EVIDENCE OF MOMENTUM INVESTING STRATEGIES AMONG U.S. MUTUAL FUNDS

Author : Davide Pesce

Student number: s3517047

Supervisor: Dr. Wolfgang Bessler

Date: 10/01/2019

Master Thesis Finance

Rijksuniversiteit Groningen

ABASTRACT:

The primary purpose of my thesis is to investigate whether and to which degree professional investors, specifically active US-mutual fund, engage in momentum behavior. In order to answer my research question, I analyze a sample of 60 US mutual funds between December 2006 and June 2018, using two methodologies. I calculate the winners-over-losers and the winners-minus-losers ratios for every quarter for each fund portfolio. Then, I replicate the statistics proposed by Grinblatt et al. (1995). This measure calculates the degree to which portfolio managers tilt their portfolio towards the three months previous winners, and get rid of the losers compared to a benchmark portfolio. Overall, my results are in line with the findings of Grinblatt et al. (1995). Further, I show that the small-medium funds (total net assets between 107,525 and 371,850 million) have the highest momentum measure. Finally, after have adjusted the fund ’s performance for the common risk factors, the funds with the highest GWT measure exhibit also significant and positive momentum coefficient factor, providing support to my analysis.

I. Introduction

Momentum trading has been probably among the most discussed puzzle in Finance ( see Jegadeesh and Titman 1993; Grinblatt et al. 1995; Carhart 1997; Griffin et al. 2003; Fama and French, 2012; etc.). Further, momentum-investing strategies are also used and documented among portfolio managers. My research has the objective to answer the question whether the so-called "smart money" investors have superior skills or rather they follow simple trading rules. To answer that I need to investigate whether we can clearly observe momentum-patterns among portfolios of professional investors, of active US-mutual funds.

Momentum strategies benefit from the fact that the stocks that performed the best in the previous period are likely to continue the trend in the future. Jegadeesh and Titman (1993) are the first who exploit the momentum effect by buying stocks that have performed well in the past and selling the ones that perform badly. What renders the momentum effect puzzling is that the superior return could not be explained by systematic risk or reaction to common factors, but rather undermine the market efficiency hypothesis. Explanations of over/under reaction have been given since De Bondt, and Thaler (1985) and academia tends to use momentum-trading strategies in order to motivate models that exhibit irrationalities.

Ever since momentum is widely used among private and professional investors and extensively studied by academics. The momentum effect is still observed nowadays, as documented by Moskowitz et al. (2012), Novy-Marx (2012), and Chen et al. (2018), to mention some. It is also studied at the international level (see Rouwenhorst, 1998; Griffin et al., 2003; Galariotis et al., 2007; Chui et al., 2010; Chaves 2012; Fama and French, 2012).

Given the momentum anomaly, that is the unexplained returns compared to the level of risk, professional investors start to use momentum-investing strategies as documented by Grinblatt et al. (1995), where they found that US-mutual fund exhibited momentum behavior, and by Carhart (1997), who shows that momentum effect explains performance persistence among mutual funds. These findings are also confirmed by later studies such as Nofsinger and Sias (1999), Titman (2001), and Muley and Kim (2008). In contrast, Gomper and Metrick (2001) find no evidence of momentum trading among professional investors and de Haan and Kakes (2011) argue that Dutch institutional investors buy losers and sell winners; Thus they are mainly contrarian investors.

In order to investigate whether there is evidence of momentum-investing strategies, I analyze a sample of 60 US mutual funds using the statistics proposed by Grinblatt et al. (1995) (thereafter GWS), which given the efficacy for the purpose it has been used in other studies (see Badrinath and Wahal 2002; De Haan and Kakes 2011; Curcuru et al. 2011). Since we do not know the actual strategy pursue by professional managers, the best we can do is to analyze the fund-by-fund portfolio and look for evidence of momentum-pattern. Therefore, after I have constructed my dataset consisting of a portfolio of 47 holdings period for each fund, I apply two different methodologies.

First of all I detect momentum behavior through two ratios, which are the "winners-over-losers” ratio and the "winners-minus-losers" ratio. Winner stocks and loser stocks are defined in the same old fashion way as determined by Jegadeesh and Titman (1993). Thus, the winners are going to be the stocks that perform the best based on the three previous returns; in contrast, the losers are the ones that perform the worst. Even though these ratios have the same intuition behind them, and therefore came to the same conclusion (i.e., momentum-patterns among mutual funds), I use both of them in order to avoid some calculation constraints. The finding for the “winners-over-losers” (“winners-minus-losers”) ratio is 1.47 (0.007) indicating that on average the US-mutual funds exhibit evidence of momentum-investing strategy.

portfolio allocations between the last and the current period would have been made. Due to the GWT, I detect how much a fund manager exposes his portfolio to past winner stocks (respect to the benchmark period) rather than past losers. Therefore, a positive value means that the mutual funds on average have a portfolio with a higher return compared with the benchmark portfolio. I calculate the descriptive statistics for the GWT measure of the all sample, finding that on average US mutual funds have the momentum measure equal to 0.28%. Overall, the results confirm the findings discover by Grinblatt et al. (1995). Finally, I calculate the Sharpe ratio, suggesting that on average the funds in my sample underperform the market.

The paper I organize as follows. The next section presents a review of the previous literature regarding momentum, and the momentum-investment strategies among professional investors. I conclude with discussing the reason why managers should use momentum rules. In Section III, I describe my data set and the methodology I apply in order to detect momentum. The results are discussed in Section IV, where first I apply the ratio above-mentioned, and then I apply the GWT measure. In the same section, I also calculate the risk-adjusted performance and the correlation between the momentum measure and return. In section V, I undertake the robustness test. Finally, Section VI concludes and discusses the limits of my thesis.

II. Literature Review Market efficiency and momentum anomaly

The efficient market hypothesis (thereafter EMH), despite its ongoing debate in the academic and corporate world, especially in the active and passive asset management industry, is the theory underlying well-functioning financial markets. Formulated by Eugene Fama in 1965, the efficient market hypothesis in its strong from states that the price of any security fully reflects all available information (publicly and privately). This could mean that the price is the present value of the correctly forecasted future cash flows and discounted with the appropriate discount rate of the underlying security. The EMH assumes that the information is freely available to all market participants, which act rationally. It follows from this that if an investor has no privileged access to information (i.e. insider information) and does not have a superior model exploiting the publicly available information content, then there is a low probability to generate consistently abnormal returns. If an outperformance occurs, this is based on pure luck and not skills. When all current information is included in the stock price, stock prices changes can only occur when new information arrives, which then is immediately incorporated in the price.

To make the EMH testable, it is broken down in three different form. The strong form was explained above. Beyond the strong form of efficiency, other two forms of the EMH was developed by Fama (1970): the weak form of efficiency and the semi-strong form of efficiency. The first states that the best estimate of future prices is the current price. The latest indicates that all public information are reflected in the stock price. Usually, when talking about market efficiency, it refers to the semi-strong form of efficiency.

Since information (price) is not predictable under EMH, but arrives somewhat random, no investment pattern can be discerned; state differently, it appears that stock prices fluctuate randomly through time. This is why it is sometimes called “random walk.”

falling securities will keep falling. (This is the called the weak-form market efficiency, where prices move in upward and downward trends, and are therefore predictable and exploitable). This anomaly offers the opportunity for momentum strategies, where previous winners are bought, and previous losers are sold. Past and current studies explain such anomaly mostly with behavioral models, although they are based on rational theory and behavior. For instance, Barberis et al. (1998), and Daniel et al. (1998) explain price persistence assuming that investor is irrational, whereas Mazouz et al. (2012) find that underreaction leads to price persistence to specific assets. Underreaction is based on the semi-strong from of market efficiency which is tested with event studies.

In contrast, De Bondt and Thaler (1985) find that there is a tendency of human behavior to overreact to new information. Obviously, individual investor will always randomly overreact or underreact to new information otherwise there would be no trades in the stock market and no price adjustment process. Only a large trading volume makes the market efficient. However, the assumption of market efficiency only states that the average price contains all information as there is continuous price adjustments (see Grossman and Stiglitz, 1980). The extent to which creates as well as opportunities for abnormal returns, called “contrarian” strategy, which consists in taking the opposite direction of the current price trend.

Some researchers provide evidence that in practice the momentum anomaly generates abnormal returns, challenging even the weak form of EMH. However, when the returns are adjusted with a multi-factor asset pricing model such as the Carhart model (Performance model), including momentum, the performance usually vanishes, suggesting that the observed “Outperformance” is only the compensation for the systematic risk the investor assumes.

Momentum Theory

Jegadeesh and Titman (1993) were the first to identify the momentum effect. They provide evidence that simple trading strategies based on buying past winner and selling past loser (i.e. “selects stocks based on their past 6-months returns and holds them for 6 months”) deliver significant excess returns of 12% per year over the period from 1965 to 1989. De Bondt et al. (1985) report contradicting results to these findings. They found that past “loser” outperformed past winners, with the former having higher returns of about 25% than the latter, suggesting systematic price reversals. De Bondt et al. (1987) confirmed this return pattern also in a follow-up study. They find that investors’ overreaction might explain the negative relationship between excess return for losers (particularly in January) and both long-term and short-term formation period performance.

Further, Carhart (1997) used the findings of Jegadeesh and Titman (1993) to explain the short-term persistence among equity mutual fund returns, documented by Hendricks et al. (1993). He concludes that the one-year momentum (i.e. buying last years’ winner mutual funds and selling last year loser mutual funds) yield a return of 8% per year.

The momentum effect it not only documented for the US-market, but also there exists sufficient evidence from other countries around the world. For example, Rouwenhorst (1998) finds for 12 European countries that a diversified portfolio invested on past winner stocks outperforms the portfolio of past loser stocks by 1 percent per months. Griffin et al. (2003) finds that the momentum strategy yields positive returns for 40 countries. The co-movement between different countries is weak, suggesting that this that the risk would be very different between these countries (i.e. country-specific risk factor) if risk is the explanation for momentum.

for the risk factors in the Fama-French three-factor model, the number of significant profitable contrarian strategies disappears. However, the momentum strategies yielding significant returns increase by one.

A cross-country study by Chui et al. (2010) report that a momentum strategy exhibits profits and persistence over time and across different countries, generating even higher Sharpe ratios when compared to US studies. Furthermore, Fama and French (2012) also performed an international study on momentum returns. They examine the following four different regions: North America, Europe, Japan, and Asia-Pacific. They extend the insights from prior studies by investigating whether momentum returns relate to the size of the firm. They conclude that there are strong momentum returns almost in all countries with the exception of Japan. With respect to the relationship between the size of the firm and momentum returns, they find that these returns are lower for smaller firms and higher for larger firms.

Moskowitz et al. (2012) rather study cross-sectional momentum, which focuses on the relative returns of securities. They document time series momentum in equity indices, commodities, and bond futures, by concentrating in their study on the security’s past return. Their results show that time series momentum is consistent across several major asset classes for the last 25 years. They perform particularly well during extreme periods, i.e. market with large return movements, either up and down, which means periods with higher volatility.

In the large field of the momentum literature, there is also evidence that “intermediate” momentum strategies deliver even larger significant returns. Novy-Marx (2012) documents a strategy that purchases winner stocks based on their intermediate horizons’ past performance, for instance 12 to 7 months prior the portfolio formation. This strategy does not only deliver superior returns but also results in significant alphas relative to Fama-French four-factor model.

Chen et al. (2018) contribute to the existing literature by studying the variation of momentum performance at the individual firm level. Moreover, the empirical asset pricing literature investigated the existence of intermediate-term momentum effects and its persistence. However, this recent study has found that persistence performance of intermediate momentum is not as widely spread for all stocks that have an extreme past-performance. In this sample, which includes common stocks traded on NYSE, AMEX, and NASDAQ, only 60% of winner and loser stocks can persistently remain in that category for more than 1 moth post-formation period and about 25% of winner and loser stocks experience a contrarian effect. Thus, they construct a persistent momentum strategy in such a way that they only buy persistent winner stocks and sell only persistent loser stocks. This strategy yields an annual return of 15%, which outperforms a price momentum strategy by 19 basis points in monthly returns. It remains significant and robust even after controlling for risk different factors (i.e. market factor; Carhart’s (1997) momentum factor; Fama-French five-factor model (2015)).

Momentum among “smart money”

Evidence regarding the use of momentum strategies among professional investors has been extensively studied over the past two decades, with somewhat conflicting results. Jensen (1968), who can be seen as the forerunner of performance evaluation of mutual funds, provide the basis for a large number of studies and a large body of literature. However, some care has to be taken before analyzing mutual fund performance: the so-called “incubation bias”, which consists in launching multiple funds privately and then opens only at the end of the evaluation period the outperforming funds to the public. This practice and its role in the development of new mutual fund is documented by Evans (2010), finding that about 23% of funds were incubated and it serves to attract additional flows, rather than identify superior manager skills. According to the results, the incubation bias leads to an upward bias of returns. In fact, what he found is that comparing incubated funds with non-incubated funds during the incubation period, the first category outperformed the second by 3.5% (on risk-adjusted basis) and the Sharpe ratio is more than twice as high as for the non-incubated funds. The outperformance on the post-incubation is no statically significant, confirming the hypothesis that incubation is not effective to identify superior managers. Further, in a study of newly launched US equity mutual funds between 1991 and 2005, Karoui et al (2009) compare the performance changes and the portfolio characteristic for the first years of the funds. The study finds that on average, the new funds outperform their peers on a risk-adjusted basis by 0.15% per month in the first 3 years after the inception and the top-performing mutual funds exhibit short-term persistence. Analyzing the risk-characteristic of fund-starts, they find that the returns exhibit a higher ratio of unsystematic, and thus diversifiable, risk (i.e. higher exposure on small and illiquid stocks), rather than superior information. It is therefore not surprising that after three years the top-performing funds decline directly to the bottom decile ranking over the three next years. Recently, Bessler et al. (2018) study the performance persistence of US mutual funds and find that fund flows and manager changes act as “equilibrating mechanisms” (Berk and Green, 2004, p. 1271) explaining mean reversion in mutual fund performance and therefore why out- and under-performance of mutual funds are unlikely to persist. This findings also confirm the empirical results found in Bessler et al. (2016), in which diseconomies of scale are founded at the winner-level and only the winner small funds with low inflows significantly outperform their peers and the four-factor benchmark.

Grinblatt et al. (1995), did one of the earliest studies on investment strategies of “smart money” in which they examine the quarterly holdings of 274 US mutual fund between 1975 and 1984. The try to detect whether funds managers purchase stocks based on their past returns. The conclusion of their study is that 77% of the funds exhibit momentum behavior and that the ones who follow the momentum investment strategy have a significantly superior performance compared to the other funds. In light of this evidence, a further investigation needs to be done, since the funds’ performance seems to be correlated to the tendency to buy past winners and therefore the positive performance of mutual funds observed in Griblatt and Titman (1993) might be explained by simple trading rule rather than superior information. In addition, Nofsinger and Sias (1999) and Sias, Starks, and Titman (2001) find that institutional traders are momentum traders when they have to buy.

Badrinath and Wahal (2002), however, argue that the studies made so far on institutional trading practices were unable to capture the full range of institutional behavior. The methodology they develop allow them to study the trading patterns of the quarterly holdings for 1,200 different institutions (pension funds, mutual funds, investment advisor, etc.) over the period from 1987 to 1995. They find that institutions, especially growth or growth-and-value funds, are “momentum investors” when they enter a new buy order, and they act as contrarian when they sell stocks or make portfolio adjustments. Similar results are also been found by Sapp and Tiwari (2004) and Sias (2004).

The conclusion of their empirical tests is that the correlation between funds’ portfolios, especially large-cap growth and core oriented funds, is significantly strong and become even stronger as time passes (i.e. the correlation have risen up to 0.54 for 2002-2006).

On the other hand, Gomper and Metrick (2001), with a methodology that looks at changes in firm-level institutional ownership, find no evidence of momentum behavior. They rather find that large institutions invest more in large and liquid stocks that have low past returns. More recent, de Haan and Kakes (2011) show that the overall behavior for Dutch institutional investors is mostly contrarian. Thus, they mainly buy past losers and sell past winners. Although, from the three types of investors that they analyzed which are pension funds, life insurance companies, and non-life insurance companies, only pension funds are systematically contrarian traders.

Reasons for momentum trading

Previous studies have also investigated the motives why smart money managers engage in momentum trading. These reasons vary across the literature and behavioral explanation such as over- and under-reaction to information, are given. For instance, De Long et al. (1990) build a model where the so-called “rational speculator”, anticipating the action of positive feedback trader, buys today knowing that the price will rise in the future due to the presence of noise traders. Furthermore, the latter also might be even more excited with respect of what the rational speculators predict, and buys even more stocks, pushing the price beyond the fundamental value. A different explanation is given by Daniel et al (1998), arguing that professional investors overweight positive private signals. Therefore, they will buy even more stocks when good news is released, which confirmed their private information, but they will not sell if bad news is released since they are more confident in their private signal.

Hong and Stein (1999) argue that professional investors are momentum trader because public information are only slowly reflected in stock prices. In a similar way, Barberis et al (1998), explain momentum behavior as a result of a conservativeness bias model (i.e. adjustment of new information insufficiently) obtaining the same results.

Agency-problem reasons are discussed by Lakonishok et al (1992). They suggest that since a contrarian strategy takes too long to be confirmed, and in meantime, managers could be fired after few quarters of poor performances, they better off engage in momentum strategies, in other words, <<… better failed conventionally than to succeed unconventionally …>> (Keynes ).

Many other studies demonstrate that institutional investors avoid small stocks (Badrinath et al. 1989; Del Guercio 1996; Falkenstein 1996; Gompers and Metrick 2001; Bennet et al 2003) and Sias (2007) argue that such a behavior lead to momentum patterns when investor get rid of small loser stocks because now these stocks are become “too small”.

III. Methodology Data and Sample Construction

portfolio holding (i.e., March, June, September, and December) for the US mutual funds from December 2006 to June 2018, ending in 47 "holding-period," from Thomson Reuters Datastream.

Further, in order to have an unbiased dataset of mutual funds (i.e., survivorship bias free), I rule out all the funds that do not exist for all the entire 11-year period. Moreover, I exclude manually all the funds for which there were no sufficient quarterly data available (i.e., March, June, September, and December) and all the fund-of-funds, meaning that my final dataset consists of 60 (see table 1), out of a total of 255, active US mutual funds with investment focus worldwide. Quarterly performance for each mutual fund is also obtained through Thomson Reuters Datastream.

I also download the time-series of the prices of the S&P500 listed stocks during the 11 years through the Centre for Research in Security Prices (CRSP). The quarterly holdings of the fund, where each stock has at least a weight in the fund of 1%, are matched with the S&P500 constituent ticker list.

Finally, the common risk factors (market, size, value and momentum) are downloaded through the professor French’s website.

Table 1 Descriptive statistics of US-mutual funds

This table contains descriptive statistics of the US active mutual funds: The mean, median, and standard deviation are reported for the total net asset of the funds (TNA), net asset value per share (NVA), expense ratio, and the ten years percentage growth. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018.

The importance of mutual fund size for performance has been documented in the literature (see Yan, 2008; Ferreira et al., 2012; and Bessler et al., 2016). Thus, it is important to sort the full sample based on fund size (i.e., last observation of the total net asset) and split it into four equal groups. The classification is displayed in the table A2 in the “Appendix” section, where, for each quartile, I report the 10-year growth rate, total expense ratio, net asset value per share, and total net asset. From the table, it is evident that the expense ratio is negatively related to the fund size, whereas net asset value per share it is positively related to the size.

A possible division is based on the self-described mutual fund category, obtaining five different investment strategy categories: growth, value, volatility, yield, and other. Although these different fund categories are present in my sample, the inconsistency of some of the categories (e.g., only for the growth category there is a sufficient number of mutual funds) made it impossible for a proper analysis base on this self-described mutual fund category.

The “Winners-over-Losers” and “Winners-minus-Losers” ratios

In order to detect institutional momentum, a first and intuitive measure, follows from the construction of the strategy proposed by Jegadeesh and Titman (1993). Thus, securities of a portfolio of mutual funds are ranked in descending order based on their three months previous returns. Once I have got stocks ranked in every

Mean Median Std.

TNA ($) _{1,938,787,097 371,850,000 7,143,668,006}

NAV ($) 28.84 23.10 20.25

Expense Ratio 1.16 % 1.15% 0.30%

quarter, I split them in ten deciles, where the top decile is called “winners” decile and the bottom decile is the “losers” decile. Then, in each quarter, I calculate the following ratio:

(1) %& =

(_&,* +&,*

Where W represents the weight of winner stocks for the fund i (i=1…60) at quarter t (t=1,2,3,4), and L is the weight of loser stocks for the fund i at quarter t. Concerning equation (1), I detect momentum strategy if ,

- >

1, and contrarian trading behavior when ,

- < 1.

However, this ratio cannot be properly calculated when the sum of losers is equal to zero. Therefore, in order to circumvent this issue, I also calculate the difference of the sum between winners and losers.

(2) %_& = (_&,*− +_&,*

Thus if (&,*− +&,* is bigger than zero the momentum behavior is detected; otherwise, a contrarian strategy is

more likely.

The GWT Measure

In order to test the extent to which US-mutual funds engage in momentum strategies I follow the statistic proposed by Grinblatt et al. (1995) and which is extensively used in many other studies (e.g., Badrinath and Wahal 2002; De Haan and Kakes 2011; Curcuru et al. 2011). I define the momentum investing as:

(3) 2(3 = ∑ ∑7 (5_6,*

689 :

*89 − 56,*;9)<6,*;=

where 5_6,* is the weight of the security j (j=1, …N) at date t (t=1,…47), and <_6,*;= is the return of security j at t, and k is the duration over which the measure of return is lagged. For convention, I stick to the short momentum effect, and therefore, I am only interested in three prior months returns, k is equal to zero. The equation (3) represents the difference between the actual mutual fund which has rebalanced the portfolio at the beginning of the period, and the portfolio that the fund would have held if the previous holdings were kept (no revisions) during the three months-period (i.e., one quarter). Due to the equation, that is the difference between two portfolio returns during the benchmark period, I am able to analyze how much a fund manager exposes his portfolio to past winners stocks (respect to the benchmark period) rather than past losers. Therefore, a positive value means that the mutual fund, on average, has a portfolio that generated a superior return than the benchmark portfolio.

Following the above-mentioned literature, I focus on the recent momentum pattern, calculating the performance of the three months returns prior to the portfolio formation. However, since fund holdings are only available quarterly, whereas stock prices are available on a monthly base, I need to calculate the three months return for each stock in the S&P500 before using the equation (3).

The Carhart four-factor model

In order to provide robustness evidence for my result, I adjust the fund ’s performance for the common risk factors and check whether the momentum factor appears to be statistically significant for my sample.

First of all, I regress the excess return (defined as the monthly fund’s return minus the risk-free rate) of a fund on the four-factor model (see Carhart 97) as described by the equation (4):

where R is the excess return of the fund i at time t; α is the intercept, that shows (when positive) that the manager has skills and is adding persistent value to the portfolio performance ; AL represent the coefficients for the market (MKT), size (small-minus-big, SMB), value (high-minus-low, HML), and momentum (MOM) factors; ε is the residual error.

Moreover, as already discussed, value and growth do not appear to be a relevant category in my sample (i.e., too few funds are in the growth/value category), and therefore, I also use a modified version of the Carhart four-factor model, where the value factor has been removed (thereafter, Carhart three-factor model).

The test is aimed to detect how much a fund is exposed to the market, size, (value), and momentum. Therefore, a fund, which has a positive result for the GWT measure should have a significant and positive coefficients for the momentum factor. The opposite is also true: a contrarian fund should have negative A_B.

IV. Empirical Results

The presentation of the main findings is structured as follows: in the first paragraph I discuss the results for the two ratios that I used; the second paragraph highlights the results obtained from the GWT measure, and in the third, I describe the momentum measure with regard to size. Finally, I calculate the Sharpe ratio and the correlation between GWT measure and related performance. All data here presented are obtained by first calculating the measures for each fund, and once I have the time-series mean, I report the average at the aggregate level for convenience. More detailed tables follow in the Appendix section.

The W/L and W-L ratios

I first use the ratios above described in order to detect whether or not momentum-investing is such a popular strategy among mutual funds. In table 2, I report the result at the aggregate level for my sample, and more detailed findings for each fund are reported on the Appendix (see table A3). What emerges, analyzing 60 mutual funds and according to the results in table 2, that take into account the average at the aggregate level for both of the ratios, is that momentum is clearly implemented by the fund managers. The data shows the degree of which winner stocks are bought and loser stocks are sold. The bigger the value is different from 1, which is the value that the portfolio would have if no momentum was used by managers, the more the managers are exposing their portfolio towards winners and get ridding of losers. The opposite is also true: the smaller the value is and differs from one, the most likely is that the fund is using a contrarian strategy. Therefore, the figure in the chart (i.e., 1.41 for the mean) indicates that the funds on average are exposed 40% more towards previous winner stocks.

Table 2 Winners-over-Losers and Winner-minus-Losers descriptive statistics

This table contains descriptive statistics for the two ratios I used: the mean, median, standard deviation, minimum and maximum are reported after having obtained the time-series mean for each fund. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

(₊ ( − +

Mean

1.41

0.007

Median

1.16

0.005

Std

0.96

0.011

Minimum

0.00

-0.009

Maximum

5.12

0.051

The GWT measure

I start my analysis by calculating the GWT measure for each fund, using k=0, corresponding to contemporaneous return. After that, I calculate the value of GWT with respect to the fund's size.

Table 3 shows the mean, median, standard deviation, the maximum and the minimum of the GWT measure for all sample, consisting of 60 mutual funds between December 2006 and June 2018. The momentum measure is presented as percentage (that is, the estimate is multiplied by 100). I use quarterly fund-specific time-series means, obtaining a mean of 0.28%. The finding reveals that, on average, US mutual funds exhibit momentum-trading strategies. More specifically, a fund that re-adjusts its portfolio toward 3-months previous winners has a 0.28% percent higher returns than it would have had if no adjusting occurred.

Even though I find a positive and statistically significant GWT measure, its magnitude is economically insignificant. What however emerges by calculating the GWT measure for each fund is that I can have the dispersion of the GWT magnitude among my sample. As shown in table 3, the range of the momentum measure vary from a minimum of -3.2% to a maximum of 1.91%, which is more economically significant than the mean figure. Furthermore, I sort my sample into momentum traders (i.e., funds with positive GWT) and contrarian traders (i.e., funds with negative GWT), as illustrated in the Appendix, finding that among my sample, it appears that momentum is widely used and only four out of 60 funds exhibit contrarian strategies.

Overall, the results are not surprising, but rather confirm the findings discover by Grinblatt et al. (1995).

Table 3: GWT descriptive statistics

This table contains descriptive statistics for the GWT measure: the mean, median, standard deviation, minimum and maximum are reported after having obtained the time-series mean for each fund. The second and third columns present the descriptive statistics based on momentum and contrarian funds. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

GWT (%) Momentum (%) Contrarian (%)

Number of Funds 60 56 4

Mean 0.28 O.40 -1.37

Median 0.25 0.27 -1.12

Minimum -3.25 0.01 -3.25

Maximum 1.91 1.91 0.00

Momentum trading with regard to size

It is broadly recognized that the importance of a company’s size on its stock return, as documented since Fama and French (1992). In the same way, the effect of a fund's size on its return has been studied since Grinblatt and Titman (1989), later on by Carhart (1997), and Chen et al. (2004), until recently by Ferreira et al. (2012). Therefore, it is necessary to examine the momentum behavior in relation to their size in terms of a total net asset.

On one hand, the larger the fund, the less likely it is to engage in momentum strategies compared to smaller-sized funds, given that the former has the advantage of using in-house research and more private information, and therefore are less prone to herding, that is following the crowd and engage in momentum behavior. On the other hand, it can also be the case that larger funds, with a more extensive portfolio, are more likely to exhibit momentum-trading strategies due to the complexity of the portfolio since they are less prone to do research at the individual stock level but instead rely on a quick and valuable source of information.

For this purpose, I use fund-specific time-series means after having them sorted in four quartiles. In Table 4, I report the main findings. It appears that the fund which is more prone to engage in momentum trading, is in the "small-medium" category (i.e., between 107,525 and 371,850 million TNA), and small funds ( i.e., under 107,525 million TNA) are the one with the lowest GWT measure of 0.29% followed by big-size funds (i.e., bigger 1,154,750 million TNA).

Table 4 GWT summary statistics by fund size (in $M)

This table presents the descriptive statistics for the GWT measure segregated by size: the mean, median, standard deviation, minimum and maximum are reported after having obtained the time-series mean for each fund. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

Momentum trading strategies and fund performance

In order to analyze the relationship between fund performance and momentum behavior I calculate the monthly fund’s Sharpe ratios for all the funds. The Sharpe ratio of a given fund in date t is equal to:

(5) FℎOPQR POSTU = V&,*− P* W&,*

"Small":

size≤107,525 107,525<size≤371,850Small-medium: 371,850<size≤1,154,750 Medium-big: size>1,154,750 Big: Number of

funds 15 15 15 15

Mean (%) 0.29 0.82 0.47 0.39

Median (%) 0.39 0.24 0.31 0.19

where V_&,* is the rolling return of fund i in month t; P_* is the risk-free in month t; and W_&,* is the standard deviation of the return for the fund i.

The average monthly Sharpe ratio for the all sample between December 2006 and June 2018 is 0.10, compared to the market Sharpe ratio of 0.16, indicating that the funds achieved low return per unit of risk. The result, depicted in table 5, also shows that, despite the average underperformance of the active mutual funds, there are funds able to outperform the market with the highest Sharpe ratio reached of 0.19. After checking for size, it appears that big size funds, on average, has the highest Sharpe ratio of 0.12, still smaller than the market’s Sharpe ratio, whereas small-size funds achieve the smallest Sharpe ratio of 0.088.

Table 5 Sharpe ratio

This table contains the descriptive statistics of the monthly time-series means for the Sharpe ratio: in the first column I depict the average correlation that take into account all the funds in my sample. The other columns represent the Sharpe ratio with regard to fund’s size. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

Moreover, as documented by Grinblatt et al. (1995), the fund performance is significantly correlated to momentum trading. Therefore, I also investigate the level of correlation between quarterly momentum measures and the related portfolio performance, lagged by two and five months. The results, presented in table 6 shows, on average, a negative correlation of 0.029 for the full sample. In unreported calculation, I find the same results after have lagged the return for one quarter in order to give the time to momentum to adjust.

Finally, I split the sample between funds with a positive correlation and funds with negative, resulting in nearly equally split categories. Table 6 also reports the average with the positive and negative correlation.

Table 6 Correlation between GWT measure and return

This table contains the descriptive statistics of quarterly time-series means for three different groups: in the first column I depict the descriptive statistics for the average correlation that take into account all the funds in my sample; in the second and the third columns I split my sample between funds with negative correlation and funds with positive correlation. The mean has a confidence level of 95%.

All

Negative

Positive

Number of funds 60 29 31

Mean -0.029 -0.175 0.112

Median 0.010 -0.142 0.104

All

Small:

size≤107,525 107,525<size≤371,850Small-medium: 371,850<size≤1,154,750 Medium-big: size>1,154,750 Big:

Std 0.185 0.147 0.073

Minimum -0.550 -0.550 0.010

Maximum 0.295 -0.005 0.295

V. Robustness Test

In order to check whether a portfolio manager is tilting his portfolio towards previous winner stocks (and get rid of past losers), I adjust the excess return of each fund for risk, using the Carhart four-factor model as represented by equation (4), focusing in particular on the AB (that is, the beta for the momentum factor).

Therefore, if a fund’s manager engages in momentum strategy, I expect a significant positive AB. The opposite

is also true: a negative significant AB indicate a likelihood of a contrarian strategy.

Since all my research focuses on single fund trading activity, but at the same time I cannot represent each fund regression’s outcome, I decide, in order to avoid outliers biases, to split the sample between funds with positive A_B, and the ones with negative beta; finally, I calculate the aggregate descriptive statistics for each group. Table 7 shows that eleven funds hold the momentum factor, and only two are significant.

As mentioned in the previous section, I consider appropriate to do the same analysis but only regressing on market, size and momentum factors. Removing the value factor (that is, high-minus-low factor), the number of funds with a positive (and significant) AB increase dramatically. The results, depicted in table 7, show that

twenty (of which only nine are significant) out of 60 funds show positive betas for momentum.

Table 7

This table contains the findings for the Carhart four-factor (first two columns) and three-factor (last two columns) regressions. I take the descriptive statistics for the two different models. I divide the results between funds with a positive A! (positive MOM), and

negative A! (negative MOM). The mean has a confidence level of 95%.

Given the fact that on average the GWT measure is economically not relevant (i.e., 0.28%), which is in line with the magnitude of the outcome of Grinblatt et al. (1995), the findings of my robustness check is not

Carhart four-factor

Carhart three-factor

Positive

MOM

Negative

MOM

Positive

MOM

Negative

MOM

surprising. Furthermore, the funds with a GWT measure economically relevant (i.e., bigger than 1%), exhibit significant positive AB when I check for the Carhart three-factor model, confirming the result of my analysis.

On the other hand, table 7 also shows that 40 (which 20 are significant) of my sample present negative beta for momentum, that is, these funds are more likely to engage in contrarian strategy. In the table A3 in the Appendix, I depict the AB for each fund.

Finally, in the results plotted in the Appendix section (see table A4), I find value and size factor have on average a negative beta, indicating that the funds in my sample are more likely to invest in big and growth companies.

VI. Conclusion

The primary purpose of my thesis is to investigate whether and to which degree professional investors, specifically active US-mutual fund, engage in momentum behavior. The implication of this is that the active managers are more likely to follow simple trading rules (i.e., previous return) rather have superior skills (a better information valuation model) and/or access to superior information.

In order to answer my research question, I follow the important study of Grinblatt et al. (1995), analyzing the portfolio for each fund, for a total number of 60 active US-mutual funds. I detect fund momentum strategies using two methodologies. The first one is based on Jegadeesh and Titman (1993) with respect to the way they construct a portfolio in order to identify winners and losers; after that, I calculate the winners-over-losers and the winners-minus-losers ratios for every quarter for each fund portfolio. The second analysis replicates the equation used by Grinblatt et al. (1995), called GWT measure. This measure calculates the degree to which portfolio managers tilt their portfolio towards the three months previous winners, and get rid of the losers compared to a benchmark portfolio. The benchmark portfolio is the one where no portfolio holding adjustment were made from one quarter to the next one, keeping the portfolio allocations from the previous quarter constant. Hence, abnormal returns are due to portfolio adjustments.

Both of the methodologies provide evidence for momentum in my sample, even though it is not economically significant. Overall, my results are in line with the findings of Grinblatt et al. (1995), which even if positive, have no economic significance.

More interesting results, however, regards the variation of the GWT measure I obtain with respect to the fund-size category. I show that the small-medium funds, that is the ones with total net assets between 107,525 and 371,850 million, have the highest momentum measure, and the smallest ones with the lowest GWT measure. By calculating GWT measure for each fund, it allows me to sort the sample based on the momentum measure, detecting the presence of four funds that exhibit contrarian strategies (i.e., negative GWT measure), which find support for the studies done by Gomper and Metrick (2001) and more recently by De Haan and Kakes (2011). On average, the funds on my dataset underperform the market on a risk-adjusted monthly base, with a Sharpe ratio of 0.10 against 0.16 for the market. Further, when I compare the momentum measure to the related quarterly return, I find, on average, a negative correlation of 0.029, even if the percentage of negative correlation among the funds represents only half of the sample. Finally, the robustness check suggests that although on average the funds in my sample have negative beta for the momentum factor, the funds with the highest GWT measure, have significant and positive A_B, providing support to my analysis. Hence, it is important to group the sample appropriately.

the weights of all other stocks. In principle, this is a fund flow issue. Elton et al. (2010) argue that GWT measure does not significantly change when it is controlled for fund flow.

The second issue arises, because the holding information is available only at the end of each quarter I assume in my research that the actual portfolio shifts occurred before the change in the stock price is realized. This is a simplification since the exact date of trade is unknown, but still it is the best approximation I can make.

Appendix:

Table A1: Characteristics of mutual funds in my dataset

This table contains the descriptive statistics for the characteristics of the US active mutual funds: The mean, median, standard deviation, minimum, maximum, first and third quartile are reported for the growth in the last ten years, total net asset of the funds (TNA), net asset value per share (NVA), expense ratio, and the ten years percentage growth. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

Mean Median Std Min Max Q1 Q3

Growth 10-Year 101.92% 99.44% 31.17% 30.87% 204.62% 83.57% 118.72%

NAV 28.84 23.10 20.25 6.19 120.70 16.31 33.96

Total Expense Ratio 1.16% 1.15% 0.30% 0.48% 2.40% 0.96% 1.32%

Latest Fund Value

Table A2. Growth, Net Asset Value, Expense Ratio, and Total Net Asset by Mutual Fund Size

This table contains the main characteristics ( the ten years percentage grow, the net asset value per share expense ratio, and the net asset of the funds ) of the US active mutual funds divided by size. I divide my dataset based on size in four equally-split quartile. The mean, median, standard deviation, minimum and maximum values are reported for each category. The size of the sample is 60 US-mutual funds active between December 2006 and June 2018. The mean has a confidence level of 95%.

Fund size ($M) Mean Median Std Min Max

"Small": size≤107,525 (N=15)

Percentage Growth 10 Year 89.96% 87.17% 34.33% 30.87% 138.30%

NAV 21.74 19.06 10.55 6.19 40.83

Total Expense Ratio 1.26% 1.06% 0.47% 0.76% 2.40%

Latest Fund Value (TNA) 42,881,250 34,300,000 33,762,251 3,000,000 105,400,000

Small-medium: 107,525<size≤371,850

(N=15)

Percentage Growth 10 Year 106.92% 93.57% 32.17% 74.99% 204.62%

NAV 26.98 22.85 11.96 9.13 44.81

Total Expense Ratio 1.23% 1.27% 0.12% 0.97% 1.37%

Latest Fund Value (TNA) 210,700,000 186,700,000 70,898,207 113,900,000 349,300,000

Medium-big: 371,850<size≤1,154,750

(N=15)

Percentage Growth 10 Year 101.55% 101.08% 30.57% 60.30% 188.06%

NAV 31.66 23.23 28.34 8.69 120.70

Total Expense Ratio 1.20% 1.19% 0.16% 0.84% 1.44%

Latest Fund Value (TNA) 642,273,333 630,500,000 191,554,668 394,400,000 1,021,400,000

Big: size>1,154,750 (N=15)

Percentage Growth 10 Year 109.54% 103.83% 26.35% 67.46% 178.10%

NAV 35.03 29.59 23.80 12.00 98.26

Total Expense Ratio 0.95% 0.99% 0.19% 0.48% 1.15%

Table A3

This table reports the descriptive statistics for each fund for the Grinblat et al. (1995) measure (GWT), the winner-over-loser ratio (W/L), and the winner-minus-loser ratio (W-L). In the last two columns, I report the coefficients of the momentum factor for the regression of the Carhart four-factor model (four-factor), and Carhart three-factor model (three-factor). Note: a single asterisk indicates significant at 10% level (two- tailed t-test), a double asterisk indicates significant at 5% level, and triple asterisk indicates significant at 1% level.

GWT

W/L

W-L

four-factor

three-factor

Mean

Median

Std

Mean

Median

Mean

Median

$_% $_%

AB Sustainable

Global Thematic

Fund

0.46%

0.21%

1.30%

5.12 0.64 0.02 0.01 -0.12 -0.02

AB Tax-Mgd

Wealth Appr

0.47%

0.28%

1.02%

2.45 0.98 0.00 0.00 -0.03 -0.01

Aberdeen Global

Equity Fund

0.39%

0.52%

2.62%

0.65 0.03 0.01 - -0.12* -0.07*

Alger Global Focus

Fund

-2.16%

0.09% 17.30%

2.02 1.32 0.02 0.01 -0.09*** 0.03

American Century

Global Growth

1.55%

1.47%

2.83%

2.30 1.17 0.02 0.03 0.03 0.10***

American Funds

Capital World

Grow

0.24%

0.14%

0.61%

1.84 0.06 0.01 0.00 -0.07 -0.02

BlackRock

Advantage Global

Fund;

0.05%

0.00%

0.25%

0.72 0.01 0.00 - -0.03*** 0.00

Columbia Global

Equity Value Fund

0.59%

0.47%

0.52%

1.39 0.57 0.01 0.00 -0.03 -0.04**

Columbia Select

Global Equity Fun

0.50%

0.43%

0.71%

0.92 0.41 0.01 0.00 -0.02*** 0.05

Guinness Atkinson

Global Innovators

0.42%

0.35%

0.79%

1.33 0.82 0.02 0.02 -0.10*** -0.06*

Harding Loevner

Global Equity

Portfolio

0.03%

0.00%

0.12%

0.93 0.70 0.00 0.00 -0.11 -0.04

Hartford Global

Capital

Appreciation

0.51%

0.28%

0.77%

2.90 1.36 0.01 0.01 -0.06*** -0.03

Hartford Global

Growth HLS

0.77%

0.68%

0.67%

2.99 1.52 0.03 0.02 0.00 0.09**

Invesco Global

Core Equity Fund

0.10%

0.14%

0.46%

0.96 0.52 -0.01 -0.01 -0.09* -0.07**

Invesco Global

Growth Fund

0.24%

0.21%

0.26%

1.09 0.59 0.00 0.00 -0.03** 0.02

Ivy Global Growth

Fund

0.34%

0.04%

0.73%

2.45 0.49 0.02 - -0.01*** 0.07*

Janus Henderson

Global Research

Fund

0.25%

0.25%

0.41%

1.92 1.15 0.01 0.01 -0.14* -0.06*

Janus Henderson

Global Value Fund

0.97%

0.02% 14.63%

0.46 0.02 0.00 -0.00 -0.25** -0.22***

MainStay Epoch

Global Choice Fund

0.77%

0.60%

0.99%

0.92 0.36 0.02 0.01 -0.01*** -0.01

MainStay Epoch

Global Equity Yield

0.14%

0.06%

0.31%

0.89 0.09 0.00 - -0.01*** 0.01

MassMutual

Premier Global

Fund

MFS Global Equity

Fund

0.15%

0.15%

0.15%

0.71 0.21 -0.01 -0.01 -0.06*** -0.03

MFS Global

Growth Fund

0.18%

0.17%

0.26%

1.21 0.45 0.00 -0.00 -0.10** -0.03

Morgan Stanley

Inst Global

Franchise

0.19%

0.17%

0.46%

0.33 0.03 0.00 - -0.10*** -0.05

Mundoval Fund

0.94%

0.25%

4.57%

1.19 0.17 -0.00 -0.00 -0.16*** -0.16***

Nationwide Global

Sustainable Equity

0.46%

0.37%

0.74%

1.79 0.93 0.01 0.00 -0.14* -0.11***

Oakmark Global

Fund

0.17%

0.17%

0.43%

0.98 0.37 -0.01 - -0.17 -0.19***

Oakmark Global

Select Fund

0.11%

0.13%

0.88%

0.43 0.02 -0.00 - -0.17** -0.21***

Oppenheimer

Global Fund

0.29%

0.20%

0.27%

1.33 0.64 0.01 0.01 -0.09 -0.04

SEI Inst Mgd

Global Managed

Volatility

0.05%

0.06%

0.16%

0.72 0.01 -0.00 - 0.01 0.06**

T Rowe Price

Global Stock Fund

0.21%

0.14%

0.66%

1.64 0.72 0.02 0.01 0.04* 0.00

T Rowe Price

Institutional Global

0.23%

0.14%

0.75%

1.96 0.82 0.02 0.01 -0.09* 0.01

Templeton Growth

Fund

0.21%

0.13%

0.44%

1.63 0.46 -0.00 - -0.08** -0.11***

Templeton World

Fund

0.29%

0.13%

0.82%

1.18 0.17 -0.00 - -0.12*** -0.08**

Thornburg Global

Opportunities Fund

0.44%

0.27%

0.99%

1.05 0.06 0.01 0.00 -0.09*** -0.09*

Thrivent Large Cap

USAA World

Growth Fund

0.16%

0.15%

0.15%

0.72 0.21 -0.01 -0.00 0.05** -0.03

Vanguard Global

Equity Fund

0.14%

0.09%

0.22%

1.14 0.02 0.01 0.01 -0.06** -0.04

Wasatch Global

Value Fund

0.29%

0.35%

0.53%

0.88 0.56 -0.00 -0.01 -0.09 -0.02

Wells Fargo

Intrinsic World

Eqty Fd

0.25%

0.22%

0.64%

0.54 0.04 0.00 - -0.21*** -0.12*** Table A4

This table reports the descriptive statistics of the coefficients for the Carhart four-factor and Carhart three factor regressions. From the first column on the right hand-side I plot the market factor in excess to the risk-free rate (Mkt-RF), the size factor (SMB), the value factor (HML), and the momentum factor (MOM).The mean has a confidence level of 95%.

Mkt-RF

SMB

HML

Mom

Four-factor Three-factor Four-factor Three-factor Four-factor

Four-factor Three-factor

Mean

1.01

1.01

-0.10

-0.12

-0.15

-0.07

-0.04

Median

1.04

1.04

-0.12

-0.13

-0.16

-0.08

-0.03

Std

0.13

0.13

0.13

0.13

0.13

0.07

0.08

Min

0.58

0.58

-0.37

-0.40

-0.44

-0.32

-0.30

References

Badrinath, S.G., Gay, S.D., and Kale, J.R., 1989. Patterns in institutional investment, prudence, and the managerial safety-net hypothesis. The Journal of Risk and Insurance, 56, 605-629.

Badrinath, S.G., Wahal, S., 2002. Momentum trading by institutions. Journal of Finance, 57(6), 2449–2478.

Barberis, N., Shleifer, A., and Vishny, R.W., 1998. A model of investor sentiment. Journal of Financial Economics, 49, 307-343.

Barroso, P., Santa-Clara P. 2015. Momentum has its moments. Journal of Financial Economics, 116,111-120.

Bennett, J., R. Sias, and L. Starks, 2003. Greener pastures and the impact of dynamic institutional preferences. Review of Financial Studies, 16, 1203-1239.

Bessler, W., Blake, D., Luckoff, P., and Tonks, I., 2018. Fund Flows, Manager Changes, and Performance Persistence. Review of Finance, 1911-1947.

Bessler W., Kryzanowski, L., Kurmann P., and Lückoff P., 2016. Capacity effects and winner fund performance: the relevance and interactions of fund size and family characteristics. The European Journal of Finance, 22 (1), 1-27.

Carhart, Mark M., 1997. On Persistence in Mutual Fund Performance. Journal of Finance, 1997, 57–82.

Chaves, D., 2012. Eureka! A momentum strategy that also works in Japan. Unpublished working paper. Research Affiliates, Newport Beach, California.

Chen, J., Hong, H., Huang, M., and Kubik, J.D. 2004. Does fund size erode mutual fund performance? The role of liquidity and organizatio.. American Economic Review, 94(5), 1276 -1302.

Chui, A., Titman, S., and Wei, J., 2010. Individualism and momentum around the world. Journal of Finance, 65, 361–392.

Curcuru, S.E., Thomas, C.P., Warnock, F.E., Wongswan, J., 2011. US international equity investment and past and prospective returns. American Economic Review, 101(7), 3440–3455.

Daniell, K., Hirshleifer, D., andSubrahmanyamm, A., 1998. Investor psychology and security market under- and overreactions. Journal of Finance, 53, 1839-1885.

Daniel, K., Moskowitz, T., 2016. Momentum crashes. Journal of Financial Economics, 122, 221-247.

De Bondt, W., Thaler, R., 1985. Does the stock market overact? Journal of Finance, 40 (3), 793-805.

De Haan, L., Kakes, J., (2011) Momentum or contrarian investment strategies: evidence from Dutch institutional investors. Journal of Banking Finance, 35(9), 2245–245.

De Long, B.J., Shleifer, A., Summers, L., and Waldmann ,R., 1990. Positive feedback investment strategies and destabilizing rational speculation. Journal of Finance, 45 (2), 379-395.

Del Guercio, D., 1996. The distorting effect of the prudent-man laws on institutional equity investment. Journal of Financial Economics, 40, 3-62.

Elton, E.J., Gruber, M.J., Blake, C.R., Krasny, Y., Ozelge, S., 2010. The effect of holdings data frequency on conclusions about mutual fund behavior. Journal of Banking Finance, 34(5), 912-922.

Evans R., 2010. Mutual Fund Incubation. Journal of Finance, 65 (4), 1581-1611.

Falkenstein, E.G., 1996. Preferences for stock characteristics as revealed by mutual fund portfolio holdings. Journal of Finance, 51, 111-135.

Fama, E.F., 1965. The behaviour of stock market prices. Journal of Business, 38, 34-105.

Fama, E., French, K., 2012. Value, size, and momentum in international stockreturnss. Journal of FinancialEconomicss, 105, 457–472.

Ferreira, M. A., Keswani, A, Miguel, A. F., and Ramos, S. B., 2012. The Determinants of Mutual Fund Performance: A Cross-Country Study. Review of Finance, 17, 483-525.

Galariotis E. C., Holmes P., Xiaodong S. Ma, 2007. Contrarian and momentum profitability revisited: Evidence from the London Stock Exchange 1964–2005. Journal of Multinational Financial Management, 17, 432-447.

Gompers, P., Metrick, A., 2001. Institutional investors and equity prices. Quarterly Journal of Economics, 116, 229-260.

Grinblatt, M., Titman, S., 1989. Mutual fund performance: an analysis of quarterly portfolio holdings. Journal of Business, 62(3), 393–416.

Grinblatt, M., Titman, S., 1993. Performance Measurement without Benchmarks: An Examination of Mutual Fund Returns. Journal of Business, 1993, 66(1), 47-68.

Grinblatt, M., Titman, S., Wermers, R., 1995. Momentum Investment Strategies, Portfolio Performance, and Herding: A Study of Mutual Fund Behavior. American Economic Review, 85 (5), 1088- 1105.

Grossman, J. S., Stiglitz, E.J., 1980. On the impossibility of informationally efficient markets. American Economic Review, 70 (3), 393-408.

Hendricks, D., Jayendu P., Zeckhauser, R., 1993. Hot hands in mutual funds: Short-run persistence of performance, 1974-88. Journal of Finance, 48, 93-130.

Hong, H., Stein, J.C., 1999. A unified theory of underreaction, momentum trading and overreaction in asset markets. Journal of Finance, 54, 2143-2184.

Jegadeesh, N., Titman, S., 1993. Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. The Journal of Finance, 48 (1), 65-91.

Jensen, M., (1968) The Performance of Mutual Funds in the Period 1945-1964. Journal of Finance, 23, 389-416.

Lakonishok, J., Shleifer, A., and Vishny, R.W., 1992. The impact of institutional trading on stock prices. Journal of Financial Economics, 32, 23-43.

Karoui, A., Meier, I., 2009. Performance and characteristics of mutual fundstarts. The European Journal of Finance, 15, 487-509.

Mazouz, K., Alrabadi. D.W.H., and Yin, S., 2012. Systemic liquidity risk and stock price reaction to shocks. Accounting and Finance, 52, 467-493.

Moskowitz T., Ooi Y., Pedersen L., 2012. Time series momentum. Journal of Financial Economics, 104, 228–250.

Mulvey, J., Kim W. C., 2008. Active Equity Managers in the U.S.: Do the Best Follow Momentum Strategies? Journal of Portfolio Management, December 2008, 126-134.

Nofsinger, J., Sias. R., 1999. Herding and Feedback Trading by Institutional and Individual Investors.” Journal of Finance, December 1999, 2263–2295.

Novy-Marx R., 2012. Is momentum really momentum?. Journal of Financial Economics, 103, 429 - 453. Roberts, H., 1967. Statistical versus clinical prediction of the stock market. Unpublished Manuscript, Center for Research in Security Prices, University of Chicago.

Rouwenhorst, K.G., 1998. International momentum strategies. The Journal of Finance, 53(1), 267–284. Sapp, T., Tiwari, A., 2004. Does Stock Return Momentum Explain the ‘Smart Money’ Effect?. Journal of Finance, 2605–2622.

Sias, R., Starks, L., Titman, S., 2001. The Price Impact of Institutional Trading. Unpublished working paper.

Yan, X. S., 2008. Liquidity, Investment Style, and the Relation between Fund Size and Fund Performance. Journal of Financial and Quantitative Analysis, 43 (3), 741-768.

Kenneth French. 2018. Kenneth R. French – Data Library. [ONLINE] Available at

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html#Research. [Last access 26