In the US, are mutual funds able to beat the market?
- Looking back at US mutual fund performance from January 2006 till December 2015.ABSTRACT
In this paper we analyze the performance of mutual funds versus the performance of the market, to see if mutual funds are able to beat the market. To broaden our research, we also investigate if mutual fund size has an effect on mutual fund performance. Based on the Lipper Classification Codes, we divide our mutual fund database into three subcategories, Small-, Mid-, and Large-Cap. In order to measure performance, we use the Carhart (1997) four-factor model. We find no evidence of superior performance of mutual funds over the performance of the market.
Name: Paul Gravestein
Student number: 10440186
Study: Economics and Business Track: Finance and Organization Specialization: Finance
Statement of Originality
This document is written by Student Paul Gravestein who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents
3
Table of contents
I. Introduction p. 4
II. literature review p. 5
- Mutual funds p. 6
- Previous research findings p. 6
- Size effect p. 9
III. Data p. 9
IV. Methodology p. 10
- Hypothesis p. 11
V. Empirical results p. 12
- Total sample results p. 12
- Small-, mid-, large-cap results p. 13
VI. Conclusion p. 14
VII. Discussion p. 15
4
I. Introduction
In 1973, Burton Malkiel stated that a monkey could do a just as good job at selecting a portfolio of stocks, by throwing darts at newspaper’s financial pages whilst being blindfolded, as opposed to an expert carefully selecting a portfolio. Burton Malkiel is a professor of economics at Princeton University and wrote a book, ‘A Random Walk Down Wall Street’, about his statement. If he’s right, this means that financial experts and thus mutual funds are not able to outperform the market. This leads us to the following question: why pay ‘experts’ to select a portfolio on behalf of you, if you could do just as good or even better if you would just randomly select a portfolio yourself?
Over the past few decades mutual funds have become more and more popular among investors (p.784, Gruber, 1996). From 1976 till 2009, the assets under management by mutual funds increased from 51$ billion to 11,1$ trillion (p. 81, Khorana and Servaes, 2012). According to ICI (Investment Company Institute), in 2011 nearly a quarter of US households’ investments were managed by an investment company of some sort. Despite the financial crisis, the total number of US mutual funds grew to nearly 8.000 in 2013, which held 13,5$ trillion assets under management (p. 96, Bodie et al., 2014).
As reported by Gruber (1996), one possible reason for the popularity of mutual funds is that they are priced and sold at net asset value. Since mutual funds are so called ‘open-end’ funds, they have to be priced at net asset value. If this wouldn’t be the case, there would be an arbitrage opportunity. Because mutual funds are traded at net asset value, management ability may not be priced. This would be the case if prices weren’t priced at net asset value. In the latter case the ‘price’ you would have to pay for management ability is the amount the price deviates from net asset value.
Likewise, Malkiel (2013) found that overconfidence by institutional investors is likely to play an important role in the increased popularity of mutual funds. He states that many investors truly believe that mutual fund managers can select the best investment
opportunities, even if empirical research shows otherwise. Furthermore, he states that the prices in the form of fees are of influence too. This is because many investors judge the effectiveness and quality of a mutual fund based on the price charged for the service. As a result, investors value mutual funds with the highest fees the most.
5
Past research on the performance of mutual funds is inconclusive. For instance, Wermers (2000) finds that mutual funds are able to beat the market. Whilst on the hand, Fama and French (2010) among others find that mutual funds generally underperform the market. This leaves us with a so called ‘puzzle’ about who’s right, and thus has led us to formulate the following research question:
In the US, are mutual funds able to beat the market?
In this study we focus on the performance of US mutual funds over a ten-year period, from January 2006 till December 2015. In order to get the average return of mutual funds we first make an index consisting of 7.483 US mutual funds returns. Each of those mutual funds returns is measured monthly. To measure if mutual funds outperform the market we use Carhart’s (1997) four-factor model. Our main point of focus within this model is the alpha, α, of the mutual fund index. The alpha is our measure of risk adjusted performance of mutual funds. If the alpha, looking at the robust t-statistic, deviates positively and significantly from 0, then we conclude that mutual funds outperform the market (p.2559, Kosowski et al., 2006).
The paper is organized in the following way. Section two places this paper in the relevant literature. Section three describes the data and samples on which we conduct our research. Section four comments on the methodology which we use. Section five shows the main empirical results. Section six shows our concluding note based on our empirical results. Section seven discusses on our results. Final section eight will be our list of references.
II. Literature review
Our literature review is divided into three subchapters, mutual funds, previous research findings, and size effect. In the first subchapter we clarify what mutual funds actually are. In the second subchapter we summarize previous literature on the performance of mutual funds. In the third and final subchapter we take a closer look at the ‘size effect’ on mutual fund performance.
6
Mutual funds
The term mutual fund is the common name for open-end investment companies (p. 96, Bodie et al., 2014). Basically, the function of mutual funds is to pool funds of individual investors and invest in a wide range of securities or other assets. The services provided by those mutual fund can be divided into four categories:
Administration & record keeping Diversification & divisibility Professional management Reduced transaction costs
An investor can buy and sell mutual fund shares directly from and to the mutual fund itself. Trading on those shares is only possible once a day after the market closes. Since mutual funds are open-end investment companies their shares are priced and traded at net asset value.
As mentioned early, mutual funds have become more popular, but the services provide by mutual funds do come at a price. There are four general classes of fees.
Front-end load Back-end load Expenses 12b-1 fees
A front-end load is basically a commission your need to pay to the fund when you buy a share. On the opposite, a back-end load is a sort of ‘exit-fee’ which you need to pay when you sell your share back to the fund. The last two fee classes are operating expenses and 12b-1 fees. Operating expenses are fees to keep the fund running. To this class belong administration expenses and the salaries of fund managers. Finally, 12b-1 fees are fees to pay for marketing and distribution costs. All those loads, expenses, and fees vary between each and every mutual fund, via this way one can choose to invest according to his own taste.
Previous research findings
One of the earliest research on mutual funds has been done by Jensen in 1968. He conducts a research in which he analyses the performance of 115 mutual funds over the
7
period 1945-1964. In order to measure a mutual fund’s performance, he uses his own Jensen’s alpha (1968). He comes to the conclusion that not only on average mutual funds aren’t able to outperform a buy-the-market-and-hold-it policy, but also that there is very little evidence that individual funds are able to significantly better job than the market.
Carhart (1997) uses his own four-factor model, which we use as well, to evaluate the performance of mutual funds. His model is an extension to the Fama & French (1993) three-factor model. To analyses the performance of mutual funds he takes a sample of all known US mutual funds over the period 1962 to 1993. Just like we do, Carhart also focuses on his model’s alpha to evaluate a mutual fund’s performance. Overall, he finds that mutual funds are not able to outperform the market. Although the top-decile mutual funds are able to earn higher returns than the market, most mutual fund generate pretty much the same returns as the market. Note that, this is after fees and expenses, but before any front-end or back-end loads. Carhart states that after those loads even investing in those top-decile funds would generate a negative return. For the lower-decile funds it’s even worse. Those funds underperform the market by about twice their investment costs.
Opposed to Jensen and Carhart, Wermers (2000) finds a positive return from investing in mutual funds over investing in a market index. Based on a newly constructed database, he performs an analysis of the mutual fund industry. This database is a merger between a data base of equity holdings of mutual funds and one containing monthly net returns, yearly expense ratios, turnover levels and other finical characteristics. By comparing the newly composed database to market he find that mutual funds’ stock holdings were able to
outperform the market by 1,3 percent per year net of fees and expenses. Of this 1,3 percent about 60 basis points is due to higher average returns of the stocks held by the fund.
Whereas the remaining 70 basis points is due to talent and skills of mutual fund managers in picking stocks that beat the market.
A different type of research has been done by Cremers and Petajisto (2009). They take a closer look at the performance of a mutual fund with respect to the funds ‘Active Share’ from 1980 till 2003. Active Share is a measure which represents the share of portfolio holdings that differ from the holdings of a certain benchmark, which in both their and our case is the market index holdings. In other words, an active manager attempts to
outperform a benchmark by taking positions that are different from the benchmark. Thus, if he is able to outperform the benchmark then one of the reasons for this would be that he
8
has superior stock selection skills. They find that fund with the highest Active Share are able to outperform their benchmark by as much as 6.5% per year. However, they also found that the lowest Active Share funds aren’t able to beat their benchmark. All in all, they conclude that at least some fund managers indeed have selection skills and are able to gain abnormal returns for mutual funds.
Kosowski et al. (2006) test weather alphas generated by the Carhart (1997) four-factor model of the best performing mutual funds, the so called ‘stars’, are due solely to luck or due to expert investment skills. If the performance of a mutual fund is solely based on luck, then examining the performance of those funds wouldn’t say anything about the
investment ability and skills of those funds’ managers. In order to test whether luck plays a significant role they examine the performance of 2.118 US mutual funds. They find that the performance of both the best and worst mutual funds isn’t solely due to luck. This means that mutual fund managers, of ‘stars’, at least have some degree of expert investment skills and influence on the performance.
Similar to Kosowski et al., a recent study one the performance of UK equity mutual funds has been conducted by Cuthbertson, Nitzsche and O’ Sullivan (2005). They also run a
Carhart four-factor regression to analyze the performance of mutual funds and the role of luck within the returns of those funds. Again, this is the same model as the one we use. However, instead of looking at the US market, they focus on the UK mutual fund market. In their research they used a sample of 935 mutual funds and looked at the period from 1975 till 2005. The aim of their research is to distinguish the effect of luck from the effect of skill on the performance of mutual funds. They come to the conclusion that between 5 and 10 percent of the top performing UK mutual funds generate abnormal returns due to skill instead of luck. This is broadly consistent with the empirical findings based on the US mutual fund market of Kosowski et al. (2006). The other top performing funds outperformed the market due to only luck. Furthermore, they conclude that the worst performing mutual funds aren’t just unlucky, but instead they demonstrate bad skill.
Finally, one of the most recent studies has been conducted by Fama and French (2010). Based on their own Fama and French three-factor model (1993) and Carhart (1997) four-factor model they evaluate the performance of US mutual funds from 1984 till 2006. Their conclusion is that mutual fund managers in general aren’t able to outperform the market.
9
This doesn’t mean that there are no mutual funds which outperform the market, but their performance isn’t significant enough to show up in the aggregate results.
Size effect
Most literature on the relation between mutual fund size and performance suggest that there’s a negative relationship between size and performance, the so called ‘size effect’. Chen et al. (2004) even find strong evidence that size erodes mutual fund performance. Their findings indicate that larger mutual funds often hold more illiquid portfolios. This illiquidity leads to a decrease in performance. Those findings are contrary to the view that larger funds have more available resources and economies of scale benefits, which would lead them to have superior performance.
III. Data
Data on mutual funds will be generated from the CRSP (Center for Research in Security
Prices). From the CRSP Survivor-Bias-Free US Mutual Fund Database we collect our sample,
which contains monthly returns and total net asset value for all open-end US mutual funds. These returns are net returns, which means that those are the returns after deduction of fees and expenses, but before any front- or back-end loads deduction (p. 1660, Wermers, 2000). We will focus our research on a ten-year period starting from January 2006 till December 2015.
To control for survivorship bias, we include only those mutual funds which have data available for the entire period in our sample. However, according to Malkiel (1995) and Elton, Gruber and Blake (1996), this will lead our mutual fund returns to be slightly
overestimated. This is due to the fact that funds who merged or liquidated during our time horizon are excluded from our sample. These funds are often merged or liquidated as a result of bad performance. By doing so, we exclude a group of mostly bad performing funds from our sample, which can lead to an overestimation of the mutual fund returns.
In total our sample contains the monthly returns of 7.483 mutual funds over a ten-year period, which add up to a total of 897.960 observations. The Fama French 3-factors, the momentum factor, market return, and risk-free rate will all be obtained from Kenneth French’s own website.
10
Table I: Summary Statistics of Excess Return per Category, and Carhart Four-Factors
The table shows the descriptive statistics of the excess return of our total sample and subcategories. Besides that, the descriptive statistic of the individual factors of the Carhart Four-factor model, Ri,t - Rf = αi + β1,i (RMkt – Rf)t + β2,i SMBt + β3,i HMLt + β4,i MOMt+ εi,t , are also presented. The period is from January 2006 till December 2015.
Variable Observations
Number
of Funds Mean Median
Standard
Deviation Min Max Excess return Total Sample 897960 7483 0.3722 0.4804 2.8756 -12.9125 7.7769 Excess return Small-Cap 860280 7169 0.3727 0.4806 2.8857 -12.9808 7.8105 Excess return Mid-Cap 22200 185 0.3733 0.6226 2.8913 -12.3332 7.7914 Excess return Large-Cap 15480 129 0.3415 0.5220 2.3179 -9.9525 5.8863
Mkt-Rf 120 - 0.6087 1.2350 0.4577 -17.23 11.35
SMB 120 - 0.0960 -0.0050 2.3270 -4.25 5.79
HML 120 - -0.1153 -0,1550 2.4035 -9.67 7.65
MOM 120 - 0,1555 0.7300 4.0964 -24.83 11.30
Table I shows the descriptive statistics of our sample. Based on the Lipper
Classification Codes we divide our sample of the total market into three subcategories.
Mutual funds with a total net asset value of less than one billion belong to the small-cap category, funds with one to five billion in total net asset value belong to the mid-cap
category, the other funds with a total net asset value of more than five billion belong to the large-cap category. Besides descriptive statistics of those categories and the entire sample, table I also shows descriptive statistics of the Carhart (1997) four-factors which we use.
IV. Methodology
The model which we use to analyze the performance of mutual funds is the Carhart (1997) four-factor model, which is an extension to two other models, starting with the CAPM (Capital Asset Price Model). Sharpe (1964) was the first one to describe the model and together with Jensen’s (1968) alpha you get a model which looks as follows. In the model, the excess return of an asset is explained by alpha, a volatility measure in relation to market excess return, and an error term. In our study RMkt includes all NYSE, AMEX, and NASDAQ
11
α since this is the performance measure of an asset or as in our case of a mutual fund. In other words, a positive value for Jensen’s alpha means that a fund outperforms the market.
Ri,t – Rf = αi + β1,i (RMkt – Rf)t + εt
Based on the model described above Fama & French (1993) build their own three-factor model. They add two variables, SMB and HML, to the model. SMB stand for small minus big and is included in the model to mimic the risk returns related to size. Like SMB, HML which stand for high minus low, is included in the model to mimic the risk related to book-to-market equity. The Fama and French three-factor model explains almost 90% of the
diversified portfolio returns, whereas the CAPM with Jensen’s alpha accounts only for about 70% of the diversified portfolio returns (p.21, Fama and French, 1993). The Fama and French three-factor model looks as follows.
Ri,t – Rf = αi + β1,i (RMkt – Rf)t + β2,i SMBt + β3,i HMLt +εi,t
Finally, Carhart (1997) adds the variable MOM to complete his four-factor model which we will be using to analyze the performance of mutual funds. MOM stands for momentum and is included in the model to capture the one-year anomaly. The complete model is shown down below.
Ri,t – Rf = αi + β1,i (RMkt – Rf)t + β2,i SMBt + β3,i HMLt + β4,I MOMt+ εi,t
All data for the risk-free rate as well as data for the SMB, HML and MOM variables comes from Kenneth French’s own website. The mutual fund returns are obtained from the CRSP Survivor-Bias-Free US Mutual Fund Database.
Hypothesis
As explained in more detail in the literature review section, there’s not really a concluding note on weather mutual funds are able to outperform the market or not. Yet, investors continue to pour money into mutual funds in pursuit of superior performance. To
12
mutual funds over the past decade, whereas most recent previous research doesn’t analyze the performance further than 2006. We use the same model as used in most of the previous research, namely the Carhart (1997) four-factor model. All of the above lets us formulate the following hypothesis.
H0 : US mutual funds have the same performance as the market.
H1 : US mutual funds outperform the market.
Translated into factors of the Carhart (1997) four-factor model, our hypothesis looks the following.
H0 : αus mutual funds = 0
H1 : αus mutual funds > 0
V. Empirical results
In this section we present and analyze our empirical results. As mentioned before, we have divided our panel data into three subcategories based on the Lipper Classification Codes. First, we look at the panel regression performed on the total sample, afterwards we analyze the panel regression results for all three of the subcategories. In all of our panel regressions we use the robust t-statistic. Our main point of focus are the alphas. If they differ
significantly and positively from zero, then this indicates outperformance of mutual funds over the market (p.2559, Kosowski et al., 2006).
Total sample results
Our total sample consists of 897.960 observations and 7.483 mutual funds. Table II shows the panel regression for the entire sample. The alpha of this regression has a value of -0,0123, which means that mutual funds underperform the market by 0,0123 percent per month. However, this isn’t significant at any level. So based on this regression we do not reject the null hypothesis.
13
Table II: Panel Regression of Total Sample
The table shows the regression results of the following model: Ri,t - Rf = αi + β1,i (RMkt – Rf)t + β2,i SMBt + β3,i HMLt + β4,i MOMt+ εi,t . The period is from January 2006 till December 2015.
Regression Total Sample
Mkt-Rf 0.6196*** (25.68) SMB -0.0515 (1.80) HML -0.0831* (-2.49) MOM -0.0461* (-2.56) α -0.0123 (-0.19) Observations 897960 Number of funds 7483 R-squared 0.9500 t-statistics in parentheses. *: significance at 10% level. **: significance at 5% level. ***: significance at 1% level.
Small-, mid-, large-cap results
To broaden our research, we also investigate if mutual fund size plays a role in mutual fund performance. Table III shows the results from our regression on the three subcategories. The alphas of small- and mid-cap are both negative but not significant. They indicate that on a monthly basis, mutual funds underperform the market with 0,0131 and 0,0107 percent, respectively. On the other hand, large-cap mutual funds have a positive alpha of 0,0301 percent. This is in contrast with the view of Chen et al. (2004), in which they find strong evidence that size erodes mutual fund performance. However, this outperformance of mutual funds over the market isn’t significant, so we do not reject the null hypothesis based on any of the subcategories.
14
Table III: Panel Regression of Small-, Mid-, Large-Cap Mutual Funds
The table shows the regression results of the following model, for the Small-, Mid-, and Large-Cap
subcategories: Ri,t - Rf = αi + β1,i (RMkt – Rf)t + β2,i SMBt + β3,i HMLt + β4,i MOMt+ εi,t . Based on the Lipper Classification
Codes each mutual fund is assigned to one of the subcategories. The period is from January 2006 till
December 2015.
Regression Small-Cap Mid-Cap Large-Cap
Mkt-Rf 0.6212*** 0.6307*** 0.5141*** (25.37) (35.19) (36.02) SMB 0.0540 0.0076 -0.0276 (1.88) (0.27) (-1.23) HML -0.0846* -0.0492 -0.0501* (-2.50) (-1,83) (-2.44) MOM -0.0465* -0.0407 -0.0298* (-2.57) (-2.17) (-2.19) α -0.0131 -0.0107 0.0301 (-0.20) (-0.18) (0.63) Observations 860280 22200 15480 Number of funds 7169 185 129 R-squared 0.9495 0.9572 0.9559 t-statistics in parentheses. *: significance at 10% level. **: significance at 5% level. ***: significance at 1% level.
VI. Conclusion
In this paper we investigate the performance of mutual funds over the past decade. We first describe the main characteristics of mutual funds and reviewed existing literature on the performance of mutual funds. Afterwards, we use a sample of 7.483 mutual funds in total, each with data available for the entire period, to perform our own research on the
performance of mutual funds. To analyze the performance, we use Carhart (1997) four-factor model. Our main point of focus of the model is the alpha. If alpha is positive and significant, mutual funds are able to outperform the market. To broaden our research and analyze the effect of mutual fund size on performance, we divide our sample into three subcategories, small-, mid-, and large-cap. We run a regression both on those subcategories and on the total sample.
When we look at the performance of mutual funds in our total sample, we see that they underperform the market by 0,0123 percent on a monthly basis. This in line with
15
findings of Fama and French (2010) among others, but in contrast with findings of for example Wermers (2000). However, since the alpha we find isn’t significant, we do not reject the null hypothesis which states that mutual funds have the same performance as the market.
The evaluation of the performance of the small-, mid-, large-cap samples show that the first two have a slight negative alpha of -0,0131 and -0,0107 percent, whilst the large-cap sample has a slightly bigger and positive alpha of 0,0301 percent. This means that the bigger the mutual fund, the better their performance. Those results are in contrast with conclusion of Chen et al. (2004), who states that mutual fund performance erodes with an increase in size. Again, neither one of the alphas we find is significant, so we do not reject the null hypothesis.
Ultimately, based on the Carhart (1997) four-factor model, we do not find a significant outcome about whether US mutual funds are able to outperform and beat the market or not.
VII. Discussion
Based on previous literature, there isn’t a general position about the performance of mutual funds with respect to the market. This leads our results to be in cohesion with some of the previous literature and in contrast with others.
With our regression on the three subcategories we find that our large-cap results are in not in line with the view of Chen et al. (2004). One possible reason for the positive alpha of the large-cap category, opposed to the negative alpha found with the small- and mid-cap sample, could a form of ‘success bias’. With this we mean that mutual funds which have the best performance, and thus the highest alpha, also attract the most investors as a results of their performance. As a consequence, mutual funds with the highest alpha will because of that also be the funds that belong to the large-cap category. When analyzing our results we come to the conclusion that instead of size eroding mutual fund performance, like Chen et al. (2004) found, we find that size enhances mutual fund performance.
For further research, we think it’s interesting to investigate if there’s a difference in performance among mutual funds which invest in different industries. This can be done by
16
splitting up a mutual fund database, based on the individual fund’s investment targets. We are not able to do this since the necessary data to do so was unavailable.
Finally, another way to broaden and enrich our research, is to look if there’s a difference in mutual fund performance between US and non-US mutual funds.
17
VIII. References
Bodie, Z., Kane, A., Marcus A.J.,[2014]. Investments. McGraw-Hill.
Carhart, M.M., 1997. On persistence in mutual fund performance, Journal of Finance 52, 57– 82.
Chen, J., Hong, H., Huang, M. and Kubik, J., 2004. Does fund size erode mutual fund performance? The role of liquidity and organization, American Economic Review 94(5): 1276–1302.
Cremers, K.J.M., Petajisto, A., 2009. ‘How Active Is Your Fund Manager? A New Measure That Predicts Performance’, The Review of Financial Studies 22, 3329-3365.
Elton, E.J., Gruber, M.J. and Blake, C.R., 1996. Survivorship bias and Mutual Fund Performance, The review of financial studies 9, 1097-1120.
Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds, Journal
of Financial Economics 33, 3–56.
Fama, E., French, K., 2010. Luck versus skill in the cross section of mutual fund returns,
Journal of Finance 65, 1915–1947.
Gruber, M.J., 1996. Another puzzle: The growth in actively managed mutual funds, Journal
of Finance 51, 783-810.
Jensen, M., 1968. The performance of mutual funds in the period 1945–1964. Journal of
Finance 23, 389–416.
Khorana, A., 2012. What Drives Market Share in the Mutual Fund Industry. Review of
Finance 16, 81-113.
Kosowski, R., Timmermann, A., Wermers, R., and White, H., 2006. Can mutual fund “stars” really pick stocks? New evidence from a bootstrap analysis, Journal of Finance 61, 2551– 2595.
Malkiel, B.G., 1973 [2011]. A Random Walk Down Wall Street. W. W. Norton.
Malkiel, B.G., 1995. Returns from investing in equity mutual funds: 1971-1991, Journal of Finance 50, 549-572.
Malkiel, B.G., 2013. Asset Management Fees and the Growth of Finance, Journal of
Economic Perspectives 27, 97-108.
Sharpe, W.F., 1964. Capital Asset Prices: A theory of market equilibrium under conditions of risk, Journal of Finance 19, 425-42.
18
Wermers, R., 2000. Mutual fund performance: An empirical decomposition into stock-picking talent, style, transaction costs, and expenses, Journal of Finance 55, 1655-1703.
