Comparing mutual fund and ETF performance in emerging markets

Comparing Mutual Fund and ETF Performance in Emerging Markets

Abstract

Emerging markets experience a fast growing economy, however, the possibility exists that emerging capital markets do not behave quite efficient yet. When the efficient market hypothesis does not hold one would assume opportunities to obtain risk-adjusted abnormal returns would present themselves. This can be an indication for the presence of abnormal returns in emerging markets compared to developed capital markets. If these abnormal returns would be present, one would assume that active portfolio management would lead to higher risk-adjusted returns than passive portfolio management. In this thesis I will examine whether active portfolio management in developing countries leads to abnormal returns compared to developed countries by comparing mutual funds with ETFs. Mutual funds will represent the active side of portfolio management, whereas ETFs will serve as a proxy for passive portfolio management since they track market indices.

Stella Hak 10592156

Economics and Finance June 2016

2

Table of contents

Introduction 3

Literature review

Efficient Market Hypothesis 5

Emerging markets 6

Active management and the Capital Asset Pricing Model 6

Mutual fund past performance 6

Mutual fund performance measures and persistence 9

ETFs, characteristics and performance 11

Hypotheses 13

General approach and data 14

Regressions 17

Results 19

Conclusions 24

Literature 25

3

Introduction

The main goal of investing is mainly to make a profit. When one invests into a portfolio of assets there are two main investing strategies, either an active management portfolio or a passive market portfolio. One of the ways active managers make a profit is by investing where similar securities are priced differently, buying the underpriced security and selling the overpriced, a process known as arbitrage. Considering the efficient market hypothesis states that all available information is reflected in stock prices, this would make arbitrage opportunities nonexistent. Another way of active management is to attempt to outperform the market through market timing and security selection. Many research has been done for developed markets on the performance of mutual funds and the overall conclusion is that mutual funds, on average, underperform the market index. Since emerging markets are assumed to be less efficient than developed markets one would assume this might indicate the potential existence of abnormal returns.

According to the IMF, in the next 2-3 years, approximately 70% of Global GDP growth will be represented by emerging markets1_{. In contrast to developed markets, which experience}

more stagnant growth, one would presume this would create profitable investment opportunities for excess returns and portfolio diversification. Additionally, diversification benefits would arise through low correlations between developing and developed markets. For these reasons investors find it appealing to invest in emerging markets, either through actively managed mutual funds or passively managed exchange-traded funds (ETFs) and index funds.

In contrast to passively managed funds mutual funds are pooled investments that provide liquidity and allow investors to enjoy economies of scale by investing in a wider variety of assets instead to diversify their risk away (Cuthbertson, Nitzche, and O’Sullivan, 2010). The extended portfolio of a mutual fund provides a greater risk-return relationship compared to a smaller portfolio or even a single security. Mutual funds are actively seeking for ways to ‘’beat the market’’ through superior security selection. Because mutual funds are actively managed they charge management fees which, one would assume, causes lower returns.

Alternatively, while managers in actively managed funds are trying so seek profitable investment opportunities continuously, passively managed index funds aim at mimicking a broad market index. Exchange traded funds (ETFs) are groups of securities that are designed to mirror certain sectors of an index. Since the objective of ETFs is to give exposure to indices, ETFs are useful to investors who want to passively track an index (Hehn, 2005). ETFs are a

4 rapidly growing class of financial products and are often promoted as being tax efficient (Poterba and Shoven, 2012).

This thesis will rely on the assumption of imperfect capital markets in emerging countries to study the possibility of above normal returns with active portfolio management, mutual funds, compared to the passively managed ETFs. It is of interest to study whether the hypothesis either can or cannot be rejected since it will provide us with information about the functioning of emerging markets, thus if developing markets are efficient or not, as well as the possible opportunity for investors to produce above normal returns.

Starting with a literature review in which the failure of the efficient market hypothesis in emerging markets will be justified together with the definition of emerging markets. The Capital Asset Pricing Model (CAPM) and its anomalies will be discussed. Subsequently, past performance and the persistence of this performance of mutual funds will be discussed. Possible obstacles with the evaluation of mutual fund performance will be explained after. Then the characteristics and performance of ETFs will be analyzed. Following, the hypothesis conducted from the literature review will be given. Hereafter the composition of the data and deliberation on how it is obtained will be given. Next the results will be discussed and further interpretation of these results will be given. Finally a conclusion of the overall outcome will be provided and a possible indication for further appropriate research.

5

Literature review

Justification for the assumption of inefficient capital markets

According to the efficient market hypothesis a stock reflects all available information and therefore it should not be possible to be able to consistently outperform the market using active portfolio management. When the efficient market hypothesis holds, prices respond quickly to relevant information and reflect their intrinsic value, hereby making it impossible to achieve abnormal returns. Since developed capital markets are considered to be efficient and highly liquid they are therefore less sensitive to opportunities to obtain risk-adjusted abnormal returns. Thus on average the return on a passive portfolio, which invests in the market index, outperforms the actively managed portfolio when returns have been adjusted for expenses.

The opportunity for abnormal returns is partly dependent on the assumption of inefficient capital markets and is therefore highly important for the validity of this thesis. This assumption is justified by the adaptive market hypothesis by Lo (2004) who acknowledges that the assumption of market efficiency is problematic. Lo provided several implications which motivate the use of active management, among others he states that arbitrage opportunities do arise over time and particular investment strategies might not perform equally in differing markets.

Additionally, a great deal of research has shown that emerging capital markets are located somewhere between weak form and semi strong form of the efficient market hypothesis (Alexakis, Patra, & Poshakwale, 2010; Balaban & Kunter, 1997; Cooray & Wickremasinghe, 2008). Balaban and Kunter (1997) showed the occurrence of significant deviations from the efficient market hypothesis when concerning changes in market liquidity.

Furthermore, Coorey & Wickremasinghe (2008) showed that even the semi strong form might not be supported by empirical evidence because of the high degree of interdependence in emerging stock markets. Sharpe (as cited in Elton, Gruber, & Blake, 1996) who investigated 100 of the largest mutual funds, ranked them on the basis of risk adjusted excess returns (alpha) and analyzed whether past alphas are related to future alphas.

Carhart (1994) arrived at the same conclusion as Sharpe, which is that past performance contains information about future performance (as cited in Elton, Gruber, and Blake, 1996). Aforementioned implies stock prices reflect information varying between information entailed in historic prices and generally available public information, corresponding with a weak or semi strong form of the efficient market hypothesis.

Furthermore, Grossman and Stiglitz (1980) constructed a model in which informed investors pay a cost to obtain information about asset prices in order to earn above normal

6 returns in contrast to uninformed investors. They argued that when information is costly prices cannot reflect all available information (Grossman and Stiglitz, 1980). It also has been

verified that when the efficient markets hypothesis is true and information is costly, competitive markets break down because those who spent resources on information would receive no compensation. Stiglitz mathematically proved that markets are not perfectly efficient and thus prices do not reflect all available information about the stock price and that pricing information happens imperfectly. In this thesis the uninformed investors will be represented by the passively managed ETFs, whereas the actively managed mutual funds correspond with investors who have obtained private information.

Emerging markets

Using the definition as described by Eling and Faust (2009) emerging markets are those countries that are in the process of rapid growth and industrialization. De Santis and Imrohoroglu (as cited in Eling & Faust, 2009) concluded that emerging markets have substantial growth opportunities but are also more volatile than developed markets because of the existence of political and economic risks.

According to Harvey (1995, as cited in Rouwenhorst, 1999) most emerging markets have low correlations with other stock markets when compared to developed markets. Low correlations imply possible diversification benefits (Markowitz, 1952 as cited in Wüsten & Stephen, 2012). Huij and Post (2011) concluded that emerging markets demonstrated better performance than US funds. This would be consistent with the view that the efficient market hypothesis does not hold in emerging hence, less efficient, markets. Since we use the assumption that emerging markets are less efficient than developed markets, one would assume that besides diversification opportunities, opportunities to obtain risk-adjusted abnormal returns would present themselves as well.

Analyzing active management and the Capital Asset Pricing model

To evaluate whether investments can generate higher risk-adjusted returns in emerging markets than in developed markets we will analyze the Capital Asset Pricing Model (CAPM). Markowitz (1952, as cited in Wüsten & Stephen, 2012) found most individuals solely consider expected return and expected risk when analyzing an asset. Since both risk and return are passively correlated a tradeoff has to take place. The CAPM is a useful tool for estimating expected returns of stocks as well as entire portfolios. CAPM states that the return of any stock can be separated into systematic risk and idiosyncratic risk, which is firm-specific. The idea of the CAPM is that investors are only compensated for systematic risk, since they can diversify all

7 firm-specific risk away (Grinold & Kahn, 1999). The sensitivity of a stock’s or portfolio’s risk to the market is measured by beta, 𝛽.

Beta is the covariance of the stock or portfolio with the market, divided by the variance of the market (Boďa & Kanderová, 2014).

𝐶𝑜𝑣(𝑟𝑠, 𝑟𝑚)

𝑉𝑎𝑟(𝑟𝑚)

Where

𝐶𝑜𝑣(𝑟𝑠, 𝑟𝑚) = 𝑐𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑎 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑤𝑖𝑡ℎ 𝑡ℎ𝑒 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛

𝑉𝑎𝑟(𝑟𝑚) = 𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛

Beta thus measures a stock’s or portfolio’s sensitivity to market risk and is easily derived with the use of a linear regression of a stock’s of portfolio’s excess returns (𝑟𝑠− 𝑟𝑓) on market excess

returns (𝑟𝑚− 𝑟𝑓). The intercept of the line, the alpha captures the abnormal performance of a

security. 𝑟𝑠− 𝑟𝑓 = 𝑎𝑠+ 𝛽𝑠(𝑟𝑚− 𝑟𝑓) + 𝜀𝑠 In which 𝑟𝑠 = 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑟𝑓 = 𝑟𝑖𝑠𝑘 − 𝑓𝑟𝑒𝑒 𝑟𝑎𝑡𝑒 𝑎𝑠 = 𝑎𝑙𝑝ℎ𝑎 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝛽𝑠 = 𝑏𝑒𝑡𝑎 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑟𝑚 = 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛 𝜀𝑠 = 𝑒𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚 𝑜𝑓 𝑡ℎ𝑒 𝑟𝑒𝑔𝑟𝑒𝑠𝑠𝑖𝑜𝑛

This equation states that the return of a stock or portfolio, adjusted for the risk-free rate, should equal the alpha of the stock plus the stock’s beta times the market return adjusted for the risk free rate. The error term of the regression is represented by 𝜀𝑠 and should have an expectation

8 of 0. When we take expectations on both sides the expected residual return should be zero (Grinold & Kahn).

The CAPM equation states

𝐸(𝑟𝑠) − 𝑟𝑓 = 𝛽𝑠(𝐸(𝑟𝑚) − 𝑟𝑓) Where 𝐸(𝑟𝑠) = 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑟𝑓 = 𝑟𝑖𝑠𝑘 − 𝑓𝑟𝑒𝑒 𝑟𝑎𝑡𝑒 𝛽𝑠 = 𝑏𝑒𝑡𝑎 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑜𝑟 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝐸(𝑟𝑚) − 𝑟𝑓= 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑚𝑎𝑟𝑘𝑒𝑡 𝑎𝑑𝑗𝑢𝑠𝑡𝑒𝑑 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑟𝑖𝑠𝑘 − 𝑓𝑟𝑒𝑒 𝑟𝑎𝑡𝑒

This formula states that the expected return of a stock or portfolio is determined by the risk-free rate, the expected excess return of the market, and the beta of the stock or portfolio. The fact that according to the CAPM there is no residual excess return explains why investors should hold the market portfolio, when investors all hold the same expectations and markets are efficient. The CAPM is a useful tool for determining the expected return of a security or portfolio, though, one should bear in mind that the CAPM only uses systematic risk for the determination of the expected return of an asset; it thus assumes there is only one risk to account for.

As described by Lin (2004) a single-market factor, like the CAPM, is not enough to capture the variations in excess returns for emerging market funds. Seeing that we just confirmed the assumption of the failure of the efficient market hypothesis in emerging markets this ‘’fact’’, concerning the impossibility of abnormal returns, might not hold. The CAPM equation is used to determine the required rate of return.

Alpha is used to determine how much the realized return varies from the expected return calculated with CAPM. Alpha can therefore be seen as a measure of performance that compares the realized return with the return that should have been earned given the amount of systematic risk, the beta. Alpha is the intercept of the CAPM line. The success of an active portfolio strategy relies on the existence of alpha. Alpha is described as the difference between a

9 fund’s expected risk adjusted return and the actual realized return. While using the market index as a benchmark there will be looked for excess returns of a fund compared to the benchmark and this will provide a fund’s alpha. If the active portfolios generate positive alphas this will be an indication of the effectiveness of active portfolio management. Alpha is the residual or, intercept, from the model. A positive alpha figure indicates the fund has performed better than its beta would predict. When calculating alpha we do assume that market risk, as measured by beta, is the only relevant risk factor.

Mutual fund past performance

In this section past performance of mutual funds will be analyzed. To achieve above normal returns has generated a great deal of interest among investors and especially the ability for active portfolio managers to accomplish this objective. A great variety of research exists on the persistence of performance of mutual funds. The predominant outcome is the, on average, underperformance of mutual funds compared to passively managed index funds (Bollen and Busse, 2005; Carhart, 1997; Elton et al., 1996; Hendricks et al., 1993, as cited in Blitz and Huij, 2012).

A study has shown that on average managed funds underperform their benchmarks by 1% (Kosowski et al. 2006, Barras et al. 2010, Cuthbertson et al. 2008, as cited in Cuthbertson, Nitzche, and O’Sullivan, 2010). According to Cuthbertson, Nitzche, and O’Sullivan (2010) the key drivers for the bad performance of the mutual funds are fees, expenses and high turnover. Portfolio turnover is a measure of how frequently assets within a fund are bought and sold by managers. Higher turnover ratios imply more frequent buying and selling and thus higher transaction costs.

A counter argument concerning turnover is made by Dey. Dey (2005) investigated the determinants of portfolio turnover and the relation between expected returns on stock exchange indexes and their turnover. Dey (2005) described the existence of a positive relation between turnover and growth. According to Dey (2005) this relation suggests that choosing a higher fraction of growth stocks in a portfolio might increase turnover, but, since growth stocks are riskier than value stocks, the expected return on such a portfolio will increase, thus only reflecting higher returns due to higher risk.

Grinblatt and Titman (1989) argue that if managers do have superior investment strategies they will collect additional profits through higher expenses, which will result in depletion of the abnormal returns obtained for the investor. Furthermore, they do indicate superior fund performance may exist among aggressive growth and growth funds and funds with small net asset values (Grinblatt and Titman, 1989).

10 One way to measure if a fund outperforms the market is by looking at a fund’s alpha. As described above alpha is the difference between a fund’s expected risk adjusted return and the actual realized return. If the active portfolios generate positive alphas this will be an indication of the effectiveness of active portfolio management.

Gottesman and Morey (2007) investigated the predictability in emerging market fund performance (as cited in Huij & Post, 2011). Gottesman and Morey regressed expense ratio, turnover, fund size, manager tenure and past performance to study whether one or more of these variables had a significant effect on fund performance. Expense ratio seemed to be the only noteworthy variable which affected fund performance. Expense ratio and fund performance exhibit an inverse relationship, so funds with lower expenses showed better performance on average.

One more important concern to point out when looking at mutual fund performance is the possible existence of survivorship bias. Most mutual funds disappear if they consistently underperform for a period of time. When funds are analyzed, and the funds which are already terminated are not included, one would expect the results of mutual funds to overestimate the past returns. Malkiel (1995) argued that most data sets that are used for the assessment of mutual fund performance exclude funds that have terminated operations. When these funds are excluded mutual fund performance can be seen as having superior risk-adjusted performance. However, when all funds are included and evaluated mutual funds do not seem to outperform the market (Malkiel, 1995). Therefore it is important to mitigate the effect of survivorship bias and include funds which already terminated as well.

Concluding, most previous studies either find insignificant abnormal returns for actively managed funds or when abnormal returns do exist they are diminished as a result of management fees and other expenses.

Mutual fund performance measures and persistence

Investors are not only interested in the past performance of mutual funds, but they are concerned about the persistence of this performance as well. As discussed above, multiple factors might affect performance of mutual funds and the persistence of this performance. In their pursuit of opportunities to obtain risk-adjusted abnormal returns mutual funds tend to have higher stock turnover, they thus engage more frequently in buying and selling of stocks. According to research high stock turnover has a negative effect on mutual fund returns (Carhart, 1997). Carhart (1997) states that expense ratios and load fees have a negative impact on mutual fund performance as well. Carhart simultaneously illustrated that mutual funds with high alphas demonstrate above average alphas and expected returns in subsequent periods. Unfortunately,

11 these results are not robust to model misspecification because the same model is used to estimate both periods.

To investigate the persistence of mutual fund performance Malkiel (1995) divided a sample of mutual funds in top and bottom half performers. Malkiel analyzed these funds over two periods to determine whether they were able to maintain their position. He found that between 1970 and 1980 the percentage of repeated winners was 65.1%. Over the period 1980-1990, 51.7% of last year’s winners were next year’s winner. This would indicate that persistence in mutual fund performance is possible.

Elton, Gruber, and Blake (1996) concluded that when future performance is evaluated over 3-year periods, selection on the former 3-year alpha probably contains more information about future performance than any other time horizon, which provides us with a good implication of a proper time horizon for our data analysis.

ETFs: characteristics and performance

The introduction of Exchange Traded Funds (ETFs) took place in 1993 and they have become popular investment vehicles since then (Anderson, Born, and Schnusenberg, 2010). ETFs are securities that track different kinds of indexes, this characteristic appeals to investors who want to invest in specific sectors or country markets (Maduro & Ngo, 2008). Unlike mutual funds ETFs trade on stock exchanges. ETFs are well diversified and are therefore usually lower-risk investments than individual stocks (Hehn, 2005). In addition they are characterized by low costs and high liquidity (Maduro & Ngo, 2008). Since they are passive investment products their turnover is usually much lower than the turnover of mutual funds, which lowers their costs (Anderson, Born, and Schnusenberg, 2010). Because of this, ETFs are normally characterized by lower expense ratios which makes them more attractive for investors. Also, unlike mutual funds, ETFs are priced continuously as they can be purchased at existing market prices throughout the day, while mutual funds are only priced once a day. This continuous pricing makes it is attractive for investors who seek to track benchmark indices or investors who want to gain exposure to a certain market segment (Anderson, Born, and Schnusenberg, 2010).

Investors do incur a brokerage commission when they either buy or sell an ETF, which is an additional expense when compared to mutual funds. As for international ETFs Khorana et al. (1998) suggest that ETFs may be capable of providing a more effective, low-cost strategy of diversifying internationally than closed-end mutual funds (as cited in Anderson, Born, and Schnusenberg, 2010).

Nonetheless, Jares and Lavin (2004) find that since trading hours between the US and the Japanese and Hong Kong markets are not overlapping deviations in the market arise for

12 those markets and the value of the underlying securities (as cited in Anderson, Born, and Schnusenberg, 2010). Moreover, they state that these deviations are positively related to the consecutive ETF returns and, consequently, may create profit opportunities. Harper et al. (2005) compare the risk and return performance of ETFs available for foreign markets and closed-end country funds. They arrived at the conclusion that ETFs exhibit higher mean returns than foreign closed-end funds, furthermore they find that closed-end funds exhibit negative alphas. This suggests that a passive investment strategy with the use of ETFs may be superior to an active investment strategy using closed-end funds (as cited in Anderson, Born, and Schnusenberg, 2010).

According to Blitz and Huij (2012) there are multiple studies that have investigated the performance of ETFs that track equity indices in the US, all arrived at the same conclusion which is that ETF performance is predictable to a high degree of accuracy. ETFs often manage to stay near their benchmark indices. Additionally, these studies found that ETF expenses and returns seem to have a negative one-on-one relationship2

_.

Blitz and Huij (2012) examined the performance of global emerging market (GEM) ETFs. They found that GEM ETFs exhibit higher levels of tracking error than developed market ETFs and that, on average, GEM ETFs underperform their benchmark indices by 85 basis points (Blitz and Huij, 2012). They also questioned whether half of the funds they examined even could be classified as passive funds due to the substantially high levels of tracking errors. Although trading costs are higher in emerging markets than in developed markets, emerging market ETFs do not seem to underperform developed market passive funds. One of the explanations for the higher trading costs, yet not underperforming emerging market ETFs, might be that emerging market ETFs achieve higher revenues.

13

Hypotheses

On the grounds of the inapplicability of the efficient market hypothesis the assumption of imperfectly priced assets is adequate. When opportunities to obtain risk-adjusted abnormal returns exist, value creation through active portfolio management is highly attainable. Therefore the hypothesis states that active portfolio management, thus mutual fund performance, is superior to passively tracked market indexes, ETFs, in emerging markets.

H0 : Mutual funds outperform ETFs in emerging markets when compared to a benchmark index.

In agreement with the above mentioned literature adjustment for expenses need to be made for a proper analysis. Hence, the altered hypothesis states

H0 : When adjusted for expenses mutual funds outperform ETFs in emerging markets when compared to a benchmark index.

14

General approach and data

To investigate whether active management is associated with higher returns as compared to passively managed funds data on mutual funds as well as ETFs, both in emerging markets, is needed. Evaluating this possible relationship will be done by the use of multiple Ordinary Least Squares (OLS) regressions. Elton, Gruber, and Blake provided us with a proper time horizon for the data, namely three years. The past three years, 2014-2016, will be used.

The MSCI Emerging Market Free index is an index representing the emerging markets which can be used as a benchmark. Funds which were classified by Morningstar as emerging market funds were acquired using DataStream. According to Morningstar’s definition of emerging markets funds, these funds predominantly invest in emerging market equities. The same was done for ETFs. All returns are measured in US dollars.

A remarkable observation concerning the return data of the mutual fund returns as well as the ETF returns is that they were mostly negative throughout the three-year period. As seen in the table, retrieved from The Wall Street Journal(5)_{, returns in emerging markets have been}

falling since, approximately, 2014.

For the MSCI Emerging Market Index a same pattern occurs, the past threeyear net return was -4.95%3_{. This might be due to falling commodities prices, especially oil during the period}4_.

Besides, China faced a stock market crises since Augustus 2015, in that year the annual performance of the MSCI EM index was -14.92% while China had the largest individual country weight of 23.64%. According to The Wall Street Journal growth in emerging markets relies

3 _{https://www.msci.com/resources/factsheets/index_fact_sheet/msci-emerging-markets-index-usd-net.pdf} 4 _{http://www.mraassociates.com/2015-year-in-review-market-commentary/}

15 intensely on China’s demand for metals, energy, and food5_{. Another possible explanation of the}

losses is the weakening of the foreign currencies relative to the US dollar. All things considered we should bear in mind that a more stable period of financial markets would give a better representation of the differences between mutual fund and ETF performance.

For both mutual funds and ETFs at least one year of data, ergo 12 data points, was required to be included in the sample. The complete sample consists of 63 mutual funds, with the main focus on all equity funds, and 41 ETFs, all classified as emerging markets.

To investigate whether stock markets give abnormal returns there will be look at the alpha, which is considered as a measure of the active return on investment. As described earlier the CAPM formula is

𝑟𝑠− 𝑟𝑓 = 𝑎𝑠+ 𝛽𝑠(𝑟𝑚− 𝑟𝑓) + 𝜀𝑠

As explained previously the betas are measures of a stock's or portfolio’s return sensitivity when compared to the market index, in this case the MSCI. The MSCI is used to measure equity performance in emerging markets. The MSCI Emerging Markets Index is a free-float weighted equity index that captures large and mid-cap representation across 23 Emerging Markets countries6_{. For ETFs a logical assumption would be a beta of 1 as ETFs track benchmark indices}

and, consequently, their sensitivity should be equal to the market, whereas for mutual funds this relationship probably does not hold. Although ETF betas should have a value around 1, it is still important to calculate their betas as well.

Since some anomalies arise with the use of the CAPM, and especially in emerging markets, additional variables were extracted; turnover ratios, expense ratios, and management fees in the case of mutual funds. Opposite to mutual funds, ETFs are passively managed funds for which no adjustments need to be made for manager fees and 12b-1 fees. A 12b-1 fee is considered an operational expense and it is thus included in a fund’s expense ratio. As described earlier on, there is conflicting literature on the effect of turnover on mutual fund performance. Cuthbertson, Nitzche, and O’Sullivan (2010) stated that high turnover is a key driver to bad performance since higher turnover leads to higher transaction costs. This view is supported by Carhart (1997) who observed a negative relationship between turnover and fund performance as well. Dey (2005) however, pointed out the existence of a positive relation between turnover

5_{http://www.wsj.com/articles/emerging-market-fund-investors-too-slow-to-catch-peak-returns-1448273392}

16 and growth. Turnover ratios will therefore be included to investigate whether it has an either negative or positive effect on returns. Concerning expense ratios, Carhart (1997) states that these have a negative impact on mutual fund performance. This view is confirmed by Gottesman and Morey (2007), who found that expense ratios seemed to be the only noteworthy variable which affected fund performance. Gottesman and Morey found the existence of an inverse relationship between expense ratios and performance. Additionally Blitz and Huij (2012) explored multiple studies on performance of ETFs, presenting a negative one-on-one relationship between expenses and returns. For the risk-free rate in the CAPM formula the Treasury bill rate is used7_{, this because emerging market government bonds cannot be}

considered as riskless investments in general.

When funds had multiple share types invested some were deleted from the sample to prevent that these funds were overrepresented. Justification for this elimination is provided for the reason that additional share types do not carry new information, they only differ in terms of fee structure. The regression will not suffer either as the fees are included in the expense ratios and net returns will be used.

17

Regressions

The aim of this thesis is to investigate whether mutual funds outperform passively managed ETFs throughout a period of three years. This will be done based on regression analysis, specifically, the widely used Ordinary Least Squared (OLS) regression. All regressions are done with the statistical software package, Stata.

Whether active management in emerging markets has a causal impact on returns shall be examined by regressing the dependent variable, Y (returns), on multiple relevant factors. First, all CAPM betas, and subsequently the alphas, will be calculated. The betas are calculated by regressing the market’s excess return against the stock’s excess returns. The alphas are represented by the intercepts of the regression, and should be zero. In the case of imperfect capital markets we do, however, assume a potentially positive intercept of the regression for the case of mutual funds. Positive alphas would mean that, adjusted for exposure to market risk, a mutual fund has outperformed the market portfolio.

This regression solely probably suffers from omitted variable bias, which occurs when an omitted variable correlates with both the dependent and the independent variable. Especially since we do not assume perfect capital markets and as Lin (2004) stated earlier: a single-factor model is not sufficient to capture the variations in excess returns in emerging markets. To control for omitted variable bias additional variables, on the basis of our literature review, are introduced, namely; turnover ratios, expense ratios, and the three year betas which were obtained above. Funds of which turnover percentage or expense ratio were not available were still included.

The final regression formula is

𝑦𝑖 = 𝛼0,𝑖 + 𝛽1𝑋1,𝑖 + 𝛽2𝑋2,𝑖 + 𝛽3𝑋3,𝑖 + 𝜀𝑠 In which 𝑦𝑖 𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑠 𝑡ℎ𝑒 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑡 ℎ𝑒 𝑚𝑢𝑡𝑢𝑎𝑙 𝑓𝑢𝑛𝑑 𝑜𝑟 𝐸𝑇𝐹 𝑋1,𝑖 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑠 𝑡ℎ𝑒 𝐶𝐴𝑃𝑀 𝛽 𝑋2,𝑖 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑠 𝑡ℎ𝑒 𝑒𝑥𝑝𝑒𝑛𝑠𝑒 𝑟𝑎𝑡𝑖𝑜 𝑋3,𝑖 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑠 𝑡ℎ𝑒 𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 𝑟𝑎𝑡𝑖𝑜

18 In this paper special emphasis is given to turnover and expense ratios to investigate whether they have a significant effect on returns. Turnover ratio can be used as a measure for portfolio liquidity. The turnover ratio also reflects the degree of active management since active funds trade more regularly. Expense ratios are expressed as a percentage of a fund’s net assets.

Heteroscedasticity

With regression analysis one can encounter a few problems, among one of them is the possibility of heteroscedasticity. Heteroscedasticity might arise with the use of regression analysis, which means that the variance of the error term is not constant. When the errors are heteroscedastic they are biased which leads to incorrect conclusions about the significance of the regression coefficients. A test for heteroscedasticity is available in Stata, the Breusch-Pagan / Cook-Weisberg test, which displayed a Chi2 statistic of 4.77, with a p-value of 0.0289 for the mutual fund data, as shown in table 4 of the appendix. These results indicate that the variance is not constant and thus the problem of heteroscedasticity arises. To prevent too much elaboration on statistical matters, for heteroscedasticity is controlled with the option of robust standard errors. For ETFs no robust standard errors were needed since the Chi2 statistic was 0.01 with a p-value of 0.9422 as shown in table 10 of the appendix.

19

Results

The main goal of this thesis is to determine whether active portfolio management in emerging markets adds value. To begin with, the main results of all variables individually will be discussed together with their implications and possible explanations, thereafter a comparison between the performance of mutual funds and ETFs will be made on the basis of another regression analysis. All regressions and other statistical tests can be found in the appendix.

For mutual funds the average year beta of all funds was 1.18, for ETFs the average 3-year beta was 1.09. These results indicate that mutual funds, on average, moved with 118% for every 100% move of the market. ETFs were slightly less affected by market movements historically, with a 109% change for every 100% move of the market, which seems like a logical result as ETFs track market indices.

The 3-year average alpha of the mutual funds was approximately -0.25 whereas the average three year alpha for ETFs was -6.51. Summary statistics on mutual funds and ETFs can be found in table 1 and 7, respectively, of the appendix. Hence, both mutual funds and ETFs underperformed the market during the last three years. An alpha of -6.51 is quite a striking results since ETFs carry less costs on average due to lower turnover and expense ratios and no management fees. The alpha for index ETFs should approximately match the benchmark’s beta. So based upon the ETFs level of risk they underperformed significantly. As discussed earlier emerging markets are facing a downturn.

According to an article published by Reuters on the 2nd _{of March this year emerging}

market equities underperformed developed market equities by 50% in the last five years8_{. This}

could be one of the implications for the negative returns of both mutual funds and ETFs, which were on average -4.31% and -4.96% respectively, as shown in tables 1 and 7 of the appendix.

In general expense ratios of mutual funds are much higher than for ETFs, specifically in the data of this thesis expense ratios were 0.58% on average for ETFs, compared to 1.28% for mutual funds, which seems like a logical result since higher expense ratios for mutual funds are expected. For neither mutual funds nor for ETFs the expense ratio displayed a significant result. Results of the regressions of the mutual funds can be found in table 5 of the appendix, in which an insignificant p-value of 0.191 is shown for the expense ratio coefficient. In table 11 of the appendix an even more insignificant result concerning the effect of expense ratios on ETF returns is shown, namely 0.690. On the report of the Investment Company Institute9 _{is shown}

that expense ratios for mutual funds have fallen considerably. This might be due to increased

8_{http://www.reuters.com/article/emerging-flows-idUSKCN0W40DW} 9_{http://www.icifactbook.org/fb_ch5.html}

20 competition and economies of scale, or due to the fact that expense ratios and a funds’ assets are inversely related which would simply indicate that funds’ assets have risen.

The turnover ratio for ETFs was 38.79% on average, whereas the turnover ratio for mutual funds was 51.91%, as shown in table 1 and 7 of the appendix. A puzzling outcome of the regression analysis is that turnover ratios have a significant effect, at the 1% level, on ETF returns, as well as on mutual fund returns. This in contrast to the apparently insignificant effect of expense ratios. One would might expect the difference to be much larger, though, in the figure seen below is appears that turnover ratios have experienced noticeable drops in the last 6 years.

According to Morningstar turnover ratios above 100% reflect high buying and selling activity, which is one of the indications for active portfolio management. Since the mutual funds represented the active side of investing one could wonder whether this sample of mutual funds is indeed a correct representation of active portfolio management considering the average mutual fund turnover ratio was ‘’only’’ 51.91%.

Merging of mutual fund and ETF data

To determine whether there are noticeable differences between mutual fund and ETF performance, and the determinants of this performance, the two data sets are merged. To compare both samples a file in which the alphas, betas, expense ratios, turnover ratios and returns for both the mutual funds as well as the ETFs are present is created. Mean differences of

21 all variables are calculated using paired t tests, as shown in the appendix, table 13-17. The mean difference between the average three year returns were not indicated as significant by the test statistic, as shown in table 13. Differences in alphas, as shown in table 14 of the appendix, were significant with a p-value lower than 0.00001 which indicates the underperformance of ETFs given their betas. In comparison with the alphas, the betas did not show any significant difference between mutual funds and ETFs. The t-statistic on the difference in betas between mutual funds and ETFs revealed a value of 0.52, which corresponds with a p-value of approximately 0.3029. Furthermore, the difference in turnover ratios is shown in table 16 and was not significant either, this might be attributed to the fact that turnover ratios of mutual funds have lowered over, approximately, the last decade. Another possibility might be that this specific sample included mutual funds with relatively lower turnover ratios. Expense ratios did show a significant test result, which was expected on the basis of our literature review. Table 17 in the appendix displays a t-statistic of 8.5780 which is significant at the 1% level.

Mostly insignificant results

In the interest of further research it is important to know what the main drivers are for performance and to what insignificant outcomes in this thesis can be attributed. Therefore a few possibilities will be discussed.

Sample

Though the sample should be statistically significant, it might be more significant if more mutual funds and ETFs were included. Smaller samples are also relatively more sensitive to outliers. Outliers are considered as data points which are located far outside the norm of a variable. Outliers therefore increase the variance of the error and they reduce the power of statistical tests. Another interesting distinction in data can be made on the basis of location characteristics. One could look at different regions, for example Southeast Asia and South America, and analyze if there are different returns per location.

Finally, since the sample represented the last three years in which the emerging market industry was mainly characterized by negative returns, probably a more stable financial period would lead to a more controlled comparison. Now the sample probably suffers from selection bias, which occurs when proper randomization is not accomplished, which makes the sample unrepresentative. One possible additional factor that relates to this problem is home country bias. Home country bias arises by cause of a preference which investors have towards their home country. Investors who suffer from home country bias tend to hold more optimistic

22 expectations about their domestic markets when compared to foreign markets. Therefore investors might continue to hold pessimistic expectations towards emerging markets.

Omitted variables (and possible further research)

Considering that the results were quite unusual, and they might even be considered odd, there might be factors that affect returns missing. To examine whether important variables are missing the Ramsey test for omitted variables is executed for both mutual fund and ETF data. In the case of mutual fund data the test statistic displayed a F-statistic of 7.30 with a p-value of 0.0004 which indicates that important data has been left out in the regression analysis, as shown in table 3. The same applies to the ETF data, only slightly less significant with an F-statistic of 3.45 and a p-value 0.0296, which can be seen in table 9 of the appendix. Thus for both data samples there seem to be relevant variables missing, therefore, with a view on possible further research a few potential variables will be discussed.

As stated by Banz (1981, as cited in Rouwenhorst, 1999) small stocks earn higher average returns than large stock over long time periods. In this thesis no distinction is made concerning the size of the stock, for this reason one could investigate the effect of different stock sizes on returns in emerging markets.

Otten and Bams (2004) found that a four-factor model including market betas, portfolio size, book-to-market, and PR1YR (one year momentum in stock returns) is best able to explain mutual fund returns.

Other research of Fama and French (1992, 1996, as cited in Rouwenhorst, 1999) found that value stocks with high book-to-market (HML), price-to-earning, or cash flow to price values outperform growth stocks of which these factors are lower. Moreover, it seems that high returns stocks for a period between three months and one year outperform stocks with poor prior performance.

Additionally, one could make a distinction between different types of industries, for example; healthcare, technology, industrials etc. Multiple suggestions exist for further research on differences between mutual fund performance and ETF performance.

Previous research, as described above, was mainly conducted on developed markets, therefore it is still interesting to examine whether these seemingly relevant variables affect performance in emerging markets as well.

Another important variable which might influence the returns and attractiveness of both mutual funds and ETFs are taxes. Mutual funds are subject to capital gain taxes, which they must pass through to their shareholders. Joel Dickson and Shoven (1995) and Dickson et al. (2000) point out that this tax burden makes investing in mutual funds relatively less attractive

23 when compared to the tax burden on investors who simply buy and hold a portfolio of securities (cited in Poterba and Shoven, 2012). Poterba and Shoven (2012) compared the pre-tax and after-tax returns of ETFs with the largest equity index fund. They discovered that both the pre-tax and the after-pre-tax returns of the fund were somewhat higher than the ETF. For further research the effect of taxes can be studied as well. To conclude with a final suggestion, since returns are correlated with higher P/E ratios in developed markets, this might be an interesting variable to add to the regression to examine whether the same holds true for emerging markets.

Added variable plot/partial regression plot

When a linear regression with only one independent variable is done the results can easily be inserted into a scatter plot. Unfortunately, with multiple variables this is impossible since one would need a multidimensional plot. A potential solution is to generate scatter plots of each variable individually against the independent variable, the returns. This creates a two dimensional plot in which there is a one-to-one correspondence between one of the dependent variables and the independent variable. The dependent variable has the same coefficient and standard error as in the original regression. In this way there can be looked at the relationship between a dependent variable and the independent variable once all other predictors are accounted for. Values of the variables should be interpreted as the effect of a change in the relevant variable on the return with all other dependent variables given. As seen in the appendix, tables 6 and 12, added variable plots for all data variables have been executed. In the case of mutual funds the data for each of the variables contains some outliers. Outliers do contribute to the slope of the regression which possibly results in a bias. To obtain a better estimate of the slope of each variable, these outliers should be investigated to be able to conclude whether they are due to variability in the measurement or it might be due to an experimental error.

24

Conclusion

Considering the assumption that emerging markets are less efficient throughout this thesis, one would assume more opportunities to obtain risk-adjusted abnormal returns would have presented themselves in emerging markets. In the last three years however, emerging markets seemed to have not only underperformed their benchmark index, but they have underperformed the developed markets as well. The findings do confirm a large amount of past research on mutual fund performance, which predominantly indicated the, on average, underperformance of mutual funds. Whereas for ETFs, their performance in emerging markets is in agreement with the research op Blitz and Huij (2012) who found that emerging markets ETFs underperform their benchmark index. The MSCI Emerging Fund Index returns for the last three years were -4.95% (annualized) itself. According to Morningstar, on the 23rd _{of December}

2015, emerging markets have been underperforming developed markets by approximately 14% for the last five years. As mentioned earlier the underperformance of emerging markets might be due to a stronger dollar and falling commodity prices. Additionally, this underperformance might consists in the near future due to the possibility of home country bias.

Many literature exists on relevant variables concerning both mutual fund and ETF performance. The decision on which variables to include in the regression analysis was made on the basis of the literature review, as well as the availability of data. CAPM betas were included, together with expense ratios and turnover ratios for both mutual funds and ETFs.

The results of the regression analysis were mostly inconclusive. They do suggest that there is a negative relationship between a fund’s returns and its expense and turnover ratio, however it did not appear to be statistically significant in the case of mutual funds. However, for ETFs the turnover did seem to have a significant negative effect on the returns.

Concerning the hypothesis, there does not seem to be a significant difference in performance between the risk-adjusted returns of active management and passive management. It is important, though, to point out that ETFs serve as a representative for passive management, therefore one must be careful generalizing the results to broader market indices.

It is worth noting that since emerging markets are, almost consistently, experiencing negative returns for the past three years, results might be meaningless. This must be taken into account when deriving conclusions from the sample. The sample might not be an appropriate representative and a more stable period of financial markets might give a better description.

25

Literature

Anderson, S., Born, Jeffery A, & Schnusenberg, Oliver. (2010). Closed-End Funds,

Exchange-Traded Funds, and Hedge Funds Origins, Functions, and Literature / (Innovations in

Financial Markets and Institutions; 18). Boston, MA: Springer-Verlag US.

Alexakis, C., Patra, T., & Poshakwale, S. (2010). Predictability of stock returns using financial statement information: Evidence on semi-strong efficiency of emerging Greek stock market. Applied Financial Economics, 20(16), 1321-1326.

Balaban, E., & Kunter, K. (1997). A note on the efficiency of financial markets in a developing country. Applied Economics Letters, 4(2), 109-112.

Blitz, & Huij. (2012). Evaluating the performance of global emerging markets equity exchange-traded funds. Emerging Markets Review, 13(2), 149-158.

Boďa, & Kanderová. (2014). Linearity of the Sharpe-Lintner Version of the Capital Asset Pricing Model. Procedia - Social and Behavioral Sciences, 110, 1136-1147.

Dey, M. (2005). Turnover and return in global stock markets. Emerging Markets Review, 6(1), 45-67.

Carhart, M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance, 52(1), 57-82.

Cooray, A., & Wickremasinghe, G. (2008). The Efficiency of Emerging Stock Markets: Empirical Evidence from the South Asian Region. The Journal of Developing Areas, 41(1), 171-183.

Cuthbertson, K., Nitzsche, D., & O'Sullivan, N. (2010). Mutual Fund Performance: Measurement and Evidence 1. Financial Markets, Institutions & Instruments, 19(2), 95-187

Grinold, Richard C., & Kahn, Ronald N. (1999). A Quantitative Approach for Providing Superior

Returns and Controlling for Risk. McGraw-Hill. Retrieved from

26 Grossman, Sanford J., & Stiglitz, Joseph E. (1980). On the impossibility of informationally

efficient markets. American Economic Review, 70, 393.

Elton, E., Gruber, M., & Blake, C. (1996). The Persistence of Risk-Adjusted Mutual Fund Performance. Journal of Business, 69(2), 133-157

Eling, & Faust. (2010). The performance of hedge funds and mutual funds in emerging markets.

Journal of Banking and Finance, 34(8), 1993-2009.

Grinblatt, M., & Titman, S. (1989). Mutual Fund Performance: An Analysis of Quarterly Portfolio Holdings. Journal of Business, 62(3), 393-416.

Hehn, Elisabeth. (2005). Exchange Traded Funds. Berlin, Heidelberg: Springer Berlin Heidelberg.

Huij, & Post. (2011). On the performance of emerging market equity mutual funds. Emerging Markets Review, 12(3), 238-249.

Lo, Andrew W. (2004). The adaptive markets hypothesis: Market efficiency from an evolutionary perspective. Journal of Portfolio Management, 15.

Lin, W. (2013). Is there alpha in institutional emerging-market equity funds? Journal of Portfolio

Management, 39(4), 106-117

Madura, J., & Ngo, T. (2008). Pricing behavior of exchange traded funds. Journal of Economics

and Finance, 32(1), 1-23.

Malkiel, B. (1995). Returns from Investing in Equity Mutual Funds 1971 to 1991. Journal of Finance, 50(2), 549-572.

Otten, R., & Bams, D. (2004). How to measure mutual fund performance: Economic versus statistical relevance. Accounting & Finance, 44(2), 203-222.

Poterba, J., & Shoven, J. (2002). Exchange-Traded Funds: A New Investment Option for Taxable Investors. American Economic Review, 92(2), 422-427.

27 Rouwenhorst, K. (1999). Local Return Factors and Turnover in Emerging Stock Markets. Journal

of Finance, 54(4), 1439-1464.

Wüsten, N., & Stephan, A. (2012). Active Portfolio Management in the German Stock Market: A CAPM Approach.

28

Appendix

Mutual funds

Table 1. Mutual fund summary statistics

*Ratios and fees are displayed in percentages

29

Table 3. Ramsey test for omitted variables for mutual funds

Table 4. Breusch-Pagan / Cook-Weisberg for heteroscedasticity in mutual fund data

30

31

Exchange Traded Funds

*Ratios are displayed in percentages

Table 8. Correlations ETF variables

32

Table 10. Breusch-Pagan / Cook-Weisberg test for heteroscedasticity in ETF data

33

34

T-tests on mean differences between mutual funds and ETFs

Table 13. Mean differences in return between mutual funds and ETFs

35

Table 15. Mean differences in betas between mutual funds and ETFs

36