Do Better Educated Mutual Fund Managers Outperform Their Peers?

Do Better Educated Mutual Fund Managers Outperform Their Peers? By P.F. van Laarhoven Tilburg University School of Economics and Management Supervisor: A. Manconi Master s program in Finance 22-08-2014 Abstract I examine the relationship between managerial characteristics and mutual fund performance. Using a novel dataset of 2,807 US mutual fund managers, I test whether the managerial characteristics of the mutual fund manager in charge with a strong focus on education - are related to the performance and characteristics of the funds that she manages. I find that the quality of the undergraduate institution - measured by SAT scores, and the presence of specific degrees like having an MBA/CFA designation positively and predominantly statistical significantly influences the performance of the concerned mutual fund. I also find that managers that hold an MBA degree take on more idiosyncratic risk to outperform the market. On the other hand, this does not apply to managers that hold a CFA designation, who tend to be less exposed to idiosyncratic risk.

Table of Contents I. Introduction... 4 II Hypothesis Development... 7 II.I Literature review... 7 II.II Hypotheses... 9 H1 Mutual fund managers who attended higher and more education tend to earn higher excess returns.... 10 H2 Better educated mutual fund managers tend to manage bigger firms which have more Investment Inflow.... 10 H3 Mutual fund managers that hold a degree in finance or beta related studies tend to outperform degrees that are indirect related to the mutual fund industry.... 10 H4 Education has an influence on the investment style of mutual fund managers.... 10 H5 Managerial characteristics influence the extent to which mutual fund managers hold on systematic risk.... 11 II. Data and methodology... 11 III.I Exogenous variables... 11 III.II CRSP Mutual Fund Information... 13 III.III Endogenous variables... 14 III.IV Summary statistics and Correlation Matrix... 16 III. Empirical results... 16 IV.I Simple Excess return OLS... 16 IV.III Risk Adjusted Performance Measures... 18 IV.IV Sharpe (1964), Fama-French (1992) and Carhart (1997)... 19 IV.V Education specific performance measures... 19 IV.VI Fama-French Factors OLS... 20 VI.VII Systematic and Idiosyncratic risk... 21 IV.VIII Endogeneity: Discussion... 22 IV. Conclusions... 23 V. References... 25 Articles... 25 Books... 25 VI. Appendix... 26 Table I... 26 Table II... 27 2

Table III... 28 Table IV... 29 Table V... 30 Table VI... 31 Table VII... 32 Table VIII... 34 Table XI... 35 3

I. Introduction Mutual funds or investment companies play an important role in the U.S. stock market. Nearly 25% of household financial assets consist of positions in these funds. There is, particularly in the U.S., a wide range of mutual fund operators. In light of the extant literature, it might come as a surprise that mutual funds are so popular among retail investors. Basically, at least since the work of Malkiel (1979), we have noticed that on average mutual funds do not beat the market, i.e. they do not outperform a diversified market index. He states in his classic finance book A Randow Walk Down Wall Street that asset prices follow a random walk and that investors therefore cannot outperform the market averages. However, (cf. Kacperczyk et al., 2005) find that at least some mutual funds are able to earn abnormal profits. The question if mutual funds can outperform their market averages is therefore an important and popular finance question, since positive risk-adjusted returns has implications for market efficiency. Besides Malkiel (1979) and Kacperczyk et al. (2005), other former studies tried to shed light on this question as well with mixed results. Where other studies in general tried to measure if the risk-adjusted return is positive from an investor prospective only, will this study follow another approach and look especially to the relationship between performance and managerial characteristics. Investors seem to seek for specific investment strategies to realize higher returns than the market will offer. CBS News (July 31, 2012) reported for instance that the endowments at Ivy League universities such as Yale and Harvard are exceptionally successful, causing many investors both individual and institutional to consider replicating their investment strategy. This development could perhaps clarify that managerial characteristics and especially education, has an influence on 4

the performance of investment entities. On the other hand, a recent article by Business Insider (April 17, 2014) reveals that having an MBA is a waste of money, since participants faces besides the huge tuition fees also the opportunity costs of not getting a full salary during that time. In this study, I try to look cross-sectionally at how performance of mutual funds is related to the concerning characteristics of the mutual fund manager. Hence, it is irrelevant for this study if the average risk-adjusted returns of the mutual funds within the sample are positive. I want to examine whether there do exist cross-sectional differences in performance by looking at multiple time periods for the same managers and mutual funds. For instance, when having an MBA predicts positively performance, then can this be in place even if all the funds within the sample have negative risk-adjusted performance, since managers who are holding an MBA degree will have on average less negative performance. I use a sample of 2,708 US mutual fund managers and try to look whether managerial characteristics are related to mutual fund performance. In addition to Chevalier and Ellison (1999a), I included several new educational variables to examine the effect on fund performance. I observe that educated managers (proxied by the presence of their degrees) are able to earn significantly higher risk-adjusted returns. This finding is not just statistically significant, but also economically relevant: one standard-deviation higher SAT sore of the manager s undergraduate institution (corresponding to the difference between Yale and Alfred University 1 ) is associated with higher alpha of one basis point per year. Furthermore, I document similar results to that of Gottesman and Morey (2006) regarding the relationship between fund size and having an MBA, since larger firms tend to hire non-mba managers. Besides, I find a strong negative relationship between gender and firm size as well as investment inflow. Female managers are less likely to manage a large 1 SAT score of respectively 2240 and 1340. 5

mutual fund and do have less investment inflow. Moreover, I have tried to analyse the investment strategies of mutual fund managers. I observe that managers that attend an undergraduate institution with lower requirements for admission, hence a higher admission rate, tend to load more on Small minus Big strategies. In addition to this, age and tenure seem to have a minor impact on different strategies. Where other studies like Chevalier and Ellison (1999a) report that older managers tend to use more momentum strategies, does it not seem to matter for this study. Finally, I find that managers that hold an MBA degree do take on more idiosyncratic or idiosyncratic risk to outperform the market. On the other hand, this does not count for managers that hold a CFA designation. They tend to hold on more systematic risk and less idiosyncratic risk. The remainder of the thesis proceeds as follows. In Section I is the Hypothesis Development formulated. The Data and Methodology is described in Section II. Section III presents the empirical results, where I have tried to examine the relationship between managerial characteristics and fund performance. In section IV are the conclusions drawn. Section V sums up a reference list and section VI contains the appendix where all the tables and results are presented. 6

II Hypothesis Development The rewarding and interest for mutual funds is actually quite odd since there is a lot of competition and the uncertainty about the future values of the investments exists largely. One of the oldest questions in finance (Jensen, 1968) is whether mutual funds do add value and if they can outperform passive benchmarks. II.I Literature review Former studies regarding fund performance have revealed several different outcomes, but in general we might say that mutual funds do not add value according to these papers. Jensen (1968) stated that the 115 mutual funds within their sample where not able to beat the, so called, buy-and-hold-strategy which indicates the index investors. They document very little evidence that any individual fund was able to outperform their passive benchmarks. Furthermore, these conclusions hold even when the funds gross returns of management expenses where measured, which means that they assumed that their management fee was zero and they only faces brokerage commissions. According to Gruber (1999) mutual funds offer on average a negative risk adjusted return and investors are better off by buying index funds, but since future performance are somewhat predictable from lagged performance, sophisticated investors are still interested. Wermers (2000) document that on the one hand mutual funds do outperform a broad market index. However, taken transactions and fees into account, they outperform with 1% per year comparing to the index. They reported their results over the period 1975 to 1994 and 7

stated that mutual funds held stock portfolios that outperform a broad benchmark index by 1.3 percent per year. 60 basis points are due to higher average returns, while 70 basis points are because of talents in picking stocks that beat their benchmark portfolios. On the other hand, when taken net returns into account, they underperformed the benchmark. This difference of 2.3% (-1% and +1.3%) is for 70 basis points duo to underperformance of nonstock holdings while the remaining 160 basis points are due to expenses and transaction costs. In addition, Ding, B., and R. Wermers (2009) find that experience and (advisor-level) stock picking track-record of a fund manager are correlated with following-year performance. Thereby, they document that larger boards are associated with better performance, when the manager has more than 10 year experience. Chevalier and Ellison (1995b) looked more to the characteristics of the fund itself. Contrary to these results, (cf. Kacperczyk et al., 2005) examined the relationship between industry concentration and the performance of actively managed U.S. mutual funds between 1984 and 1999 and they find, on average, that industry concentration performs better. Previous mentioned papers have examined if there is a possibility to outperform and if there is evidence for persistence over time. Regardless of whether it is, an interesting question that arises is where it comes from. Chevallier and Ellison (1999a) looked to the managerial characteristics instead of the fund. They document in their paper Are Some Mutual Funds Managers Better Than Others? Cross-Sectional Patterns in Behavior and Performance that higher education has a positive influence on fund performance. They used a sample of 492 managers who had sole responsibility for a growth or growth and income fund for at least some part of the 1988-1994 8

period. They gathered the majority of their data from Morningstar and used variables as managerial SAT scores, having an MBA or not, age and tenure to explain the variation in the mutual fund returns. First, they examined that higher SAT Scores (in this example 1355 or 1142) and having an MBA leads to higher simple excess returns of respectively 98.6 and 63.1 basis points. Secondly, they regress beta, log of assets, expense ratio% and turnover ratio on the latter variables and find among other that higher educated fund managers tend to manage higher beta funds. Therefore, a risk-adjusted test is also executed. Notably for this measure is that they reported a drop of the coefficient MBA from 0.63 to 0.04, indicating that the higher returns achieved by MBAs are essentially completely attributable to their taking on more systematic risk. Finally, their overall conclusions are that mutual funds managers, who attended more selective undergraduate institutions, seem to earn higher risk-adjusted returns. Golec (1996) shows in his paper The Effect of Mutual Fund Managers Characteristics on Their Portfolio Performance, Risk and Fees that a fund s performance, risk and fees are significantly impacted by a manager s characteristics. They document that investors can expect better risk-adjusted returns from younger managers with an MBA degree that have longer tenure at their funds. Furthermore, they examined that a large management fee signals superior investment skill which leads to better performance. II.II Hypotheses Concluding, this research will shed light on the question if these results are gathered by the manager of the fund or the fund itself. In comparison with, for instance, Chevalier and Ellison (1999a) this research will be an addition to the existence literature because it goes into more depth regarding to the differences in education. Besides higher education (having an MBA 9

degree or not), this study will focuses as well on broader education. This allows me to formulate the following hypothesis: H1 Mutual fund managers who attended higher and more education tend to earn higher excess returns. According to Bertrand and Schoar (2003), realizations of all investment, financing and other organizational strategy variables appear to systematically depend on the specific executives in charge. Therefore I want to measure the influence of managerial characteristics on fund characteristics. This enables me to formulate the next hypothesis: H2 Better educated mutual fund managers tend to manage bigger firms which have more Investment Inflow. Gottesman and Morey (2006) argue that following former studies students that attend liberal arts schools might have had more individual attention than students that went to larger, research-oriented institutions. Therefore, they include this independent variable in their study, to examine the influence of this difference. In addition to this, I would like to examine the influence of a having a different degree. This encourages me to formulate the following hypothesis: H3 Mutual fund managers that hold a degree in finance or beta related studies tend to outperform degrees that are indirect related to the mutual fund industry. H4 Education has an influence on the investment style of mutual fund managers. According to Chevalier (1999b), younger managers tend to hold less idiosyncratic risks and have more conventional portfolios. In this study, we would like to examine this further with the additional educational variables. Therefore, I have developed the following hypothesis: 10

H5 Managerial characteristics influence the extent to which mutual fund managers hold on systematic risk. II. Data and methodology The majority of the data that is used in this thesis is gathered via intensive online 2 data searching. In collaboration with four others students, we have created from a raw dataset of 27,707 mutual fund managers a unique file of 2,807 mutual fund managers with accompanying mutual funds. From this starting point we have searched for all kind of characteristics of the manager, with a strong focus on education, address details, (ex)-industry and further personal characteristics. From that point on, everyone worked independently on their thesis. As a matter of relevance, this study uses only the characteristics that are related to education with some additional control variables and therefore the following exogenous variables are included: III.I Exogenous variables Age: This variable is denoted in date of birth and is founded in 69% of the cases. In the case that it could not be found, an assumption is made with the tenure variable by subtracting 21 years from this variable. Tenure: This variable indicates the year in which they entered the industry of investing. This variable is founded in 81% of the cases. If it could not be found explicitly on the internet, the year when the degree is granted is denoted. In the case that it could not be 2 Examples of sources that frequently used are: Forbes, LinkedIn, Business Week, Edgar online (SEC filings), ZoomInfo, Morningstar, Intellius, mutual fund websites, The Economist, Yahoo Finance, Reuters, Bloomberg, CRSP Mutual, The Wallstreet Transcript. 11

found at all, an assumption is made with the age variable by adding 21 years up to this variable. Gender: A zero is listed when it concerned a man, while a 1 is listed when it concerned a female. This variable is founded in 100% of the cases. Bachelor institution: the name of the institution where the manager has received his bachelor is listed. This variable is founded in 84% of the cases. When the variable could not be found, an assumption is made that the information is not publicity available instead of the manager has no degree. In other words, I assume that all mutual fund managers have at least a bachelor degree. of the cases. Bachelor subject: the subject of the bachelor is listed. This variable is founded in 54% 2 nd bachelor, MSc, MA, MBA, CFA, CPA and PH.D.: A zero is listed when the concerning managers does not have one of these titles, while a 1 is listed when he/she has. These variables are founded respectively in 3%, 8%, 6%, 52%, 48%, 4% and 4% of the cases. For the name of the bachelor institution, I have added two additional proxies, since this variable consist of 632 unique universities: SAT scores: this variable might reflect the manager s ability effort, connection and/or the quality of his degree. The SAT 3 is a standardized test widely used for college admissions in the United States. This test consists of three sections (Mathematics, Writing and Critical Reading) where a score of 200-800 can be obtained on each of the three sections (total 600-2400). Most universities report upper and lower bounds for all of the three sections, while some schools only report these bounds for the writing and mathematical section. 3 Data sources: http://collegeapps.about.com/od/collegeprofiles/p/ and http://www.satscores.us/ 12

The bounds are supposed to be constructed so that the middle of 50% of students attending the school lies between the upper and lower bounds. For instance, Harvard University and Princeton University report the following scores: Harvard University Princeton University SAT Math: 710/800 SAT Math: 690/800 SAT Critical reading: 700/800 SAT Critical reading: 680/800 SAT Writing: 710/800 For Harvard, the average of the bounds per section is taken, which results in a SAT score of 2260 for. For Princeton, the average of the bounds is taken, which results in 1485. In addition to set things equal, we have multiplied this score by 1.5 to get an estimator of the SAT score when the writing section was not missing, which results in a SAT score of 2227.5. This variable is founded in 92% of the cases. In 7% of the cases, the university is located outside the US and the university therefore didn t report a SAT score. The remaining 1% could not be found. Admission rate%: if publicity available, the admission rate% for all universities is listed. This variable is founded in 92% of the cases. III.II CRSP Mutual Fund Information As stated before, after removing all duplicates and keeping only unique mutual fund managers, 2,807 managers are left. The data file, containing these managers, is merged with a CRSP Mutual Fund Database 4 which contains information about mutual funds descriptive 4 Survivorship bias free. 13

information, monthly mutual fund returns and Fama-French factors for a chosen period of 1980 5 to 2013. III.III Endogenous variables In order to clarify the cross-sectional relationship between the managerial characteristics and mutual fund returns, I have calculated several measures which will be used as endogenous variables in this study. Simple excess return of the mutual fund is an endogenous variable that is calculated by subtracting the risk-free return (One Month Treasury Bill Rate) from the total return at the end of the month. Risk-adjusted return of the mutual fund is calculated by determining the alpha 6 that is generated by the manager in charge. Alpha is Jensen s measure to determine performance adjusted for systematic risk. It measures the portfolio return that is attributable to the manager s skill (or luck). In this study, alpha is calculated in three ways. First, alpha is calculated by using the Capital Asset Pricing Model 7 with the One Month Treasury Bill rate as risk-free rate and a value-weighed market index as the market portfolio. In addition to systematic risk, market capitalization, the book-to-market-ratio and past returns are taken into account since they have significant influence on explaining cross-sectional patterns in stock returns regarding the modern finance literature. The High minus Low (HML) portfolio is a zero-investment portfolio constructed by subtracting the returns of low book-to-market ratio stocks from the returns of high book-to-market ratio stocks. The Small minus Big (SMB) portfolio is a zero-investment portfolio constructed by subtracting the returns of large market capitalization firms from the stock returns of small market capitalization firms. These two 5 Since I capture with this time window all the data that is needed for my sample. 6 See Jensen (1968) 7 Equation to determine Alpha : (1) 14

additional portfolio s leave us the alpha according to Fama-French (1992) 8. Finally, the momentum (MOM) is added to calculate the alpha according to Carhart (1997) 9, which consists of a zero-investment portfolio constructed as the spread between the performance of stocks that are in the top 30 percent of returns in the 12 prior months and those that are in the bottom 30%. mutual fund. Investment inflows determines the total investment inflows by investors in (or out) the (4) Systematic risk is measured by beta. Systematic risk is the type of risk that is impossible to completely avoid. It cannot be mitigated through diversification. It is the risk that is inherent to the entire market segment. Idiosyncratic risk is measured by the residual variation in portfolio return after accounting for variation due to systematic risk. It measures the degree of portfolio diversification. Total Net Assets is measured by taking the logarithm of the Total Net Assets. It is included as a control variable for size. 8 (2) 9 (3) 15

III.IV Summary statistics and Correlation Matrix Table I lists the summary statistics for all the variables. We observe that the average beta is less than one (0.88) and that all the alphas are on average negative, which implies that the average mutual fund manager underperforms the market. Furthermore, we see that the average mutual fund manager is 57 year old, works for 32 years in the industry, holds an MBA (69%) and CFA designation (52%) and is a man (91%). Table II presents the correlation between the variables. The results indicate a strong negative correlation between SAT scores and Admission rate% of -0.89, which confirms according to ones expectations that higher SAT universities do allow less applicants. Furthermore, we see a correlation coefficient of -0.04 between CFA and MBA, which means that having an MBA does according to my results not directly encourages to become a CFA. The correlation between having an MBA degree and SAT scores are positive (0.13). This implies that managers from higher educational prestige schools are more likely to become an MBA than managers from lower prestige schools. III. Empirical results My goal in this section is to explain whether mutual fund managers can influence the crosssectional distribution in mutual fund returns. IV.I Simple Excess return OLS Table III shows the first linear regression analysis where the simple excess return on a share is regressed on a set of manager characteristics that are representative for the average mutual fund manager. The dependent variable is the simple excess return of the mutual Fund, 16

which is regressed on different managerial characteristics, including the average SAT score of the managers undergraduate institution (divided by 100), a dummy variable that takes the value one if the manager has an MBA or CFA and zero otherwise, the manager s age and the managers tenure in investment years. 10 T-values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. The point estimates suggests that managers who have an MBA and CFA degree and who attended a higher SAT school earn on average higher returns, but none of the coefficients are statistically significant. This also applies to the age and tenure of the mutual fund manager. Furthermore, an increase in the admission rate% will lead on average to a decrease in the simple excess return of the mutual fund, which means that a manager who has attended a undergraduate institution which is easier to apply for, the simple excess return of the mutual fund will decrease. Again, the coefficient is not statistically significant. If we look besides statistical significance, to economic significance, we see quite similar results. has a mean of 0.0049 and, for instance, has a standard deviation of 2.1879. If we look to the estimator β of we see a coefficient of 0.0002. A one standard deviation increase in, i.e. = 0.0002, will result in a change in the dependent variable of the average of the of 0.0002 2.1911 = 0.0004. Relative to, this is a 0.082% increase. 10 Regression equation: (5) 17

IV.II Mutual Fund Characteristics and Fund Performance Table IV examines whether mutual fund manager characteristics are correlated with different characteristics of the mutual fund itself. In table IV the outcome of these regressions are tabulated and it shows in the first place that manages that attend a higher SAT undergraduate institution tend to manage larger funds which have more investment inflow. According to Chevalier and Ellison (1999a), I do find contrary results. They document that age and firm size are negatively significant related, while tenure and firm size are positively significant related. The impact of the manager s age on firm size is slightly negative and statistically significant at the 1% level, but the economic significance seems to be very low since the coefficient is almost zero. Furthermore, the relationship between tenure and firm size is statistically negatively related to firm size. Managers with more investment experience seem to manage smaller funds. Regarding the relationship between MBA and firm size, I do find similar results to that of Gottesman and Morey (2006), since larger firms tend to hire non- MBA managers. This coefficient is statistically significant at the 1% level. Finally, we see a strong negative relationship between gender and both endogenous variables. Female managers are less likely to manage a large mutual fund and do have less investment inflow. Both coefficients are highly statistically significant. IV.III Risk Adjusted Performance Measures Besides the simple excess return as a dependent variable, I want to look to a risk-adjusted performance measure to determine more accurate the relationship between manager characteristics and mutual fund performance. Table V includes the CAPM alpha 11 as dependent variable. First, in row 1, the same regression equation is presented as in table III for comparison. In row 2, CAPM alpha is regressed on the managerial characteristics. We observe that all the variables except for tenure move from non-significant to highly significant 11 See equation (3). 18

at the 1% level. The causal impact is similar to that of row 1. Managers that hold an MBA degree tend to generate a positive alpha, all else equal. Row 3 and 4 show an OLS where all the exogenous variables used in this study are included and log of assets and investment inflow are added as control variables. The added variables do not seem to explain the variation in the dependent variables, since almost all coefficients are quite similar to zero. IV.IV Sharpe (1964), Fama-French (1992) and Carhart (1997) Table VI compares the regression equations for the three different alphas. Besides the alpha according to Sharpe (1964), the other measures of alpha are included 12. The Fama-French factors tend to explain the most variability of the data around its mean, since this OLS regression has the highest R-square (0.06). Furthermore, the coefficients are quite similar to each other. Each row indicates that a manager that holds an MBA seems to outperform a non- MBA manager based on the dependent variable alpha. One possible simple explanation for this could be that managers that hold an MBA are more intelligent than other managers. Another possible explanation is, that these managers do have better connections due to their education and are therefore exposed to more information than other non-mba managers. IV.V Education specific performance measures Table VII examines more education specific performance measures of mutual funds. The bachelor subject of the manager s undergraduate institution is taken into account and the simple excess return and CAPM alpha are regressed on these characteristics. One might expect that mutual fund managers that hold a degree in Finance would be better informed regarding investments than a manager that holds a degree in History. What we see, for instance, is that a manager who holds a bachelor degree in Social Sciences seems to 12 See equation (1), (2) and (3) 19

underperform a manager who holds a bachelor degree in Economics/Finance/Business/Accounting/Marketing. The coefficient is significant at the 1% level. Though, the economic significance seems not to be in place on the other hand. Generally speaking, we do not observe a strong relationship between this education variable and fund performance, since most variables are not statistical or economic significant. IV.VI Fama-French Factors OLS Table VIII shows the regression coefficients of the Fama-French factor weights on the managerial characteristics. As mentioned before, recent finance literature has described that besides market risk, three other portfolios might determine the return on investment. Fama- French (1992) explored that the stocks of small firms have consistently outperformed the stocks of large firms. They created a portfolio where stocks of large firms are sold and stock of small firms are bought, which we refer to as the small minus big portfolio. They stated also that the shares of firms with a high book to value of assets divided by market value of assets outperform the market portfolio. They refer to this portfolio as the high minus low portfolio, where shares of high book-to-market stocks are bought and of low book-to-market stocks are sold. Carhart (1997) adds a fourth portfolio, where stocks of last year s outperformers are bought and stocks of last year s underperformers are sold, which we refer to as the momentum. The table tells us that managers that attend an undergraduate institution with lower requirements for admission, hence a higher admission rate, tend to load more on small minus big strategies. This coefficient is statistical significant at the 1% level. The economic significance is also very high, since has a mean of 0.00444851 and has a standard deviation of 0.2373. The latter variable reflects a coefficient of 0.0032. A one standard deviation increase in, i.e. 20

= 0.0032, will result in a change in the dependent variable of of 0.032 0.2373 = 0.0076. Relative to the average Small minus Bi, this is a 170% increase. Furthermore, age and tenure seem to have a minor impact on different strategies. Where other studies like Chevalier (1999a) report that older managers tend to use more momentum strategies, does it not seem to matter for this study. VI.VII Systematic and Idiosyncratic risk Table XI measures the impact of managerial characteristics on holding on different kind of risks. We would expect that MBA is positively related to systematic risk and negatively related to idiosyncratic risk. MBA s are taught that only beta receives compensation in the market and therefore they will try to outperform the market by only taking one more systematic risk rather than idiosyncratic risk. However, the table shows the opposite of one s expectations. MBA is positively related to beta, but the coefficient is not statistically significant. If we look to the relationship between residual standard deviation (idiosyncratic risk) and MBA we see again a positive (and now significant at the 1% level) relationship. This implies that managers that hold an MBA degree do take on more idiosyncratic or idiosyncratic risk to outperform the market. On the other, this does not count for managers that hold a CFA designation. They tend to hold on more systematic risk (not significant) and less idiosyncratic risk. 21

IV.VIII Endogeneity: Discussion I tried to examine whether managerial characteristics are related to mutual fund performance. I used several educational variables to explain the variation in the dependent variables. Even then, my findings may be exposed to endogeneity issues. First of all, it could face reverse causality problems. I tried to examine whether education causes alpha. Hence, I want to measure whether gained knowledge or connections due to education of the mutual fund manager influences the performance of mutual funds. Since it seems to be very unlikely that managers leave the industry to get more education, I do not expect reverse causality problems. Furthermore, my results could be exposed to the omitted variable bias. It is hard to distinguish smarts from education. Smart managers could also have had an easier time acquiring an education. Hence, there could be an omitted variable Smarts - that influences both education and alpha and there might be no relationship between education and alpha, but only smarts is what driving alpha. To address issues like this, I have added several control variables like age, tenure and firm size. However, there could be other omitted variables that explain the variation in alpha, like place of birth, residence and other personal characteristics. 22

IV. Conclusions This study examines the relationship between managerial characteristics and mutual fund performance. It shed light on the question whether the quality and difficulty of the obtained degrees of the manager concerned do influences the performance of mutual funds. Similar to Chevalier and Ellison (1999a), I examine if the difficulty of the undergraduate institution that the manager in charge has completed measured by SAT scores - has an impact on the performance of the fund as well as if the manager holds an MBA or not. In addition to this, this study measures the influence of holding a CFA/CPA designation, having an MSc/MA/2 nd bachelor degree, being a Ph.D. as well as the influence of the subject of the bachelor degree that is obtained. Furthermore, it contains besides the control variables age and tenure, the fund size and the total investment inflow within the mutual fund. Besides the relationship with fund performance, it measures whether managerial characteristics are related to fund characteristics, investment style and taking on specific kinds of risk. First, we observe that the SAT scores of the manager s undergraduate institution and the presence of an MBA and CFA designation is positively related to the simple excess return of the mutual fund. Although, none of the coefficients is statistically or economic significant. Second, I document similar results to that of Gottesman and Morey (2006) regarding the relationship between fund size and having an MBA, since larger firms tend to hire non- MBA managers. This coefficient is statistically significant at the 1% level. Furthermore, I find a strong negative relationship between gender and firm size as well investment inflow. Female managers are less likely to manage a large mutual fund and do have less investment inflow. Both coefficients are highly statistically significant. 23

Third, if I correct systematic risk and look to the relationship between risk-adjusted performance and managerial characteristics we observe again that SAT scores, holding a CFA designation and having an MBA is positively and this time significant related to alpha. Almost all coefficients are highly statistically significant. Nevertheless, the economic significance seems to be small. Fourth, we observe that managers that attend an undergraduate institution with lower requirements for admission, hence a higher admission rate, tend to load more on small minus big strategies. This coefficient is statistical significant at the 1% level. The economic significance is also very high. Furthermore, age and tenure seem to have a minor impact on different strategies. Where other studies like Chevalier and Ellison (1999a) report that older managers tend to use more momentum strategies, does it not seem to matter for this study. Fifth, we find that managers that hold an MBA degree do take on more idiosyncratic risk to outperform the market. On the other hand, this does not count for managers that hold a CFA designation. They tend to hold on more systematic risk (not significant) and less idiosyncratic risk. 24

V. References Articles Bertrand, M., and A. Schoar, 2003, Managing with Style: The Effect of Managers on Firm Policies, Quarterly Journal of Economics 118(4), 1169-1208. Chevalier, J., and G. Ellison, 1999a, Are Some Mutual Fund Managers Better Than Others? Cross-Sectional Patterns in Behavior and Performance, Journal of Finance 54(3), 875-899. Carhart, Mark, 1997. On perstistence in mutual fund performance. Journal of Finance 52 (1), 57-82 Chevalier, J., and G. Ellison, 1999b, Career Concerns of Mutual Fund Managers, Quarterly Journal of Economics 114(2), 389-432. Cohen, L., A. Frazzini, and C. Malloy, 2010, Sell-Side School Ties, Journal of Finance 65(4), 1409-1437. Ding, B., and R. Wermers, 2009, Mutual Fund Performance and Governance Structure: The Role of Portfolio Managers and Boards of Directors, Working paper. Fama, E. F., and K. R. French, 2010, Luck versus Skill in the Cross-Section of Mutual Fund Returns, Journal of Finance 65(5), 1915-1947. Golec, Joseph H., 1996. The effects of mutual fund managers characteristics on their portfolio performance, risk and fees. Financial Services Review 5, 133-148. Kacperczyk, M., C. Sialm, and L. Zheng, 2005, On the Industry Concentration of Actively Managed Equity Mutual Funds, Journal of Finance 60(4), 1983-2011. Books Malkiel, B. G., 1973, A Random Walk Down Wall Street: The Time-Tested Strategy for Successful Investing, W. W. Norton and Co. 25

VI. Appendix Table I Descriptive Statistics Summary statistics for all of the variables used in the analysis are presented. Simple excess return (%) is calculated by subtracting the risk-free return rate (One Month Treasury Bill Rate) from the total return at the end of the month. Beta is the coefficient of the market portfolio, determined by regressing the fund s monthly returns minus the risk-free rate on the monthly returns of the market portfolio minus the risk-free rate. The HML, SMB, and MOM weights are the coefficient from a regression of the fund s monthly returns on the market returns minus the risk-free rate and the returns of subtracting the returns of low book-to-market ratio stocks from the returns of high book-to-market ratio stocks. The SMB portfolio is a zero-investment portfolio constructed by subtracting the returns of large market capitalization firms from the stock returns of small market capitalization firms. The MOM consists of a zero-investment portfolio constructed as the spread between the performance of stocks that are in the top 30 percent of returns in the 12 prior months and those that are in the bottom 30%. The HML portfolio is a zero-investment portfolio constructed by buying high book-to-market stocks and selling low book-to-market stocks. Alpha4 is the excess return from this four-factor model in percent per year. The log of assets is the logarithm of the monthly asset value. The manager characteristics variables includes the SAT score of undergraduate institutions which are conveniently divided by 100, the admission rate in percent of the undergraduate institution, dummy variables that takes the value of one if the manager has an MBA, Ph.D., CFA, CPA, MSc, MA or 2 nd bachelor and zero otherwise, the managers age and the manager s tenure in investment year. Variable # of Obs. Mean Std. Dev. Min. Max. Excess return 137396 0.0049 0.0493-0.1598 0.1404 Beta 137379 0.8834 0.3503-4.5790 4.3288 Unsys. risk 137379 0.0676 0.0553 0.0000 3.1174 Alpha1 137379-0.0002 0.0045-0.0708 0.0734 Alpha3 137379-0.0008 0.0042-0.1700 0.0824 Alpha4 137379-0.0008 0.0041-0.1488 0.1350 SMB 137379 0.2133 0.3315-4.9595 2.7710 HML 137379 0.0803 0.3406-6.7372 2.7221 MOM 137379 0.0013 0.1652-5.7772 2.1292 Manager SAT (/100) 137396 19.0389 2.1879 13.7000 22.3500 Admission rate% 137396 0.4276 0.2361 0.7200 0.9300 MBA 137396 0.6934 0.4611 0 1 PH.D. 137396 0.2653 0.4415 0 1 MA 137396 0.2854 0.4516 0 1 MSc 137396 0.3002 0.4583 0 1 CPA 137396 0.0436 0.2041 0 1 CFA 137396 0.5236 0.4994 0 1 2nd bachelor 137396 0.0441 0.2052 0 1 Cash-in-flow 137395 16.9132 2.2496 2.2499 75.4944 Log of Assets 137396 4.2258 2.4121-2.3026 9.4468 Age 137396 1956 11.2260 1926 1978 Tenure 137396 1981 10.0485 1955 2000 Gender 137343 0.0872 0.2821 0 1 26

Table II Correlation Matrix This table presents the correlation between the different variables. Variable name [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] Excess return [1] 1.000 Beta [2] 0.016 1.000 Unsys. risk [3] -0.002 0.267 1.000 Alpha1 [4] 0.064-0.173-0.166 1.000 Alpha3 [5] 0.063-0.050-0.134 0.790 1.000 Alpha4 [6] 0.059-0.103-0.146 0.753 0.908 1.000 SMB [7] 0.021 0.318 0.335 0.003-0.062-0.093 1.000 HML [8] 0.007-0.401-0.086 0.218-0.101-0.038 0.005 1.000 MOM [9] 0.013 0.161 0.159 0.104 0.158-0.134 0.114-0.226 1.000 Manager SAT (/100) [10] 0.015 0.056 0.041 0.104 0.111 0.098 0.056-0.003 0.018 1.000 Admission rate% [11] -0.011-0.048-0.045-0.110-0.104-0.091-0.062-0.010-0.023-0.829 1.000 MBA [12] 0.002 0.012 0.037 0.039 0.046 0.025 0.016-0.032 0.063 0.133-0.137 1.000 PH.D. [13] 0.002 0.019 0.026 0.001 0.010-0.005-0.036-0.017 0.022 0.052-0.054 0.305 1.000 MA [14] 0.002 0.025 0.009 0.018 0.026 0.015-0.063 0.006 0.015 0.017-0.014 0.264 0.806 1.000 MSc [15] 0.002 0.056 0.018 0.007 0.015-0.003-0.022-0.021 0.025 0.043-0.052 0.208 0.796 0.746 1.000 CPA [16] 0.000-0.038-0.023 0.041 0.025 0.022 0.001 0.010 0.007-0.066 0.053 0.002-0.025-0.051-0.046 1.000 CFA [17] 0.001 0.006-0.018 0.014 0.007 0.005 0.014 0.002 0.006 0.002 0.040 0.093-0.041-0.036-0.015-0.001 1.000 2nd bachelor [18] 0.004 0.038 0.012 0.032 0.049 0.043-0.015-0.045 0.028 0.005-0.007-0.050-0.007-0.028-0.031 0.026-0.057 1.000 Investment Inflow [19] 0.058 0.164-0.069 0.152 0.170 0.156-0.021-0.009 0.047 0.113-0.089 0.008 0.010 0.041 0.046 0.000 0.015-0.013 1.000 Log of Assets [20] 0.022-0.001-0.158 0.171 0.199 0.200-0.058 0.007-0.018 0.147-0.139 0.012 0.029 0.070 0.023-0.013 0.008 0.009 0.384 1.000 Age [21] -0.005 0.074 0.077 0.028 0.038 0.033 0.044-0.113-0.004-0.059 0.062 0.026-0.057-0.032-0.012 0.036 0.088 0.029-0.044-0.070 1.000 Tenure [22] -0.006 0.095 0.087 0.023 0.031 0.025 0.035-0.136 0.000-0.058 0.046 0.016-0.019-0.006 0.018 0.026 0.069 0.010-0.058-0.074 0.828 1.000 Gender [23] -0.004-0.042 0.034-0.020-0.025-0.028-0.004 0.000 0.044-0.005-0.014 0.033 0.063 0.047 0.029 0.000 0.029 0.044-0.034-0.040 0.045 0.036 1.000

Table III Mutual fund Performance and Manager Characteristics The dependent variable is the simple excess return of the mutual fund at the end of the month which is regressed on a set of manager characteristics, including the average SAT score of the managers undergraduate institution (divided by 100), a dummy variable that takes the value one if the manager has an MBA and zero otherwise, the manager s age and the managers tenure in investment years. T- values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. Independent Variables Coefficients Constant 0.051004 (-1.48) Manager SAT (/100) 0.000162 (-0.96) Admission rate% -0.000974 (-0.34) MBA 0.000099 (-0.36 CFA 0.000096 (-0.09) Age 0.000002 (-1.13) Tenure -0.000027 (-1.35) Gender -0.006360 (-1.91) R 2 0.0002 No. of observations 137343 28

Table IV Fund Characteristics and Manager Characteristics The dependent variable are the log of total assets and the investment inflow, which are regressed on a set of manager characteristics, including the average SAT score of the managers undergraduate institution (divided by 100), a dummy variable that takes the value one if the manager has an MBA and zero otherwise, the manager s age, the managers gender (M/V), a dummy variable that takes the value one if the manager has an MBA/CFA/CPA/Ph.D./MA/MS/2 nd bachelor and zero otherwise and the managers tenure in investment years. T-values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. Independent Variables Log of Assets (1) Dependent Variables Investment inflow (2) Constant 33.8190 131.682 (26.35)*** (20.36)*** Manager SAT (/100) 0.1030 0.670 (19.65)*** (25.36)*** Admission rate% -0.6801 0.517 (-13.97)*** (2.11)** MBA -0.1120-0.224 (-7.54)*** (-2.99)*** CFA 0.1082 0.439 (8.33)*** (6.71)*** CPA 0.0299 0.800 (-0.95) (5.05)*** PH.D. -0.3675-3.341 (-12.78)*** (-23.06)*** MA 0.8841 2.186 (35.24)*** (1703)*** MSc -0.2500-0.437 (-10.32)*** (17.91)*** 2nd bachelor 0.1670 0.015 (5.34)*** (-2.77)*** Age -0.0042-0.080 (-4.07)*** (2.95)*** Tenure -0.0117-0.080 (-10.3)*** (-13.92)*** Gender -0.3442-1.264 (-15.12)*** (-11.03)*** R 2 0.0300 0.0183 No. of observations 137343 137342 29

Table V Risk Adjusted Performance Measures The dependent variables are the simple excess return of the mutual fund at the end of the month and the CAPM alpha which are regressed on a set of manager characteristics, including the average SAT score of the managers undergraduate institution (divided by 100), a dummy variable that takes the value one if the manager has an MBA/CFA/CPA/Ph.D./MA/MS/2 nd bachelor and zero otherwise, the manager s age, the managers tenure in investment years and the log of total assets of the mutual fund at the end of the month. T-values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. Independent Variables Simple excess (1) Alpha (2) Dependent Variables Simple excess/full (3) Alpha/full (4) Constant 0,0510-0,0287 0,0216-0,0392 (1.91)* (-11.82)*** (-0.81) (-16.32)*** Manager SAT (/100) 0,0002 0,0001 0,0000 0,0000 (1.48) (-8.21)*** (-0.05) (-4.15)*** Admission rate% -0,0010-0,001478-0,0011-0,0014 (-0.96) (-16)*** (-1.11) (-15.06)*** MBA 0,0001 0,0002 0,0001 0,0003 (0.34) (-8.03)*** (-0.4) (-9.74)*** CFA 0,0001 0,0001 0,0000 0,0001 (0.36) (-4.61)*** (-0.1) (-3.76)*** CPA 0,0000 0,0000 0,0010 (-7.1)*** (-0.01) (-17.53)*** PHD 0,0005-0,0039 (-0.89) (-7.36)*** MA -0,0009 0,0003 (-0.16) (-7.35)*** MSc -0,0005 0,0000 (-0.9) (-0.37) 2nd Bachelor 0,0011 0,0007 (-1.7)* (-12.23)*** Age 0,0000 0,0000 0,0000 0,0000 (0.09) (-7.1)*** (-0.13) (-5.99)*** Tenure 0,0000 0,0000 0,0000 0,0000 (-1.13) (-0.16) (-0.32) (-3.5)*** Gender -0,0064-0,0037-0,0004-0,0003 (-1.35) (-8.68)*** (-0.81) (-6.12)*** Log of Assets 0,0000 0,0002 (-0.41) (-42.17)*** Investment Inflow 0,0002 0,0000 (-19.78)*** (-33.92)*** R 2 0,0002 0,0150 0,0034 0,0516 No. of observations 137343 137326 137342 137325 30

Table VI Alpha and Managerial Characteristics The dependent variables are the alphas from a regression analysis, where the excess return per share is regressed on respectively one (excess return on the market), three (the latter variable, HML and SMB) and four (the latter variables and MOM) portfolio s. These alpha s are regressed on a set of manager characteristics, including the average SAT score of the managers undergraduate institution (divided by 100) and the admission rate, a dummy variable that takes the value one if the manager has an MBA/CFA/CPA/Ph.D./MA/MS/2 nd bachelor and zero otherwise, the manager s age, the managers tenure in investment years, the log of total assets of the mutual fund at the end of the month and the investment inflow. T-values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. Independent Variables CAPM (1) Dependent Variables Fama-French (2) Carhart (3) Constant -0.0392-0.0480-0.0423 (-16.32)*** (-21.84)*** (-19.37)*** Manager SAT (/100) 0.0000 0.0001 0.0001 (4.15)*** (11.25)*** (9.92)*** Admission rate% -0.0014-0.0006-0.0005 (-15.06)*** (-6.61)*** (-5.57)*** MBA 0.0003 0.0003 0.0002 (9.74)*** (12.36)*** (6.31)*** CFA 0.0001 0.0000 0.0000 (3.76)*** (-0.37) (-0.05) CPA 0.0010 0.0006 0.0005 (17.53)*** (11.19)*** (10.01)*** PH.D. -0.0004-0.0003-0.0003 (-7.36)*** (-5.91)*** (-6.26)*** MA 0.0003 0.0003 0.0004 (7.35)*** (6.48)*** (8.26)*** MSc 0.0000 0.0000-0.0002 (-0.37) (0.36)*** (-3.65)*** 2nd Bachelor 0.0007 0.0010 0.0009 (12.23)*** (18.73)*** (16.08)*** Age 0.0000 0.0000 0.0000 (5.99)*** (9.95)*** (8.48)*** Tenure 0.0000 0.0000 0.0000 (3.5)*** (3.39)*** (2.50)** Gender -0.0003-0.0003-0.0003 (-6.12)*** (-8.17)*** (-8.56)*** Log of Assets 0.0002 0.0003 0.0000 (42.17)*** (51.62)*** (31.51)*** Investment Inflow 0.0000 0.0000 0.0003 (33.92)*** (37.14)* (53.33)*** R 2 0.0516 0.0643 0.0581 No. of observations 137325 137325 137325 31

Table VII Education specific performance measures The dependent variables are the simple excess return of the mutual fund at the end of the month and the CAPM alpha which are regressed on a set of manager characteristics, including the average SAT score of the managers undergraduate institution (divided by 100), a dummy variable that takes the value one if the manager has an MBA/CFA/CPA/Ph.D./MA/MS/2 nd bachelor and zero otherwise, the manager s age, the managers tenure in investment years and the log of total assets of the mutual fund at the end of the month. Furthermore, dummy variables are included which take the value on if the subject of the manager at the undergraduate institution was Languages, Law, Mathematics/Econometrics/Chemistry/Physics, Others or Social Sciences and zero if it is an Economics/Finance/Business/Accounting/Marketing subject. T-values are in parentheses and the symbols *, ** and *** denote statistical significance at the 10%, 5% and 1% levels. Independent Variables Simple excess (1) Dependent Variables Alpha (2) Simple excess/full (3) Alpha/full (4) Constant 0.0386-0.0286 0.0091-0.0417 (-1.09) (-10.21)*** (-0.26) (-15.16)*** Manager SAT (/100) 0.0001 0.0001 0.0000 0.0001 (-0.76) (10.21)*** (-0.25) (7.87)*** Admission rate% -0.0013-0.0002-0.0005 0.0001 (-1.02) (-1.60)*** (-0.37) (0.53)) MBA 0.0001 0.0001 0.0005 0.0002 (-0.29) (4.72)*** (-1.26) (7.77)*** CFA 0.0000-0.0007 0.0000-0.0001 (-0.08) (2.51)** (-0.10) (-4.25)*** CPA -0.0003 0.0003 (-0.39) (4.10)*** PH.D. 0.0005-0.0004 (-0.75) (-7.05)*** MA -0.0008 0.0004 (-1.17) (6.88)*** MSc 0.0000-0.0001 (-0.08) (-3.00)*** 2nd Bachelor 0.0013 0.0006 (-1.75)* (10.74)*** Age 0.0000 0.0000 0.0000 0.0000 (-0.62) (10.71)*** (-0.82) (9.66)*** Tenure 0.0000 0.0000 0.0000 0.0000 (-0.06) (-3.31)*** (-0.57) (-0.29) Gender -0.0002-0.0002-0.0004-0.0001 (-0.36) (-4.12)*** (-0.59) (-1.17) Log of Assets -0.0001 0.0003 (-1.05) (45.85)*** Investment Inflow 0.0003 0.0000 (16.78)*** (21.21)*** Languages -0.0002-0.0001 0.0004 0.0002 32