **, * indicate if the coefficient is significant at the 1%, 5% or 10% level.
Selective and Market timing abilities of the managers of mutual funds
Upper market Lower market Market timing Alpha 𝛽!! N R-‐Squared Alpha 𝛽!! N R-‐Squared 𝛽!!− 𝛽 !!
Sample only developed countries 0.100 (0.92) 0.561*** (6.31) 14566 0.0829 -‐ 0.015 (-‐1.05) 0.327*** (27.92) 14988 0.0641 0.234* (1.43)
Sample only BRIC countries 0.023 (1.21) 0.302*** (2.98) 7863 0.0023 0.018 (0.89) 0.212*** (3.40) 8056 0.0501 0.090 (1.02) Total sample 0.058 (0.77) 0.409*** (5.56) 22429 0.0157 -‐0.003 (-‐0.58) 0.269*** (9.67) 23044 0.0363 0.140** (1.97)
Appendix 1.3
Table 3: This table presents the short-term performance of mutual funds for the total data sample for upper and lower markets.
The Malkiel t-statistics are presented and ***, **, * indicate if the coefficient is significant at the 1%, 5% or 10% level.
Performance Persistency Time period 1999-‐ 2000 2000-‐ 2001 2001-‐ 2002 2002-‐ 2003 2003-‐ 2004 2004-‐ 2005 2005-‐ 2006 2006-‐ 2007 2007-‐ 2008 2008-‐ 2009 2009-‐ 2010 2010-‐ 2011 2011-‐ 2012 2012-‐ 2013 Upper Market WW 471 717 115 53 78 598 637 522 251 30 32 531 715 646 WL 393 184 345 104 55 388 349 476 1012 18 4 125 73 517 LW 389 184 18,4 21 437 393 354 481 9 12 816 284 278 533 LL 474 890 870 959 522 584 628 518 724 1150 1604 786 717 624 Malkiel – test 2.65*** 17.75*** -‐10.72 -‐4.07 1.99** 6.68*** 9.17*** 1.46* -‐21.41 1.73* 4.67*** 15.85*** 22.87***. 3.78*** Lower Market WW 565 860 138 54 94 717 764 626 301 36 38 637 858 775 WL 471 221 414 124 66 465 419 571 1214 22 5 150 87 620 LW 467 220 22 25 553 471 423 577 11 14 98 340 333 239 LL 569 1068 1044 1151 626 701 754 622 869 1380 1924 943 861 749 Malkiel-‐test 2.92*** 19.43*** -‐11.75 -‐5.25 2.21** 7.33*** 10.03*** 1.59* -‐23.46 1.83* 5.03*** 17.36*** 25.08*** 4.15***