Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
MUTUAL FUNDS
PERFORMANCE, TYPES
AND IMPACTS ON STOCK RETURNS
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or
by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no
expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of information
contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in
rendering legal, medical or any other professional services.
FINANCIAL INSTITUTIONS
AND SERVICES
MUTUAL FUNDS
PERFORMANCE, TYPES
AND IMPACTS ON STOCK RETURNS
DONALD EDWARDS
EDITOR
New York
Copyright © 2017 by Nova Science Publishers, Inc.
All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted
in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying,
recording or otherwise without the written permission of the Publisher.
We have partnered with Copyright Clearance Center to make it easy for you to obtain permissions to
reuse content from this publication. Simply navigate to this publication’s page on Nova’s website and
locate the “Get Permission” button below the title description. This button is linked directly to the
title’s permission page on copyright.com. Alternatively, you can visit copyright.com and search by
title, ISBN, or ISSN.
For further questions about using the service on copyright.com, please contact:
Copyright Clearance Center
Phone: +1-(978) 750-8400 Fax: +1-(978) 750-4470 E-mail: info@copyright.com.
Independent verification should be sought for any data, advice or recommendations contained in this
book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to
persons or property arising from any methods, products, instructions, ideas or otherwise contained in
this publication.
This publication is designed to provide accurate and authoritative information with regard to the subject
matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in
rendering legal or any other professional services. If legal or any other expert assistance is required, the
services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS
JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A
COMMITTEE OF PUBLISHERS.
Additional color graphics may be available in the e-book version of this book.
ISBN: 978-1-53610-659-6
Preface vii
Chapter 1 A Review of Performance, Screening and Flows
in Screened Mutual Funds 1
Ainulashikin Marzuki and Andrew C. Worthington
Chapter 2 Does the Choice of Performance Measure Matter
for Ranking of Mutual Funds? 73
Amporn Soongswang and Yosawee Sanohdontree
Chapter 3 Mutual Fund Prediction Models Using Artificial
Neural Networks and Genetic Programming 93
Konstantina Pendaraki, Grigorios Ν. Beligiannis
and Alexandra Lappa
Index 129
PREFACE
asset value, while GP’s result’s outperforms ANN’s results in the prediction of
mutual funds’ return. Overall, our experimental results showed that both
ANNs and GP comprise useful and effective tools for the development of
mutual fund prediction models.
In: Mutual Funds ISBN: 978-1-53610-633-6
Editor: Donald Edwards © 2017 Nova Science Publishers, Inc.
Chapter 1
ABSTRACT
Islamic mutual funds (IMFs) continue to grow as an alternative
investment vehicle for investors wishing to integrate Islamic values and
secular financial objectives in their investments. The most distinctive
feature of IMFs lies in screening strategies based on the application of
Shariah (Islamic law). Conventionally, this involves the application of
exclusionary screening, whereby fund managers screen out companies
involved in certain activities, including riba (interest), gharar
(uncertainty), and maysir (gambling), and prohibited products from their
portfolios as prescribed by the Quran, Sunnah and related Islamic texts.
The central outcome is that the managers of IMFs, unlike those of
conventional mutual funds (CMFs), necessarily access only a subset of
the population of investments available. This has dramatic implications
for many conventional dimensions of mutual fund behavior, including
performance, the flow of funds, and the fund selection behavior of
investors. This chapter reviews the theoretical and empirical literature
relating to mutual fund performance, screening, and fund flows. The
literature on performance starts with a discussion of the development of
*
Corresponding author: ainulashikin@usim.edu.my.
2 Ainulashikin Marzuki and Andrew C. Worthington
1. INTRODUCTION
Islamic finance is one of the fastest growing segments of the global
finance industry, comprising financial institutions, products and services
complying with Shariah (Islamic law) (Gait and Worthington, 2008). While
the practice of Islamic finance in the modern world only commenced with
savings institutions in 1963 (in Egypt and Malaysia), it has now spread to
many other types of financial products and services, including banking, funds
management (including mutual funds), takaful (Islamic insurance) and sukuk
(Islamic bonds). Islamic financial products have now proliferated across
almost every aspect of contemporary financial services, with comparable
products complimenting those found in the conventional finance sector.
Consequently, the number of financial institutions offering Islamic financial
products and services has also increased, from just 300 in 2005 (El Qorchi,
2005) to 628 at the end of 2009 (Lee and Menon, 2010) with operations in
more than 75 countries (El Qorchi, 2005). Additionally, the value of Shariah
compliant assets grew 25 percent from US$758 billion in 2007 to US$951
billion at the end of 2008 (International Financial Services London, 2010).
One of the fastest growing Islamic financial products is Islamic mutual
funds (IMFs), growing strongly since at least the pronouncement by the
Council of the Islamic Fiqh Academy in Jeddah in 1990 that equity investment
was permissible as long as it complied with Shariah (Nathie, 2009). Since
A Review of Performance, Screening and Flows … 3
then, many asset management companies have offered IMFs alongside their
existing conventional mutual funds (CMFs) and socially responsible
investment (SRI) funds. For example, the number of IMFs worldwide has
risen more than threefold from 200 funds in 2003 to 680 funds in 2008,
representing various types of IMFs (Eurekahedge, 2008). Concomitantly, the
value of assets managed under these funds has also grown, from US$20 billion
in 2003 to US$44 billion in 2008 (Ernst & Young, 2009). At present, equity
funds represent the largest segment of IMFs (about 40 percent), followed by
fixed income (16 percent), real estate and private equity (13 percent) with the
remainder in cash or commodities or other Islamic funds (Eurekahedge, 2008).
For the most part, these funds are concentrated in several regions, including
the Middle East and North Africa, the Asia Pacific, North America and
Europe, with more than half currently invested in the Middle East and the Asia
Pacific (International Financial Services London, 2010, p. 5).
In Malaysia, the development of IMFs is relatively more important for
several reasons. First, IMFs have a prospective role as a policy tool in the
ongoing development of the Malaysian capital market. Malaysia already has
one of the most well developed conventional and Islamic capital markets in
South-East Asia and among Islamic countries, respectively. In fact, the
Malaysian government has highlighted the importance of IMFs in its
Malaysian Capital Market 2001 blueprint. According to this, the government
will “… facilitate the development of a wide range of competitive products
and services related to the Islamic capital market” and “… create a viable
market for the effective mobilization of Islamic funds,” one of which is IMFs
(Securities Commission Malaysia, 2001).
Second, the equity market, including investment funds, is an important
buffer to the significant increase in household debt in the Malaysian economy
(Bank Negara Malaysia, 2011). However, the size of mutual fund assets
relative to the total financial assets of Malaysian household remains small
compared to other developed and developing countries. In 2010, total mutual
fund assets (net asset value) in Malaysia were RM226.81 billion (Securities
Commission Malaysia, 2012), constituting about 16 percent of the total
financial assets of Malaysian households as reported in the 2010 report (Bank
Negara Malaysia, 2011). Of this, RM24.04 billion was in IMFs, and thus they
account for about 1.67 percent of Malaysian household sector financial assets
(Securities Commission Malaysia, 2012). This size implies that there is
potential for IMFs to grow further and become the main catalyst for the
growth of the overall mutual fund industry in Malaysia (Securities
Commission Malaysia, 2011). This will not only help to support the growth of
4 Ainulashikin Marzuki and Andrew C. Worthington
the Malaysian capital markets (including the Islamic capital market) but also
the Malaysian economy as a whole.
Third, even though the asset size of IMF industry is small relative to that
of the total mutual fund industry, Malaysia’s IMF is among the largest in both
asset under management and number of funds launched in the world besides
Saudi Arabia and Kuwait [see, for instance, Securities Commission Malaysia
(2012), Eurekahedge (2008), Hoepner et al. (2011), Abderrezak (2008), and
Nainggolan (2011)]. This makes Malaysia one of the major players of the
IMFs globally. Finally, IMFs potentially appeal to not only the Muslim
investors but also to non-Muslim investors who may regard these funds as
another variant of an ethical or SRI fund. Investors who are ethically
(religiously) concerned are interested to integrate ethical (Shariah) values as
well as financial objectives in their investment decisions. The distinctive
feature of IMFs lies in their screening strategies with IMFs applying screening
based on Shariah. However, in contrast to ethical/SRI funds, IMFs mainly
apply exclusionary screening as against both positive and negative screening
in SRI funds.
The literature on performance starts with the discussion on the
development of mutual fund performance evaluation techniques and the
related theories behind it. This is important to provide a general understanding
on the importance of performance measurement and various ways to measure
mutual fund performance. We will also review the literature on fund attributes
and their influence on mutual fund performance. In the screening literature, we
review the impact of screening and differences in screening strategies to firm
and mutual fund performance. Finally, we review the literature concerning the
behavior of mutual fund investors, which uses mutual fund flows as the proxy.
In the area of mutual fund investors, we specifically focus on how IMF
investors make fund selection decision and examine if these investors are able
to select funds that are able to earn positive returns in subsequent period.
Overall, the literature on IMFs remains scarce and lags behind compared with
the literature on the CMFs and SRI funds. Thus, this section reviews related
theoretical and empirical studies on SRI screened and unscreened funds to
draw the necessary bases for the study of IMFs.
The remainder of the chapter comprises five sections. Section 2 discusses
the development of mutual fund performance evaluation, any criticisms, and
the attributes of mutual fund performance. Section 3 reviews the impact of
screening and differences in screening strategies on performance at both the
firm and portfolio level. Section 4 examines studies relating to mutual fund
flows and its relationship to past performance and other fund characteristics as
A Review of Performance, Screening and Flows … 5
well as the volatility of the mutual fund flow, both screened and unscreened
mutual funds. Section 5 reviews another strand of mutual fund flow literature
in the predictability of future performance using fund flow information or
‘smart money’. The final section of the chapter provides some concluding
remarks.
𝑅𝑖 −𝑅𝑓
𝑆𝑅𝑖 = (1)
𝜎𝑖
where 𝑆𝑅𝑖 is the Sharpe ratio, 𝑅𝑖 is the mean return of fund i over the interval
considered, 𝑅𝑓 is the average risk-free rate over the interval considered and 𝜎𝑖
is the standard deviation of return on fund i over the interval considered.
Next, Treynor (1965) introduced the Treynor ratio, which used portfolio
beta (β) as a measurement of risk similar to the CAPM model. This means that
instead of using CML, he compares portfolio risk and return to the security
market line (SML). It measures excess returns of the riskless interest rate per
unit of systematic risk, which is as follows:
𝑅𝑖 −𝑅𝑓
𝑇𝑅𝑖 = (2)
𝛽𝑖
8 Ainulashikin Marzuki and Andrew C. Worthington
where 𝑇𝑅𝑖 is Treynor’s index, 𝑅𝑖 is the average return on the mutual fund over
the measurement period, 𝑅𝑓 is risk-free rate of return and 𝛽𝑖 (beta) is the
systematic risk between the fund and the market index. 𝛽𝑖 is estimated by
regressing the portfolio return, 𝑅𝑖 with the market return, 𝑅𝑚 and divided by
the variance of the market return as follows.
𝑪𝒐𝒗 (𝑹𝒊, 𝑹𝒎 )
𝜷𝒊,𝒎 =
𝑽𝒂𝒓(𝑹𝒎 )
(3)
where 𝑅𝑖,𝑡 is the mean return on fund or portfolio i at time t, 𝑅𝑓,𝑡 is the average
risk-free rate at time t, 𝛽𝑖 represents systematic risk of the fund or portfolio
relative to the market return, 𝑅𝑚,𝑡 is the mean return on market representing
the mean-variant efficient benchmark, 𝛼𝑖 captures any excess return above
market benchmark and 𝜀𝑖 is the error term.
If managers have stock selection skill, then the excess portfolio return
should be higher than the excess market portfolio return after adjustment to the
systematic risk. The intercept of a regression in the CAPM equation captures
the additional return a manager generates. A statistically significant positive
alpha above the expected performance indicates above average performance
and alternatively, a statistically significant negative alpha indicates
underperformance.
However, these models, which rely mainly on the CAPM framework,
received criticism [see, for example, Jensen (1972) and Roll (1978)]. These
criticisms were concerned about its oversimplified assumptions, inefficiency
A Review of Performance, Screening and Flows … 9
where 𝑅𝑖,𝑡 is the return on fund i during period t; 𝛼𝑖 identifies the stock
selection ability, 𝑅𝑚𝑡 is the return on the market benchmark during period t
and (𝑅𝑚𝑡 )2 is the squared market return. The term gamma, 𝛾𝑖,𝑡 , indicates
market timing. If 𝛾𝑖,𝑡 is positive and significant then the fund manager is a
successful market timer. When they tested the model using monthly return
data of US mutual funds from 1953 to 1962, they found that there was no
significant evidence that fund managers possess market timing ability. Out of
57 mutual funds, only one fund demonstrated market timing ability.
Henriksson and Merton (1981) proposed another model to test the market
timing ability of fund managers. The intuition behind this model is similar to
the previous one developed by Treynor and Mazuy (1966), where a mutual
fund manager allocates capital between cash and equities based on forecasts of
the future market return, except now the manager decides between a small
number of market exposure levels. The model is as follows.
where 𝑅𝑖,𝑡 is the return of fund i in period t, 𝛼𝑖 is the stock selection ability,
𝑅𝑚𝑡 is the return on the market benchmark in period t and the term delta 𝛿𝑖 is
the market timing coefficient. If the value for market timing is positive and
significant, then the fund manager is a successful market timer and knows
when to enter and exit the market to take advantage of market upturns and
avoid market downturns. 𝐷𝑡 is a dummy variable that takes a value of one if
the market return is positive and zero otherwise. 𝜀𝑖,𝑡 is the error term.
Similarly, tested on 116 mutual funds in the US from 1968 to 1980, only three
funds had significant market timing ability (Henriksson, 1984). Other studies
that used this model indicated that there was little evidence that managers
possessed superior market timing abilities [see, for example, Sawicki and Ong
(2000) and Sinclair (1990)]. Later, Bollen and Busse (2001) found evidence of
market timing ability among fund managers. They used data from 1985 to
1995, which consisted of 230 funds and demonstrated that mutual funds
exhibited significant timing ability when using daily data compared to monthly
A Review of Performance, Screening and Flows … 11
𝑅𝑖,𝑡 − 𝑅𝑓,𝑡 = 𝛼3,𝑖 + 𝛽𝑀,𝑖 (𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 ) + 𝛽𝑆,𝑖 𝑆𝑀𝐵𝑡 + 𝛽𝑉,𝑖 𝐻𝑀𝐿𝑡 + 𝜀𝑖,𝑡 (8)
where 𝑅𝑖,𝑡 is the return on fund i in time t, 𝑅𝑓,𝑡 is the risk-free rate in time t,
𝑅𝑚,𝑡 is the return on a market portfolio in time t, 𝑆𝑀𝐵𝑡 is the return on
portfolio of small minus large firms in time t, 𝐻𝑀𝐿𝑡 is the return on portfolio
of high minus low book-to-market stocks in time t.
Even though the addition of these two risk factors enhanced the
explanatory power of stock returns, Fama and French (1993) claimed that this
model also suffers from another anomaly, that is, the continuation of short
term returns reported by Jegadeesh and Titman (1993). Using the works of
Jegadeesh and Titman (1993) and Bondt and Thaler (1985), Carhart (1997)
extended Fama and French’s (1993) three factor model by adding a one-year
momentum anomaly. This model is termed as the Carhart’s four-factor model
and is as follows:
𝑹𝒊,𝒕 − 𝑹𝒇,𝒕 = 𝜶𝟒,𝒊 + 𝜷𝑴,𝒊 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 ) + 𝜷𝑺,𝒊 𝑺𝑴𝑩𝒕 + 𝜷𝑽,𝒊 𝑯𝑴𝑳𝒕 + 𝜷𝑴,𝒊 𝑴𝑶𝑴𝒕 +
𝜺𝒊,𝒕 (9)
where 𝑅𝑖,𝑡 is the return on fund i in time t, 𝑅𝑓,𝑡 is the risk-free rate in time t,
𝑅𝑚,𝑡 is the return on a market portfolio in time t, 𝑆𝑀𝐵𝑡 is the return on
portfolio of small minus large firms in time t, 𝐻𝑀𝐿𝑡 is the return on portfolio
of high minus low book-to-market stocks in time t and 𝑀𝑂𝑀𝑡 is the rate of
return on portfolios of high minus low momentum (prior 1-year return) stocks
in time t.
He defined momentum as the difference between the previous best
performing and worst performing stocks. Consequently, the Fama and
French’s (1993) three factor and Carhart’s (1997) four factor models have
become very popular among academicians and practitioners and are widely
used to measure managed fund performance.
Many studies of portfolio performance compared mutual fund
performance to a single market index. However, there is problem of using a
single index model. Different types of assets held in the managed portfolio
may perform differently than the benchmarks. In a study on the cost of
information and managed funds performance, Ippolito (1989) found that
mutual funds are able to earn abnormal return and it is more than enough to
compensate for the information cost. However, on further investigation of this
result, Elton et al. (1993) found that the results of over performance were due
14 Ainulashikin Marzuki and Andrew C. Worthington
to benchmark error and extended the model from a single index (S&P 500) to
a multi-index model (three-index model). The result showed a reverse finding
and this demonstrated the importance of choosing the correct market
benchmark to explain the behavior of mutual funds. Later, Elton et al. (1996)
added another risk factor in the model to further explain the variation in the
risk factors. In their empirical work, this model includes factors such as the
S&P Index, a size index, a bond index and a value or growth index.
𝑅𝑖,𝑡 − 𝑅𝑓,𝑡 = 𝛼𝑖 + 𝛽𝑖,0 (𝑅𝑚,𝑡 − 𝑅,𝑡 ) + 𝛿𝑖′ [(𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 )𝑍𝑡−1 ] + 𝜀𝑖𝑡 (10)
where 𝑅𝑖,𝑡 is the return on fund i in time t, 𝑅𝑓,𝑡 is the risk-free rate in time t,
𝑅𝑚,𝑡 is the return on a market portfolio in time t, 𝛿𝑖′ measures the response
coefficients of conditional beta with respect to lagged public information
variables.
Testing the model empirically on a sample of 67 US equity funds from
1968 to 1990, Ferson and Schadt (1996) found that the conditional model
provided improved mutual fund performance (zero performance) compared to
the unconditional model (underperformance), Dahlquist et al. (2000) also
found that on average alpha is zero and not negative as evidenced in
unconditional findings. Christopherson et al. (1998) also argued that allowing
alpha and beta to be time varying, meaning that alpha is also interacted with
the information variables, will result in better performance measurement. Kon
and Jen (1978) found that risk is not stationary through time and suggested
using a conditional model to measure fund performance.
A Review of Performance, Screening and Flows … 15
In screened fund studies, for example, Luther et al. (1992), it was found
that performance evaluation for ethical funds is sensitive to the type of
benchmarks used. As ethical funds incline to be biased towards small cap
companies, the authors suggest that comparison of ethical funds to the small
cap index is only relevant to measure its performance. Empirically, they found
that ethical funds outperform the small cap index. Thus, this implies that
benchmark selection is important in evaluating mutual fund performance.
Efficient market hypothesis (EMH) is central in theoretical and empirical
works on investment and fund management. Fama (1970) developed EMH,
which posits that security prices reflect all information available in the market
place. According to Fama (1970), there are three forms of market efficiency –
weak form, semi-strong form and strong form. Weak form implies that
security prices already reflect all past information while the semi-strong means
security prices reflect all publicly available and past information. The strong
form asserts that security prices already anticipate past, public and privately
available information.
The implication of the EMH is that theoretically it is impossible for active
fund managers to outperform the market on a risk adjusted return basis
consistently. Thus, the main issues underlying the work of managed fund
performance is to test the market efficiency whether fund managers are able to
obtain abnormal returns. Many studies in the US reveal underperformance of
mutual funds that supports the notion of efficient market hypothesis (Fama,
1970).
However, even though the market is assumed efficient in the long run,
there is evidence that investors can exploit the market by identifying mispriced
securities. Findings from Grossman and Stiglitz (1980) indicate that in a
strongly efficient market, fund managers are able to earn superior performance
at gross return by gathering costly information. However, the gross abnormal
performance is only sufficient to compensate the cost or expenses for the
information, thus, in net, there are no above average returns. Berk and Green
(2004), based on their analytical work, found that managers’ skills are
heterogeneous, however, since skilled fund managers charge higher fees, the
high fees affect the performance of the fund, which results in under
performance or zero performance. In another study, Gruber (1996) used a
four-factor model, to investigate the performance of 270 US equity funds from
1985 to 1994. Similarly, he found that, on average, these funds earn positive
risk adjusted return before expenses. Net of expenses, these funds
underperform the benchmark, which implies that fund managers do have
superior skill. However, the amount of fees charged is more than the value
18 Ainulashikin Marzuki and Andrew C. Worthington
added. Besides the risks and returns, other factors may also influence mutual
fund performance. The following section reviews literature on the influence of
fund attributes to fund performance.
3. PORTFOLIO SCREENING
The essence of portfolio screening is the incorporation of ethical, social,
and religious values in the investment decision. Investors who buy screened
investments care about achieving their financial objectives and concern if the
fund investment objectives also complement their ethical, social, and religious
beliefs. This approach requires investors to screen out (negative screening)
companies with business activities that are not consistent with the investors’
values and include (positive screening) companies that involve in activities
that benefit their stakeholders.
However, these strategies may have a negative impact on portfolio
performance. Movement away from MPT may shift the mean variance
framework from the efficient frontier to less favorable return for a given level
of risk or, alternatively, higher risk exposure for a given level of return. The
implication is investors hold less diversified portfolios with potentially high
unsystematic risk.
According to Jones et al. (2008), what drives the performance of screened
funds are the investment strategies and the portfolio screening. Differences in
investment strategy and portfolio screenings not only influence differences in
performance between the screened and non-screened funds but also among the
screened funds. This section reviews the theoretical and empirical aspects of
screening that include the impact of screening to performance both at firm and
portfolio levels. As theoretical and empirical literature concerning IMFs is
scarce, the review substantially draws upon studies on SRI or ethical funds.
There is one and only one social responsibility of business – to use its
resources and engage in activities designed to increase its profit so long as it
stays within the rules of the game, which is to say, engages in open and free
competition without deception or fraud.
Merton (1987), Angel and Rivoli (1997) argued that ethical screening can be
considered as a kind of segmentation to the equity market where some
companies are eliminated from some of the segments. This will raise the cost
of capital of these companies and, consequently, create incentives for the
companies to change their behavior. However, whether the influence is
significant depends on the fraction of investors or screened funds excluding
the companies and, in turn, the financial performance of the screened funds.
The higher the fraction of investors excluding the companies and screened
funds that are able to show comparable risk adjusted returns to unscreened
funds the larger the impact is to influence the companies’ behavior.
Heinkel et al. (2001) developed a theoretical model to understand the
effect of ethical investing on corporate behavior in an equilibrium model with
efficient capital markets. In their model, they assumed that only two types of
investors exist: green investors and neutral investors. In addition, a firm can
choose of two types of technology: either a clean technology or a polluting
technology. The presence of ethical investing may change the corporate
behavior from using a polluting technology to a clean technology if the
fractions of green investors are larger than neutral investors are. Fund
managers employing screen strategy will exclude companies with a polluting
technology. Thus, leaving polluting firms in the small investment portfolios
and, consequently, increasing their cost of capital due to the reduction of risk
sharing opportunities. If the increase in cost of capital exceeds the cost of
being ethical, the corporate behavior of polluting firms would be affected
causing them to change to clean technology and, hence, become ethical.
Barnea et al. (2005) also investigated the effects of negative pollution
screening on the investment decisions of polluting firms. The issue is
examined in an equilibrium setting with endogenous investment decisions,
where firms are allowed to choose the level of investment. The study
concluded that negative screening reduces the incentives of polluting firms to
invest, which lowers the total level of investment in the economy.
In the aspect of religious screening and firm value, two studies examine
the impact of sins screening to firm value [see, for examples, Hong and
Kacperczyk (2009) and Derigs and Marzban (2008)]. Hong and Kacperczyk
(2009) found that excluding portfolios from investing in “sin” stocks may
impose large costs to the fund performance. This is because they report that sin
stocks outperform other stocks in their sample. In Malaysia for example, many
companies that seek listings are interested in having their name listed as
Shariah compliant IPOs. The same goes for public listed companies. The
numbers of companies announced as Shariah compliant are increasing from
24 Ainulashikin Marzuki and Andrew C. Worthington
controlled variable. The findings proved that most of the ethical fund
portfolios tend to hold small capitalization companies. They proposed to
consider a two-factor benchmark to deal with the small size effect. Adopting
this method, they found that the outperformance no longer existed. In
conclusion, there is no significant difference between SRI and non-SRI funds.
In addition to the studies in the UK and the US, there is also evidence
concerning the performance of SRI funds in other developed countries, for
example Australia (Bauer et al., 2006) and Canada (Bauer et al., 2007). Instead
of focusing in a single country assessment, other studies attempted to compare
the performance of SRI funds between countries to provide better insights into
the wider performance of ethical funds across countries [see, for example, US,
Germany and Switzerland (Schroder, 2004), Germany, the UK and the US
(Bauer et al., 2005), European markets (Kreander et al., 2005), and all over the
world (Renneboog et al., 2008a)].
Bauer et al. (2006) investigated the performance of SRI funds in Australia
by employing a conditional multi-factor model and controlling for investment
style, time-variation in betas and home bias from 1992 to 2003. They provided
no evidence of significant differences in risk-adjusted returns between SRI and
conventional funds during the sample period. However, they found that
domestic ethical funds underperformed their conventional counterparts
significantly from 1992 to 1996, whereas SRI fund performance matched
closely the performance of conventional funds from 1996 to 2003. They
suggested that, as SRI funds are new in the market, these funds experience
learning phase period before providing returns equivalent to those of
conventional mutual funds.
Kreander et al. (2005) examined the performance of SRI funds in the
broader European market. Their study includes European countries such as
Belgium, Germany, Netherlands, Norway, Sweden, Switzerland, and, the UK.
The findings indicated that the European SRI fund performance is at best par
with the conventional counterparts when comparing with the Morgan Stanley
Capital International (MSCI) World Index. However, the Swedish SRI funds
outperform the local benchmark and their performance is at par with the global
index.
A more comprehensive study of SRI performance conducted by
Renneboog et al. (2008a) examined the performance of SRI funds all over the
world and included religious screening funds in their sample. They found that
SRI funds in the US, UK, many European countries, and Asia Pacific
underperform their local benchmarks. With the exception of France, Japan,
and Sweden, SRI funds were not statistically different from the performance of
28 Ainulashikin Marzuki and Andrew C. Worthington
They concluded that Islamic and conventional indices have similar reward to
risk and diversification benefits.
Hussein and Omran (2005) examined the behavior of the Dow Jones
Islamic indices against their counterparts, Dow Jones non-Islamic indices from
1995 to 2003. They divided the sample into three sub periods: entire period,
the bull period, and the bear period. They found that Islamic indices
outperformed non-Islamic indices over the entire period and the bull market
period but underperformed the non-Islamic indices in the bear market period.
However, the results were not statistically significant. Another study by
Hussein (2005) examined the impact of Shariah screening on the performance
of FTSE Global Islamic index and Dow Jones Islamic Market index (DJIMI)
using a number of performance measurement techniques in both the short-run
and long-run from 1996 to 2003. Overall, his studied further confirmed his
earlier study that the Islamic index outperformed its counterpart during the
entire period, bull, and bear periods. He concluded that Shariah screening did
not give adverse effect to the performance of Islamic portfolio compared to
unscreened portfolios.
Albaity and Ahmad (2008) assessed the performance of the Kuala Lumpur
Composite Index (KLCI) (conventional) and the Kuala Lumpur Shariah
Iindex (KLSI) (Islamic) from 1990 to 2005 periods using Sharpe ratio,
Treynor ratio, Jensen alpha and excess standard deviation-adjusted return.
They found that KLSI had lower returns but with lower risks. However,
overall they found that KLSI index do not significantly underperform KLCI.
While all of the above studies focused on the performance of Islamic
indices, the following studies addressed the performance of IMFs in various
regions [see, for example, Elfakhani and Hassan (2005), Hayat (2006),
Abderrezak (2008), Hoepner et al. (2011), and Hayat and Kraeussl (2011)],
Malaysia [see Abdullah et al. (2007), Mansor and Bhatti (2011a, 2011b,
2011c) and Mansor et al. (2012)], Arab Saudi [see Merdad et al. (2010) and
BinMahfouz and Hassan (2012)] and Pakistan [see Mahmud and Mirza (2011)
and Razzaq et al. (2012)]. Similarly, these studies provided inconsistent
findings whether IMFs outperformed or underperformed their CMFs as well as
Islamic or conventional market benchmarks.
Elfakhani and Hassan (2005) compared the performance of 46 IMFs all
over the world to both conventional and Islamic benchmarks from 1997 to
2002. They divided the sample of mutual funds into eight sector-based
categories that included Malaysia. Their sample period was from 1997 to 2002
and employing CAPM models, they found no significant difference in
performance to either Islamic or conventional benchmarks. Hayat (2006)
30 Ainulashikin Marzuki and Andrew C. Worthington
during bearish condition (Mansor and Bhatti, 2011b). However, in their recent
work, they investigated 129 IMFs from January 1990 to April 2009 using
panel regression approach and found that IMFs do not outperform market
benchmark (Mansor et al., 2012).
Merdad et al. (2010) investigated the performance of 12 IMFs and 16
CMFs managed by the fourth largest fund managers in Saudi Arabia from
2003 to 2010. They found that IMFs outperform CMFs during bearish and
crises period but underperformed CMFs during bullish market condition. In
addition, IMFs were good at market timing in bearish and crises period while
CMFs were good at the same during bullish period.
Contradict to Merdad et al. (2010), BinMahfouz and Hassan (2012) found
there is no significant different between performance of IMFs and CMF,
regardless of benchmarks used. They used larger sample size, which includes
55 IMFs and 40 CMFs in Saudi Arabia from 2005–2010. Thus, the difference
could possibly due to the larger sample size and more than one fund managers
[unlike the previous research by Merdad et al. (2010)] manage these funds.
Mahmud and Mirza (2011) employed CAPM, Fama–French three factor
and Carhart four factor models to measure the performance of 20 IMFs and 66
CMFs from 2006–2010 in Pakistan. They found these funds underperformed
equity market benchmark. Similarly, this study does not provide statistical test
for the difference in performance between IMFs and CMFs.
Razzaq et al. (2012) evaluated the performance of 9 IMFs in Pakistan
from 2009–2010 using weekly return data. They used Sharpe, Treynor, Jensen
alpha and information ratio. In average, IMFs do not outperform market
benchmark. However, these studies in Saudi Arabia and Pakistan did not
provide statistical test for the difference in performance between IMFs and
CMFs.
In summary (refer to Table 1), there are inconsistencies in the findings and
lack of studies that uses conventional funds as benchmark besides
conventional market portfolios. Even though Abdullah et al. (2007) and
Mansor and Bhatti (2011a, 2011b, 2011c and 2012) compares the performance
of IMFs and CMFs, and also conventional benchmarks, former study did not
include the recent global financial crises while the latter lacks statistical tests
in gauging if there is significant difference in performance between these two
types of funds. These studies, including Hayat and Kraeussl (2011), did not
incorporate comprehensive performance measure including Fama–French
(1996) and Carhart (1997) four–factor models. This calls for more empirical
research that use recent data, longer sample period and comprehensive
performance measures to understand to what extent shariah screening affects
32 Ainulashikin Marzuki and Andrew C. Worthington
mutual fund performance. While the above literature focuses on whether there
are differences in performance between screened and unscreened index/funds,
the following sub section reviews on the impact of screening strategies within
performance of screened funds.
Author (Year) Sample Period No. of funds Performance measures Main findings
Hayat and Global, Msian, 2000– 145 IMFs Risk adj. performance (alpha), IMFs substantially underperform both
Kraeussl (2011) Asia Pacific, 2009 systematic risk (beta) using Islamic and conventional benchmarks
European n Jensen's version of CAPM & 2 IMFs perform worse during financial crisis
Middle East & types of robustness test (in contrast to Abdullah et al, 2007). IMFs
North America Market timing: Treynor & Mazuy are poor market timer using both para and
(1966) and non-parametric nonparametric measures
approach by Jiang (2003),
Downside risk characteristic (Ang
et al., 2006)
Mansor and Malaysia 1990– Unmatched Sharpe, Treynor, Jensen alpha, Both IMFs and CMFs exhibit positive
Bhatti (2011a, 2009 sample of 128 Modigliani Midigliani measure. selectivity and market timing ability. There
2011b, 2011c) IMFs and 350 Information ratio, Adjusted is no significant difference in both abilities
CMFs Sharpe ratio and Treynor and between IMFs and CMFs. IMFs and CMFs
Mazuy performance are similar in bearish and
bullish market condition. No hedging
function showed in IMFs.
Mansor and Malaysia 1990– 129 IMFs Single factor CAPM, Extended IMFs do not outperform single and multiple
Bhatti (2012) 2009 CAPM with multiple benchmarks, benchmarks.
Treynor and Mazuy and extended
Treynor and Mazuy
Razzaq et al. Pakistan 2009– 9 IMFs Sharpe, Treynor, Jensen alpha and On average, IMF performance is poor. They
(2012) 2010 Information ratio are unable to outperform the market
benchmark.
BinMahfouz Saudi Arabia 2005– 55 IMFs and Jensen’s alpha and Fama–French On average, there is no statistically
and Hassan 2010 40 CMFs three factor model significant difference in performance
(2012) between IMFs and CMFs regardless of
benchmark used.
A Review of Performance, Screening and Flows … 35
aspect of religious screening and firm value, Hong and Kacperczyk (2009)
found that excluding portfolios from investing in ‘sin’ stocks may impose
large costs on the portfolio performance. This is because they report ‘sin’
stocks outperform other stocks in their sample.
In the case of IMFs screening, recently Derigs and Marzban (2008)
examined the impact of different Shariah screens on the Islamic funds’ asset
universe and found inconsistencies among the Shariah boards. The same
securities may be halal in one Shariah fund and deemed non-halal in another
Shariah fund. Later, Derigs and Marzban (2009) found that Shariah screening
that employs market capitalization as the denominator provides wider
diversification potential compared to other screening methods. In this study,
the approach is to investigate how these different types of quantitative
screening affect the actual performance of Malaysian IMFs. Put differently,
how the variation in quantitative screenings, defined as screening intensity,
affects the financial performance across IMFs.
Recently, Nainggolan (2011) investigated the impact of Shariah
compliance to performance of IMFs for the period from 2008 to 2009 for
IMFs globally. She applied cross sectional OLS regression and found that for
every one percent increase in total compliance decreases fund performance by
0.01 percent per month. This implied that Shariah screening intensity
negatively affect IMF performance. However, she did not test the relationship
between Shariah screening intensity and risks as well as expenses.
Extending the work of Derigs and Marzban (2008, 2009) and Nainggolan
(2011), this study will add to the literature on the impact of differences in
financial screening to the Malaysian IMFs. This is to give further insight into
whether differences in financial ratio screening generate higher risk adjusted
returns given the wider diversification options by controlling other factors
affecting portfolio performance. While there are studies investigating the
impact of screening strategies to risk and return, literature is silent about the
impact of screening to IMFs expenses.
Sirri and Tufano (1998)], the UK (Keswani and Stolin, 2008), and other
developed capital markets. While there is some evidence relating to the
behavior of SRI or ethical fund investors [see, for example, Benson and
Humphrey (2008), Bollen (2007) and Renneboog et al. (2008a)], studies
dedicated to the behavior of IMF investors are scarce [for example, Nathie
(2009) for IMFs and Peifer (2011) for religious mutual funds]. This section
reviews literature on the determinants of fund flows that comprise financial
and nonfinancial attributes as well as the impact of fund flows to the future
fund performance (‘smart money’).
flows to measure the behavior of investors toward past performance and other
fund characteristics. These measures use information on the net asset values
(NAV) or asset under management (AUM). Only recently, separate
information on actual fund inflows and outflows are available in some
countries that give better insights on this issue (Keswani and Stolin, 2008).
Using a sample of 143 open ended mutual funds from 1965 to 1984,
Ippolito (1992) found that there was an asymmetric relationship between fund
flows and lagged performance. Investors reacted positively to the recent past
good performing funds resulting in higher inflows, but the magnitude of
outflows to poor performance was not strong. Ippolito (1992) argued that this
was likely because of the higher transaction cost where the exit fee charged by
managed funds was more expensive than the cost they incurred when they
purchased.
Gruber (1996) examined the impact of performance and other
performance determinants on money-flows into and out from 227 US mutual
funds over the period of January 1985 to December 1994. He used raw returns,
alpha from the single- and multi-factor asset pricing models as the
measurement of performance and found that investors put more money in past
good performance funds, however, the outflows for past poor performance
were not as pronounced as to past good performance funds. He suggested that
current expenses, raw returns and risk-adjusted returns contain information
about future performance over one and three year intervals.
Sirri and Tufano (1998) employed a piecewise linear model in a sample of
US growth funds from 1971 to 1990. They divided lagged fund performance
into quintiles in their objective category, and found similar findings where
money inflows to lagged recent top performance quintiles were stronger than
money outflows to past poor performance quintiles. They used annual data and
lagged one-year raw returns.
Goetzmann and Peles (1997) used mutual funds data from 1976 to 1988
and divided the performance into quartiles. Using unconstrained regression,
they found that there is a significant different pattern in terms of money
inflow-performance in the top performing quartile compared to the bottom
quartiles. The top quartiles received high fund inflows while there is no
significant outflow for the top performing funds. This implies that reward for
past performers is high cash inflows; however, the market does little in
disciplining poor performers.
Chevalier and Ellison (1997) adopted the semi parametric model to
investigate the curve demonstrated by the relationship between fund flows and
past market-adjusted returns in 1982 to 1992. Concentrating on growth and
A Review of Performance, Screening and Flows … 39
growth-and-income funds, they found that fund flows are more sensitive to
past good performance in younger funds than to older funds. While the shape
of flow-performance is quite steep and almost linear (attract stronger money-
flows in respond to past performance), older funds demonstrated a convex
shape (less sensitive to past poor performance).
Most recently, Del Guercio and Tkac (2008) employed the event study
method to investigate further how money-flows respond to past performance
measured by Morningstar star ratings for the period 1987 to 1994. Five star
rating performers and those funds newly upgraded to five star rating
performers attract significant money inflows. However, response to lower star
rating performers is not as pronounced as those for five star ratings for both
inflows and outflows.
Theoretically, investors are rational, risk averse, favor higher returns for a
given level of risk and favor lower risk for a given level of return.
Nevertheless, the above studies indicate that retail investors are not rational,
and this contradicts the efficient market hypothesis and asset pricing theories.
If retail investors are rational, theoretically they are more likely to invest more
money in top performing funds while withdrawing their money from poor
performing funds. Several studies provide a theoretical explanation why
investors do not strongly react to funds with past poor performance, as
strongly as they react to funds with past good performance. What makes
investors remain in the losing fund? Goetzmann and Peles (1997) proposed a
theory to explain this irrational behavior of retail investors. They suggest
cognitive dissonance as a possible explanation, where retail investors are
reluctant to accept that they bought poor performing funds. As a result, they
continue to keep the funds even if the performance is poor.
From a psychological perspective, Kahneman and Tversky (1979) and
Shefrin and Statman (1985) proposed the disposition effect theory to explain
the irrational behavior of investors who sell winners too soon to realize cash
gains and hold losers as they are unwilling to realize the loss. This behavior
leads to the asymmetric relationship between fund flows and fund
performance. However, whether this is true in emerging markets and for
Islamic investors remains an empirical question.
Another theoretical explanation is investors are not homogeneous.
Heterogeneity across investors makes their response to the poor performing
funds different [see Berk and Tonks (2007) and Harless and Peterson (1998)].
Investors do not process information at the same pace. Some investors respond
to the poor performing funds by immediately taking out the money while
others are very slow to respond.
40 Ainulashikin Marzuki and Andrew C. Worthington
more about social or ethical issues than the risk return characteristics (i.e.,
financial performance).
Relating to screened mutual funds, Bollen (2007) is the first author to
study the behavior of ethical investors in the US. The main issue is how the
behavior of ethical mutual fund investors differs from CMF investors. Since
the ethical mutual fund investors buy ethical mutual funds for both financial
and social objectives, the question is whether the ethical mutual fund investors
will also go for performance when deciding to purchase for ethical investment.
He found that while SRI investors are more sensitive to past good
performance, they are less sensitive to past poor performance compared to
unscreened investors. He suggested several possible explanations. First,
conventional investors may have more options to switch to other funds
compared to SR investors. Second, he suggested that SRI investors consume
nonfinancial attributes that mitigate the withdrawal of funds from the negative
performance. In addition, he investigated the volatility of money-flows in the
SRI fund in comparison to non-SRI funds and found that SRI funds have
lower volatility of money-flows. According to him, SRI investors are more
loyal than CMF investors are.
In a recent study, Renneboog et al. (2011) investigated money-flows of
410 SRI mutual funds from all over the world. Their findings supported the
work of Bollen (2007) in that SRI investors chase past good performance more
than conventional investors, and this is less pronounced for past poor
performance. The study not only investigated the impact of past performance
to net inflows but also simultaneously included other fund characteristics such
as fund size, age, risk, fund family reputation, and fund fees. In addition, SRI
investors are less sensitive to fund expenses and load fees. This study also
investigates the influence of fund characteristics on the variability or volatility
of money-flows. Smaller, younger, and riskier SRI funds have higher money-
flow volatility partly due to high marketing efforts of these funds. Volatility is
even higher in funds that experienced good recent performance and belong to a
larger fund family or to a large family because switching between funds within
the family requires lower cost. As suggested by Bollen (2007), it is interesting
to investigate whether other types of investors who are more concerned about
the non-performance funds’ attributes (similar to ethical investors) exhibit the
same behavior (similar to his findings). Thus, IMF shareholders or investors
are the most suitable candidates for generalization. As IMF investors consider
both religious and financial objective, we expect that IMF investors exhibit
different degrees of reaction to funds past performances compared to
conventional investors.
42 Ainulashikin Marzuki and Andrew C. Worthington
argued that the findings of Bollen (2007), those SRI investors were more loyal,
could be attributed to religious SRI investors.
Beside fund performance, other factors may also influence fund flows.
Mutual fund investors may also consider nonfinancial attributes in making
fund allocation decisions. This is contradictory to the standard modern
portfolio theory. These factors include fund visibility (Sirri and Tufano, 1998),
fund expenses (Barber et al., 2005; Chevalier and Ellison, 1997; Sirri and
Tufano, 1998), fund advertising (Jain and Wu, 2000; Sirri and Tufano, 1998),
investment styles (Cooper et al., 2005; Karceski, 2002), fund size (Chevalier
and Ellison, 1997; Del Guercio and Tkac, 2008; Fant and O'Neal, 2000; Sirri
and Tufano, 1998), fund age (Chevalier and Ellison, 1997; Nanda et al., 2004;
Ruenzi, 2005; Sirri and Tufano, 1998), fund risks (Oh, 2005; Renneboog et al.,
2011) and fund family characteristics (Huang et al., 2007; Nanda et al., 2004;
Sirri and Tufano, 1998).
Investors can select from thousands of mutual funds available in the
market. Financial theory predicts that rational investors do not incur search
costs. However, in reality, investors have limited ability to collect and process
information and search costs are huge for investors. Thus, many investors,
especially unsophisticated retail investors, make decisions based on the
information made available to them that is theoretically incomplete. Thus,
investors make asset allocation decisions based on funds that are more visible
to them. As a result, funds that incur higher advertising costs and appear in the
media and those with higher established reputation attract more funds despite
abnormal performance.
Sirri and Tufano (1998), Jain and Wu (2000), and Barber et al. (2005)
found that mutual fund advertising influences fund flows. Advertising
increases fund visibility and lower search costs, which, in turn, positively
influences the amount of money-flows into the fund. In addition, among all the
funds frequently advertised in influential media, growth funds attract larger
flows. Jain and Wu (2000) reported that funds that are advertised in the
financial magazines perform well prior to the advertisement and these funds
significantly attract larger future money-flows. However, the performance of
these funds does not persist, as the post performances are more unlikely to beat
the market benchmarks. They conclude that fund sponsors use past
performance information to increase assets under their management and,
44 Ainulashikin Marzuki and Andrew C. Worthington
On the other hand, higher expense ratios may also attract more money-flows
into a fund as higher expense ratios may indicate that higher marketing
expenses, which possibly increase fund visibility and consequently flows.
Barber et al. (2005) found that there is a negative relationship between fund
flows and front-end loads but that there is no relationship between fund flows
and operating expenses. This may indicate that investors are sensitive to
information that is visible to them such as front-end load expenses,
commissions, and performance compared to expense ratio. However, this does
hold true in the case of screened funds. For example, investors of SRI funds
care less about fund fees compared to conventional investors (Renneboog et
al., 2011). One potential implication of this behavior is that fund managers
may take this opportunity to invest a large amount of money in marketing to
attract more flows that will finally increase the size of assets under
management and revenue to the fund managers.
Some investors may also consider the return volatility (or total risk) in
their fund selection criteria [see, for example, Huang et al. (2007), Oh (2005),
Renneboog et al. (2011) and Sirri and Tufano (1998)]. While Huang et al.
(2007) and Renneboog et al. (2011) found that there is a negative relationship
between fund flows and return volatility, Oh (2005) found otherwise.
Renneboog et al. (2011) explained that investors realize that fund managers
have an incentive to increase returns volatility to take advantage of investors
behavior of chasing past performance thus they do not select funds with higher
returns volatility. Alternatively, in the case of Korean mutual fund investors,
Oh (2005) found that investors are putting more money into the mutual funds
as they see that return volatility or total risk as an opportunity and accordingly
invest more money into these mutual funds exhibiting these characteristics.
Nanda et al. (2004) provided further insight into the determinants of fund
flows into mutual funds. Besides fund specific characteristics, they found that
characteristics of fund family or fund sponsor also influence money-flows to a
particular fund. For example, performance of other funds also influences the
investors’ decision to invest in a fund within the same family. The authors
termed this phenomenon as fund family’s spillover effect. In addition, Ivkovic
(2002) reported that the performance of other funds in the family also
influences money-flows into the fund as well as the performance of that
particular fund itself. Cash flows are not only sensitive to the past superior
performance of the individual funds but also there is a spillover effect from the
past performance of other funds in the family to that particular fund.
Massa (2003) found that management companies with a higher degree of
product differentiation are more likely to generate low performance. However,
46 Ainulashikin Marzuki and Andrew C. Worthington
5. SMART MONEY
The last decade has seen an increased interest in research devoted to
investigating the behavior of mutual fund investors by examining the money
flow of mutual funds. Numerous empirical studies focus on the relation
between fund flows and past performance where bulk of them focuses on
mutual funds in the US [see Chevalier and Ellison (1997) Ippolito (1989), and
Sirri and Tufano (1998)], the UK (Keswani and Stolin, 2008), and other
developed capital markets. While there is some evidence relating to the
behavior of SRI or ethical fund investors [see, for example, Benson and
Humphrey (2008), Bollen (2007) and Renneboog et al. (2008a)], studies
dedicated to the behavior of IMF investors are scarce [for example, Nathie
(2009) for IMFs and Peifer (2011) for religious mutual funds]. This section
reviews literature on the determinants of fund flows that comprise financial
and nonfinancial attributes as well as the impact of fund flows to the future
fund performance (‘smart money’).
past performer by putting more money in the funds, and punish recent poor
performer by taking their money out of the funds.
In addition, recent studies had found a convex or asymmetric flow-
performance relationship [see Gruber (1996), Sirri and Tufano (1998) and
Ippolito (1989)]. This relationship implies that top performing funds enjoy
huge money inflows (stronger effect to fund inflows) while funds with poor
performance do not suffer huge money outflows (weaker effect to fund
outflows). Majority of these studies uses absolute or normalized implied net
flows to measure the behavior of investors toward past performance and other
fund characteristics. These measures use information on the net asset values
(NAV) or asset under management (AUM). Only recently, separate
information on actual fund inflows and outflows are available in some
countries that give better insights on this issue (Keswani and Stolin, 2008).
Using a sample of 143 open ended mutual funds from 1965 to 1984,
Ippolito (1992) found that there was an asymmetric relationship between fund
flows and lagged performance. Investors reacted positively to the recent past
good performing funds resulting in higher inflows, but the magnitude of
outflows to poor performance was not strong. Ippolito (1992) argued that this
was likely because of the higher transaction cost where the exit fee charged by
managed funds was more expensive than the cost they incurred when they
purchased.
Gruber (1996) examined the impact of performance and other
performance determinants on money-flows into and out from 227 US mutual
funds over the period of January 1985 to December 1994. He used raw returns,
alpha from the single- and multi-factor asset pricing models as the
measurement of performance and found that investors put more money in past
good performance funds, however, the outflows for past poor performance
were not as pronounced as to past good performance funds. He suggested that
current expenses, raw returns and risk-adjusted returns contain information
about future performance over one and three year intervals.
Sirri and Tufano (1998) employed a piecewise linear model in a sample of
US growth funds from 1971 to 1990. They divided lagged fund performance
into quintiles in their objective category, and found similar findings where
money inflows to lagged recent top performance quintiles were stronger than
money outflows to past poor performance quintiles. They used annual data and
lagged one-year raw returns.
Goetzmann and Peles (1997) used mutual funds data from 1976 to 1988
and divided the performance into quartiles. Using unconstrained regression,
they found that there is a significant different pattern in terms of money
A Review of Performance, Screening and Flows … 49
to investigate whether other types of investors who are more concerned about
the non-performance funds’ attributes (similar to ethical investors) exhibit the
same behavior (similar to his findings). Thus, IMF shareholders or investors
are the most suitable candidates for generalization. As IMF investors consider
both religious and financial objective, we expect that IMF investors exhibit
different degrees of reaction to funds past performances compared to
conventional investors.
Beside fund performance, other factors may also influence fund flows.
Mutual fund investors may also consider nonfinancial attributes in making
fund allocation decisions. This is contradictory to the standard modern
portfolio theory. These factors include fund visibility (Sirri and Tufano, 1998),
fund expenses (Barber et al., 2005; Chevalier and Ellison, 1997; Sirri and
Tufano, 1998), fund advertising (Jain and Wu, 2000; Sirri and Tufano, 1998),
investment styles (Cooper et al., 2005; Karceski, 2002), fund size (Chevalier
and Ellison, 1997; Del Guercio and Tkac, 2008; Fant and O'Neal, 2000; Sirri
and Tufano, 1998), fund age (Chevalier and Ellison, 1997; Nanda et al., 2004;
Ruenzi, 2005; Sirri and Tufano, 1998), fund risks (Oh, 2005; Renneboog et al.,
2011) and fund family characteristics (Huang et al., 2007; Nanda et al., 2004;
Sirri and Tufano, 1998).
Investors can select from thousands of mutual funds available in the
market. Financial theory predicts that rational investors do not incur search
costs. However, in reality, investors have limited ability to collect and process
information and search costs are huge for investors. Thus, many investors,
especially unsophisticated retail investors, make decisions based on the
information made available to them that is theoretically incomplete. Thus,
investors make asset allocation decisions based on funds that are more visible
to them. As a result, funds that incur higher advertising costs and appear in the
media and those with higher established reputation attract more funds despite
abnormal performance.
Sirri and Tufano (1998), Jain and Wu (2000), and Barber et al. (2005)
found that mutual fund advertising influences fund flows. Advertising
54 Ainulashikin Marzuki and Andrew C. Worthington
increases fund visibility and lower search costs, which, in turn, positively
influences the amount of money-flows into the fund. In addition, among all the
funds frequently advertised in influential media, growth funds attract larger
flows. Jain and Wu (2000) reported that funds that are advertised in the
financial magazines perform well prior to the advertisement and these funds
significantly attract larger future money-flows. However, the performance of
these funds does not persist, as the post performances are more unlikely to beat
the market benchmarks. They conclude that fund sponsors use past
performance information to increase assets under their management and,
subsequently, their incentive. Agarwal et al. (2003) extended the study on
hedge funds and reported similar findings.
Fund size and fund age may also indicate fund visibility. Large and older
funds have already been on the market and may have an established reputation
compared to smaller and younger funds. Moreover, large funds are able to
spend more on advertising and are more likely to receive media attention.
Nevertheless, Gruber (1996) argues that there is a linear relationship between
money-flows to mutual funds and fund size, where, proportionately, the larger
the fund size the larger the money-flow. Thus, using relative flows as a
dependent variable is crucial to measure the sensitivity of funds past
performance and fund size to fund flows. However, the magnitude of relative
fund flow declines with fund size where large funds tend to attract
significantly smaller relative flows than small funds (Sirri and Tufano, 1998).
Therefore, the inclusion of size effect is necessary in both the regression of the
absolute and relative flows.
Fund age may act as a proxy for investors’ awareness about the fund.
Studies found that older funds have an established reputation, which may be
good or bad depending on their performance realized in the past. Sirri and
Tufano (1998) and Barber et al. (2005) found that smaller and younger funds
attract more fund flows. Higher marketing expenses incurred to market smaller
and younger funds may explain why it attracts more fund flows. This is
because recent fund performance should be more informative for young funds
that do not have such a reputation. Chevalier and Ellison (1997), Nanda et al.
(2004), and Ruenzi (2005) reported that money-flows of young funds are more
sensitive to past performance than those of older funds. Other studies that
investigated the relationship between fund size and fund flows are Fant and
O'Neal (2000) and Del Guercio and Tkac (2008). In addition, flows have a
positive relationship with the size of the management companies and fund
visibility highlighted by the media. However, this type of fund is more likely
to charge higher expenses, possibly due to high marketing expenses.
A Review of Performance, Screening and Flows … 55
Prior literature has reported that the fund fee structure (higher load fees or
fund fees) affects fund performance where they tend to report worse
performance compared to funds that have lower loads and lower expense
ratios (Carhart, 1997). Choosing funds with low fees is advisable rather than
aiming for performance, thus, this may signal to investors to put money in low
fund fees. In other words, funds with lower fees may attract higher money-
flows and funds with higher fees attract lower flows. Studies done by Sirri and
Tufano (1998) and Barber et al. (2005) on conventional funds reveal that
increased fund fees will most likely affect the reduction in the money-flows.
On the other hand, higher expense ratios may also attract more money-flows
into a fund as higher expense ratios may indicate that higher marketing
expenses, which possibly increase fund visibility and consequently flows.
Barber et al. (2005) found that there is a negative relationship between fund
flows and front-end loads but that there is no relationship between fund flows
and operating expenses. This may indicate that investors are sensitive to
information that is visible to them such as front-end load expenses,
commissions, and performance compared to expense ratio. However, this does
hold true in the case of screened funds. For example, investors of SRI funds
care less about fund fees compared to conventional investors (Renneboog et
al., 2011). One potential implication of this behavior is that fund managers
may take this opportunity to invest a large amount of money in marketing to
attract more flows that will finally increase the size of assets under
management and revenue to the fund managers.
Some investors may also consider the return volatility (or total risk) in
their fund selection criteria [see, for example, Huang et al. (2007), Oh (2005),
Renneboog et al. (2011) and Sirri and Tufano (1998)]. While Huang et al.
(2007) and Renneboog et al. (2011) found that there is a negative relationship
between fund flows and return volatility, Oh (2005) found otherwise.
Renneboog et al. (2011) explained that investors realize that fund managers
have an incentive to increase returns volatility to take advantage of investors
behavior of chasing past performance thus they do not select funds with higher
returns volatility. Alternatively, in the case of Korean mutual fund investors,
Oh (2005) found that investors are putting more money into the mutual funds
as they see that return volatility or total risk as an opportunity and accordingly
invest more money into these mutual funds exhibiting these characteristics.
Nanda et al. (2004) provided further insight into the determinants of fund
flows into mutual funds. Besides fund specific characteristics, they found that
characteristics of fund family or fund sponsor also influence money-flows to a
particular fund. For example, performance of other funds also influences the
56 Ainulashikin Marzuki and Andrew C. Worthington
investors’ decision to invest in a fund within the same family. The authors
termed this phenomenon as fund family’s spillover effect. In addition, Ivkovic
(2002) reported that the performance of other funds in the family also
influences money-flows into the fund as well as the performance of that
particular fund itself. Cash flows are not only sensitive to the past superior
performance of the individual funds but also there is a spillover effect from the
past performance of other funds in the family to that particular fund.
Massa (2003) found that management companies with a higher degree of
product differentiation are more likely to generate low performance. However,
fund management companies have an incentive to introduce many new
products to investors. The reason is to attract higher fund flows that bring
more fees to the management companies. In addition, with an assortment of
products, asset management companies are able to compete with the
competitors based on non-performance attributes rather than financial
performance attributes. This is supported by Khorana and Servaes (2004) and
Khorana and Servaes (1999), as product innovations are able to attract more
fund flows to companies and generate a continued growth to the fund families,
especially if the new products have other special features (more differentiated)
compared to the existing products.
Besides fund characteristics as money flow determinants, there are studies
that investigate if money flows follow market returns. Most of these studies
use aggregate mutual fund flows [see, for examples, Warther (1995), Potter
(2000), Luo (2003) Remolona et al. (1997)]. They investigate if mutual fund
investors are feedback traders who move money into (out of) mutual funds
following high (low) market returns. Warther (1995) investigate the relation
between weekly and monthly fund flow and the market returns for the period
1984 to 1992. He finds that mutual fund investors are contrarian instead of
feedback traders, where they move money into mutual funds following
negative market returns. This is in contrast with studies at individual fund
level that find a positive relation between returns and subsequent flows.
Similarly, Luo (2003) find evidence that equity fund investors apply contrarian
strategy when the coefficients of the lagged market returns are significantly
negative for the period 1984 to 1998. The feedback trading strategies only
found among bond fund investors but not equity fund investors. In contrast,
Remolona et al. (1997) found no evidence that market returns have an impact
on equity fund flows for the period 1986 to 1996.
So far no research has investigated the determinants of fund flows in IMFs
and how new money-flows affect the performance of IMFs. This research
contributes to this line of literature. Renneboog et al. (2008a) studied the SRI
A Review of Performance, Screening and Flows … 57
funds globally of which Islamic funds are a subset of the sample. However,
since, according to Forte and Miglietta (2007), IMFs exhibit different
characteristics, the behavior of IMFs investors remain an empirical issue that
needs to be investigated.
CONCLUSION
Based on the review of previous literature, it is evident that there was
scant but growing research on IMFs. In relation to the performance aspect,
there was a lack of consensus among the researchers concerning the reviewed
literature. Past studies indicated that IMFs underperformed the broad market
index, especially in Malaysia.
Concerning the issue of screening, there are inconsistencies as to whether
screening affects performance at the firm level. In addition, mixed results were
found concerning whether screening strategies are able to influence a firm’s
behaviour. At the portfolio level, ample evidence was found that there is no
significant difference in performance between screened and unscreened
portfolio. However, in relation to Islamic funds in Malaysia, earlier studies
found underperformance of Islamic funds not only from the respective
benchmarks, but also from the conventional mutual funds. The literature
review findings also motivate an examination of the impact of screening
intensity rather than simply assigning a dummy variable to differentiate
between screened and unscreened funds. It is interesting to determine how the
screening intensity influences IMFs performance, risk, and expenses.
Finally, the issue of responsiveness of cash flows into screened funds has
received little attention in the Islamic investment literature. Even though there
was evidence in the area of ethical funds, these studies concentrated on the
developed markets where investors were relatively sophisticated. The issue of
IMF flows in emerging countries remains an empirical issue.
REFERENCES
Abderrezak, F. (2008). The performance of Islamic equity funds: A
comparison to conventional Islamic and ethical benchmarks. Master of
International Business and Finance, University of Maastricht, Maastricht,
The Netherlands.
58 Ainulashikin Marzuki and Andrew C. Worthington
Hakim, S. and Rashidian, M. (2002). Risk and return of Islamic stock market
indexes. Paper presented at the Economic Research Forum Annual
Meetings, Sharjah, UAE.
Hakim, S. and Rashidian, M. (2004). How costly is investors' compliance to
Sharia? Paper presented at the Economic Research Forum Eleventh
Annual Conference, Beirut.
Hamilton, S., Jo, H. and Statman, M. (1993). Doing well while doing good?
The investment performance of socially responsible mutual funds.
Financial Analysts Journal, 49(6), 62–66.
Harless, D. W. and Peterson, S. P. (1998). Investor behavior and the
persistence of poorly-performing mutual funds. Journal of Economic
Behavior & Organization, 37(3), 257–276.
Hassan, M. K., Khan, A. N. F. and Ngow, T. (2010). Is faith-based investing
rewarding? The case for Malaysian Islamic unit trust funds. Journal of
Islamic Accounting and Business Research, 1(2), 148–171.
Hayat, R. (2006). An empirical assessment of Islamic equity fund returns.
Master thesis, Free University, Amsterdam, Netherlands.
Hayat, R. and Kraeussl, R. (2011). Risk and return characteristics of Islamic
equity funds. Emerging Markets Review, 12(2), 189–203.
Heaney, R. (2008). Australian equity mutual fund size effects. Accounting &
Finance, 48(5), 807–827.
Heinkel, R., Kraus, A. and Zechner, J. (2001). The effect of green investment
on corporate behavior. Journal of Financial and Quantitative Analysis,
36(4), 431–449.
Hendricks, D., Patel, J. and Zeckhauser, R. (1993). Hot hands in mutual funds:
Short-run persistence of relative performance, 1974-1988. The Journal of
Finance, 48(1), 93–130.
Henriksson, R. D. (1984). Market timing and investment performance: An
empirical investigation. The Journal of Business, 57(1), 73–96.
Henriksson, R. D. and Merton, R. C. (1981). On market timing and investment
performance II: Statistical procedures for evaluating forecasting skills. The
Journal of Business, 54(4), 513–533.
Hillman, A. J. and Keim, G. D. (2001). Shareholder value, stakeholder
management and social issues: What's the bottom line? Strategic
Management Journal, 22(2), 125–139.
Hoepner, A. G. F., Rammal, H. G. and Rezec, M. (2011). Islamic mutual
funds’ financial performance and international investment style: Evidence
from 20 countries. The European Journal of Finance, 17(9-10), 829–850.
A Review of Performance, Screening and Flows … 65
Hong, H. and Kacperczyk, M. (2009). The price of sin: The effects of social
norms on markets. Journal of Financial Economics, 93(1), 15–36.
Huang, J., Wei, K. D. and Yan, H. (2007). Participation costs and the
sensitivity of fund flows to past performance. The Journal of Finance,
62(3), 1273–1311.
Humphrey, J. E. and Lee, D. D. (2011). Australian socially responsible funds:
performance, risk and screening intensity. Journal of Busienss Ethics,
102(4), 519–535.
Hussein, K. A. (2005). Islamic Investment: Evidence From Dow Jones and
FTSE Indices. Paper presented at the International Conference on Islamic
Economics and Finance, Indonesia.
Hussein, K. A. and Omran, M. (2005). Ethical investment revisited: Evidence
from Dow Jones Islamic indexes. The Journal of Investing, 14(3), 105–
126.
Indro, D. C., Jiang, C. X., Hu, M. Y. and Lee, W. Y. (1999). Mutual fund
performance: Does fund size matter? Financial Analysts Journal, 55(3),
74–87.
Ippolito, R. A. (1989). Efficiency with costly information: A study of mutual
fund performance, 1965-1984. The Quarterly Journal of Economics,
104(1), 1–23.
Ippolito, R. A. (1992). Consumer reaction to measures of poor quality:
Evidence from the mutual fund industry. Journal of Law and Economics,
35(1), 45–70.
Ivkovic, Z. (2002). Essays in financial economics. Degree of Doctor of
Philosophy, Yale University, United States.
Jain, P. C. and Wu, J. S. (2000). Truth in mutual fund advertising: Evidence on
future performance and fund flows. The Journal of Finance, 55(2), 937–
958.
Jegadeesh, N. and 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. C. (1968). The performance of mutual funds in the period 1945-
1964. The Journal of Finance, 23(2), 389–416.
Jensen, M. C. (1972). Optimal utilization of market forecasts and the
evaluation of investment performance. In G. P. Szego and K. Shell (Eds.),
Mathematical Methods in Finance. Amsterdam: North-Holland Publishing
Company.
66 Ainulashikin Marzuki and Andrew C. Worthington
Jones, S., van der Laan, S., Frost, G. and Loftus, J. (2008). The investment
performance of socially responsible investment funds in Australia.
Journal of Business Ethics, 80(2), 181–203.
Kahneman, D. and Tversky, A. (1979). Prospect theory: An analysis of
decision under risk. Econometrica, 47(2), 263–291.
Karceski, J. (2002). Returns-chasing behavior, mutual funds and beta's death.
Journal of Financial and Quantitative Analysis, 37(4), 559–594.
Keim, D. B. and Stambaugh, R. F. (1986). Predicting returns in the stock and
bond markets. Journal of Financial Economics, 17(2), 357–390.
Keswani, A. and Stolin, D. (2008). Which money is smart? Mutual fund buys
and sells of individual and institutional investors. The Journal of Finance,
63(1), 85–118.
Khorana, A. and Servaes, H. (1999). The determinants of mutual fund starts.
The Review of Financial Studies, 12(5), 1043–1074.
Khorana, A. and Servaes, H. (2004). Conflicts of interest and competition in
the mutual fund industry. Social Science Research Network Retrieved
from http://ssrn.com on 31 January 2010.
Klapper, L., Sulla, V. and Vittas, D. (2004). The development of mutual funds
around the world. Emerging Markets Review, 5(1), 1–38.
Klassen, R. D. and McLaughlin, C. P. (1996). The impact of environmental
management on firm performance. Management Science, 42(8), 1199–
1214.
Knoll, M. S. (2002). Ethical screening in modern financial markets: The
conflicting claims underlying socially responsible investment. The
Business Lawyer, 57(2), 681–726.
Kok, K. L., Gog, K. L. and Wong, Y. C. (2004). Selectivity and market timing
performance of Malaysian unit trusts. Malaysian Journal of Economic
Studies, 41(1 & 2), 71–85.
Kon, S. J. and Jen, F. C. (1978). Estimation of time-varying systematic risk
and performance for mutual fund portfolios: An application of switching
regression. The Journal of Finance, 33(2), 457–475.
Konar, S. and Cohen, M. A. (2001). Does the market value environmental
performance? The Review of Economics and Statistics, 83(2), 281–289.
Kothari, S. P. and Warner, J. B. (1997). Measuring long-horizon security price
performance. Journal of Financial Economics, 43(3), 301–339.
Kreander, N., Gray, R. H., Power, D. M. and Sinclair, C. D. (2005).
Evaluating the performance of ethical and non-ethical funds: A matched
pair analysis. Journal of Business Finance & Accounting, 32(7 & 8),
1465–1493.
A Review of Performance, Screening and Flows … 67
Lai, M.-M. and Lau, S.-H. (2010). Evaluating mutual fund performance in an
emerging Asian economy: The Malaysian experience. Journal of Asian
Economics, 21(4), 378–390.
Langbein, J. H. and Posner, R. A. (1980). Social investing and the law of
trusts. Michigan Law Review, 79(1), 72–112.
Lee, D. D., Humphrey, J. E., Benson, K. L. and Ahn, J. Y. K. (2010). Socially
responsible investment fund performance: The impact of screening
intensity. Accounting & Finance, 50(2), 351–370.
Lehmann, B. N. and Modest, D. M. (1987). Mutual fund performance
evaluation: A comparison of benchmarks and benchmark comparisons.
The Journal of Finance, 42(2), 233–265.
Lintner, J. (1965). Security prices, risk and maximal gains from
diversification. The Journal of Finance, 20(4), 587–615.
Low, S.-W. (2007). Malaysian unit trust funds' performance during up and
down market conditions: A comparison of market benchmark. Managerial
Finance, 33(2), 154–166.
Luo, D. (2003). Chasing the market? Driving the market? A study of mutual
fund cash flows. Yale ICF working paper.
Luther, R. G. and Matatko, J. (1994). The performance of ethical unit trusts:
Choosing an appropriate benchmark. The British Accounting Review,
26(1), 77–89.
Luther, R. G., Matatko, J. and Corner, D. C. (1992). The investment
performance of UK 'ethical' unit trusts. Accounting, Auditing &
Accountability Journal, 5(4), 57.
Lynch, A. W. and Musto, D. K. (2003). How investors interpret past fund
returns. The Journal of Finance, 58(5), 2033–2058.
Mahmud, M. and Mirza, N. (2011). An evaluation of mutual fund performance
in an emerging economy: The case of Pakistan. The Lahore Journal of
Economics, 16, 301-316.
Malhotra, D. K. and McLeod, R. W. (1997). An empirical analysis of mutual
fund expenses. The Journal of Financial Research, 20(2), 175.
Malkiel, B. G. (1995). Returns from investing in equity mutual funds 1971 to
1991. The Journal of Finance, 50(2), 549–572.
Mallin, C. A., Saadouni, B. and Briston, R. J. (1995). The financial
performance of ethical investment funds. Journal of Business Finance &
Accounting, 22(4), 483–496.
Mansor, F. and Bhatti, M. I. (2011a). The Islamic mutual fund performance:
New evidence on market timing and stock selectivity Paper presented at the
68 Ainulashikin Marzuki and Andrew C. Worthington
Chapter 2
ABSTRACT
This study examines performance of Thai equity mutual funds over
5-year time-periods of investment. A sample of 138 funds managed by
the seventeen asset management companies during the period of 2002-
2007 was analyzed using both the traditional approaches: the Treynor
ratio, Sharpe ratio and Jensen’s alpha and the Data Envelopment Analysis
(DEA) technique. The results suggest that performances evaluated using
the former measures lead to more similar fund rankings compared to
those applying the latter method. For 3-year time-period of investment,
80% of the top ten best funds ranked based on the DEA technique are the
same as those ranked using the traditional measures; however only 40%
Corresponding author: Amporn Soongswang. Graduate College of Management, Sripatum
University, 2410/2 Phaholyothin Rd., Jatujak, Bangkok 10900, Thailand. Tel: +66 2579-
1111 Ext. 3031, Fax: +66 2579-1111 Ext. 3011, Email: amporn.so@spu.ac.th.
74 Amporn Soongswang and Yosawee Sanohdontree
of those for 1-year and 5-year time-periods of investment. Thus, the use
of diverse performance measures rather than time-periods of investment
leads to different fund rankings. Finally, the analyses assert that
performance evaluation measure matters and choosing a measure is
important for ranking of Thai equity mutual funds.
INTRODUCTION
In recent decades, mutual funds have played an important role in financial
markets. As of the end of 2007, the world mutual fund industry managed
financial assets exceeding $26 trillion (including over $12 trillion in stocks),
more than four times the $6 trillion of assets managed at the end of 1996
(Investment Company Institute, 2008, cited in Ferreira, Miguel & Ramos,
2009). The number of mutual funds has grown considerably to more than 66,
000 funds worldwide at the end of 2007, approximately 40.91% or 27,000
funds are equity. Although the growth of the mutual fund industry started in
the US, where the industry plays an extremely important role in the stock
markets, this trend has widely spread to other countries around the world
(Khorana, Servaes & Tufano, 2005).
In Thailand, the mutual fund industry started with the first local closed-
end fund in 1977 with an initial size of only 100 million baht. The fund was
established by the first asset management company, Mutual Fund Company
Limited (MFC). Thai mutual funds have been classified by their objectives
and/or policies. These are equity fund, debt fund and balanced fund; open-
ended fund and closed-end fund; onshore mutual fund and offshore mutual
fund; short-term fixed income fund and long-term fixed income fund; and
other types off mutual funds such as flexible portfolio fund, fund of funds,
warrant fund, property fund, retirement mutual fund and sector fund. Their
numbers and total assets have increased over time from 240 funds and 345.80
billion baht in 1999 to 815 funds and 1,372.87 billion baht in 2007(as of April
27). 138 out of the total of 815 funds or about 5.58% were open-ended equity
funds.
Most mutual fund studies have focused on the use of the risk-adjusted
performance measure as an alternative for individual investors in selecting
investment opportunities. The Sharp ratio is probably the most widely used
Does the Choice of Performance Measure Matter for Ranking …? 75
the Treynor ratio, Sharpe ratio, Jensen’s alpha and the DEA technique. The
correlation coefficients were also computed for analyses for degrees of
correspondence between the results estimated applying different evaluation
methods.
This study makes contributions to the literature in terms of the results for
Thai open-ended equity mutual funds that adds to in this area for emerging
markets. The study also enhances understanding about methods applicable to
funds’ performance evaluation, and asserts that the choice of performance
measure influence the analysis of equity mutual funds in Thailand. Thailand is
an important emerging market in South-East Asia that reduces risk and
increases expected returns, rendering significant diversification benefits for
globally-minded investors (Khanthavit, 2001 and Bekaert & Urias, 1998).
Thus, the results can be an investment guide and applicable fund measure
alternative for both local and foreign individual investors.
This study is organized as follows: Section 1 introduction to mutual funds.
Section 2 reviews the literature of relevant fund studies and studies of fund
performance evaluation measures, addressing both traditional and new
measures. Section3 describes data and presents various methods used for
equity mutual fund analyses in this study. Section 4 includes analyses and
results while the last section provides conclusions of the study.
REVIEW OF LITERATURE
Early studies on mutual funds; see, for example, Jensen (1968) and Sharpe
(1966) support the efficient market hypothesis, but later studies such as
Bergstresser and Poterba (2002), Elton, Gruber and Blake (1996), Goetzmann
and Ibbotson (1994), Hendricks, Patel and Zeckhauser (1993) and Kempf and
Ruenzi (2008) find that past performance of mutual funds can predict future
performance. Studies e.g., Brown and Goetzmann (1995), Chevalier and
Ellison (1997), Ferreira et al. (2009), Grinblatt and Titman (1989), Gruber
(1996), Ippolito (1989), Jensen (1968), Malkiel (1995), Scholz and
Schnusenberg (2008) and Sirri and Tufano (1998) conclude that mutual funds
under-perform the market. Carhart (1997) shows that performance persistence
in his sample can be attributed to a momentum factor; meanwhile Malkiel
(1995) uses a large sample of mutual funds and finds performance persistence
during 1973-1981, but there is no evidence of persistence during 1982-1991.
Wermers (2003) reports that mutual fund returns strongly persist over multi-
year periods, which are inconsistent with those reported by Ferreira et al.
Does the Choice of Performance Measure Matter for Ranking …? 77
al. (2009) evaluate the ranking correlation of the Sharpe ratio with 13
performance measures for a sample of US mutual funds, and suggest that the
choice of performance measure is a relevant issue. Accordingly, Zakamouline
(2010) performs analyses of rank correlations between the Sharpe ratio and
some alternative performance measures, and asserts that the choice of
performance measure influence the evaluation of hedge funds. His argument is
that the Sharpe ratio is an adequate measure of performance evaluation only
where returns have normal distribution, but in the real world, there are several
categories of funds with non-normal shapes. Fund managers change frequently
the portfolio composition and leverage leading to a distribution variation
across time.
In Thailand, given a limited number of studies of equity mutual funds,
these studies have focused on closed-end funds rather than open-ended funds,
even though open-ended funds enable one to track the indexes much better
than closed-end funds (Bekaert & Urias, 1998). Moreover, they have been
restricted to the conventional fund performance measures. Although the
techniques for evaluating fund performance have been in existence for almost
40 years, there is no single technique which can serve as panacea (Agarwal &
Mukhtar, 2010). The professional fund managers do not rely on a single
measure for designing a portfolio; thus using more several and different
methods result in a range of outcomes compared to past studies. This can
increase a variety of choices of investment opportunity for individual investors
and benefit investors as guidelines for applying performance measures in fund
evaluation and ranking. At the same time, there is no doubt that the results
obtained using diverse methods are reliable.
This study evaluates performances of 138 open-ended equity mutual
funds, which were managed by the seventeen asset management companies
based in Thailand, between May 2002 and April 2007; and thus the
performances were analyzed using the Sharpe ratio, Treynor ratio, Jensen’s
alpha and DEA technique. They were then compared to those of the index of
the Stock Exchange of Thailand (SET index) whether the average fund
performance is significantly greater than the market. Correlation between the
results derived from the different measures used in the study was analyzed.
Finally, the returns were used for mutual funds ranking.
Does the Choice of Performance Measure Matter for Ranking …? 79
Treynor Ratio
rp r f
Tp
p
(1)
where Tp is the Treynor ratio, rp the portfolio return, rf the risk-free return and
p the systematic risk.
Sharpe Ratio
rp r f
Sp
p
(2)
where Sp is the Sharp ratio, rp the portfolio return, rf the risk-free return and
σ p the total risk of portfolio.
Does the Choice of Performance Measure Matter for Ranking …? 81
Jensen’s Alpha
J p rp r f p (rm r f )
(3)
where Jp is the Jensen’s measure for portfolio, rp the portfolio return, rf the risk
free return, p the systematic risk and rm the market return.
t
u o y ok
Max E k o 1
(4)
m
vx
i 1 i ik
Subject to:
t
u o y ok
Ek o 1
1 k = 1, 2,…, n
m
vx
i 1 i ik
uo 0 o = 1, 2,…, t
vi 0 i = 1, 2, …, m
for the k th DMU, xik the amount of the i th input for the k th DMU, u o the
weight assigned to the o output, v i the weight assigned to the i input, t
th th
the number of outputs, m the number of inputs and n the number of DMUs.
The inputs of the model are the weighted fees and expenses, systematic
risk and total risk. The outputs are returns, diversification and manager skill.
In Thailand, the appropriate performance benchmarks used to compare
mutual fund returns have been defined by the AIMC. These are the SET index,
which is the most widely used as Thai market benchmark for equity funds, and
82 Amporn Soongswang and Yosawee Sanohdontree
the SET 50, which is also used for equity fund benchmark. However, in this
study the SET index is selected as the performance benchmark.
The net return that an investor achieves in investing in a mutual fund
depends on dividend and capital gain or loss that comes from the change in the
net asset value. Return of the mutual funds and the market in a time-period
were calculated as:
NAVt 1 Div t t 1
( 1) 100
Fund return = NAVt (5)
where NAVt is the NAV at the buying month, NAVt+1 the NAV at the month-
end of a period and Divt →t+1 the amount of cash distributed during the period
to shareholders.
SET t 1
( 1) 100
Market return = SET t (6)
where market return is the return on the SET index, SETt the SET index at the
buying month and SETt+1 the SET index at the month-end of a period.
Risks were estimated as the expressed equation:
2
1 n
Var(r) ri ram
n i 1 (7)
where ri is the return of individual mutual fund and ram the mean rate of
returns.
rp α β rm e p
(8)
where rp is the portfolio return, α the intercept term, β the systematic risk,
rm the market return and e p the error term.
The regressing of systematic risk also provided the value of r2 that gives
the strength of correlation between the fund returns and the market indicating
the diversification.
Does the Choice of Performance Measure Matter for Ranking …? 83
where rp is the portfolio return, rf the risk free return, rm the market return, σ p
the total risk of portfolio and σ m the total risk of the market.
This study finds relationship between the results of performance indexes
calculating the Pearson’s correlation coefficient, which was computed
following the given formula.
cov(X,Y)
ρ X,Y
σ X σY (10)
RESULTS
The following section presents the results of the analyses of performances
of 138 funds. These open-ended equity mutual funds were managed by the
seventeen asset management companies in Thailand between May 1, 2002 and
April 30, 2007. The investment horizons include three time-periods from 1-
year to 5-year horizon. The performances of the funds were evaluated using
different measures and the outcomes are explained in terms of out-performing
or under-performing funds compared to the market. The main issues are the
size and signs of excess returns and whether or not the funds are significantly
out-performed. The Pearson’s correlation coefficient was also computed to
indicate whether or not there is significantly positive and high correlation
between the two results estimated using different measures. Finally, the funds
are ranked as top ten best performers based on different performance
evaluation measures.
84 Amporn Soongswang and Yosawee Sanohdontree
Time period % Out perform Mean Market Std.deviation Std. error t-stat Sig
1-Year 96 0.0033 0.0000 0.0023 0.0002 17.30 0.000
3-Year 91 0.0025 0.0000 0.0022 0.0002 11.07 0.000
5-Year 100 0.0054 0.0000 0.0019 0.0002 24.94 0.000
Does the Choice of Performance Measure Matter for Ranking …? 85
Method 1Y 3Y 5Y
Treynor vs. Sharpe 0.999** 0.989** (N = 99) 0.913** (N = 75)
Treynor vs. Jensen 0.942 **
0.971 (N = 99)
**
0.940** (N = 75)
Treynor vs. DEA 0.040 0.036 (N = 99) 0.068 (N = 75)
Sharpe vs. Jensen 0.947** 0.983** (N = 99) 0.967** (N = 75)
Sharpe vs. DEA 0.041 -0.008 (N = 99) -0.047 (N = 75)
Jensen vs. DEA -0.059 -0.062 (N = 99) -0.076 (N = 75)
**
Significant at 1% level; N = 138 except stated differently in the parentheses.
As can be seen from (1) and (2), the only difference between the two
measures is that the Treynor ratio uses the systematic risk while the Sharpe
ratio uses the total risk of portfolio. Thus, the two measures give the same
results for the various portfolios and different results if the portfolios are not
well diversified (see Noulas et al., 2005, Rao, et al., 2006 and also Ornelas et
al., 2009). This can be explained that the numerators of these traditional
measures are the same and similar for the case of the Jensen’s alpha. The risk-
adjustment used by each measure does not change relative evaluation of the
funds. Finally, it is noted that the correlation between results evaluated using
each pair of the different traditional approaches has the same pattern;
meanwhile that between results evaluated employing the far different
measures: the DEA and each of the traditional methods has the same pattern,
which is different from the former.
Table 6. Top Ten Best Performers Ranked Based on Different Performance Evaluation Measures for 1-year, 3-year
and 5-year Time-periods of Investment; and Comparisons of the Average Performances of Open-ended Equity
Mutual Funds and the Market
1-year 3-year 5-year
Treynor Sharpe Jensen DEA Treynor Sharpe Jensen DEA Treynor Sharpe Jensen DEA
ABSM ABSM BTP ABSM ABG ABG BTP BTP ABG ABG B-INFRA ABG
(Aberdeen) (Aberdeen) (BBL) (Aberdeen) (Aberdeen) (Aberdeen) (BBL) (BBL) (Aberdeen) (Aberdeen) (BBL) (Aberdeen)
B-LTF B-LTF B-LTF AYFSTECH ABSC-RMF ABSC-RMF ABSL ABSC-RMF TVF B-INFRA ABG TVF
(BBL) (BBL) (BBL) (Ayudhya) (Aberdeen) (Aberdeen) (Aberdeen) (Aberdeen) (Kasikorn) (BBL) (Aberdeen) (Kasikorn)
B-INFRA B-INFRA BKA2 B-INFRA ABSL ABSL ABG ABG B-INFRA BTP BTP B-INFRA
(BBL) (BBL) (BBL) (BBL) (Aberdeen) (Aberdeen) (Aberdeen) (Aberdeen) (BBL) (BBL) (BBL) (BBL)
BTP BTP BKA MAX DIV LTF BTP BTP BKA BERMF BTP BKA BKA KPLUS2
(BBL) (BBL) (BBL) (Siam City) (BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (Kasikorn)
BKA2 BKA2 BERMF B-LTF B-INFRA BKA BKA2 OSPD BKA KPLUS BKA2 KPLUS
(BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (Thanachart) (BBL) (Kasikorn) (BBL) (Kasikorn)
BKA BKA B-INFRA AYFLTFDIV BKA B-INFRA B-INFRA TVF BKA2 KKF TVF SCBRM4
(BBL) (BBL) (BBL) (Ayudhya) (BBL) (BBL) (BBL) (Kasikorn) (BBL) (UOB) (Kasikorn) (SCB)
ABSC-RMF BERMF B-SUB BTP BERMF BERMF BERMF B-INFRA KPLUS TDF KPLUS NERMF
(Aberdeen) (BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (BBL) (Kasikorn) (UOB) (Kasikorn) (Thanachart)
BERMF ABSC-RMF BTK ABSC-RMF BKA2 BKA2 ABSC-RMF BKA KKF BKA2 KKF AYFSTECH
(BBL) (Aberdeen) (BBL) (Aberdeen) (BBL) (BBL) (Aberdeen) (BBL) (UOB) (BBL) (UOB) (Ayudhya)
SCBLT3 SCBLT3 SCBLT3 BTK B-SUB B-SUB B-SUB BKA2 TDF APF TDF APF
(SCB) (SCB) (SCB) (BBL) (BBL) (BBL) (BBL) (BBL) (UOB) (UOB) (UOB) (UOB)
ABLTF B-SUB IBP BKA2 TVF AYFTW5 TFEQ B-SUB B-SUB B-SUB APF TDF
(Aberdeen) (BBL) (Primavest) (BBL) (Kasikorn) (Ayudhya) (Kasikorn) (BBL)t (BBL) (BBL) (UOB) (UOB)
Mean
-0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027 -0.0027
SET Index
-0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062 -0.0062
Note: The funds were operated by the Thai asset management companies stated in the parentheses.
Does the Choice of Performance Measure Matter … 87
CONCLUSION
The results show that on average, the performances of Thai open-ended
equity mutual funds significantly out-perform the market for all time-periods
of investment, when measured using the traditional measures. However, when
the DEA technique is used, the analyses of the results suggest that the funds
under-perform the market for all time-periods of investment. Thus, it is
concluded that different measures give different outcomes, due to very low
correlation between the traditional measures and the DEA technique.
In terms of fund ranking assessment, performances evaluated using the
traditional measures result in more similar fund rankings compared to those
applying the DEA technique, perhaps because the traditional fund performance
evaluation methods are based on mean-variance theory (also see Ornelas et al.,
2009 and Rao et al., 2006). The results also suggest that for 3-year time-period
of investment, 80% of the top ten best funds ranked based on the DEA
technique are the same as those ranked using the traditional measures,
compared to 40% of those for 1-year and 5-year time-periods of investment.
Thus, the usage of diverse performance measures rather than time-periods of
investment leads to different fund rankings. Finally, it is concluded that the
choice of performance evaluation methods is important for Thai equity mutual
fund rankings.
This is the first comprehensive study focusing on open-ended equity
mutual funds in Thailand. The study investigates funds’ performances
covering long different investment horizons by using several metrics.
Consequently, this study brings about more variety of outcomes and
comparison with other markets, and finally, contributes to the area of financial
economics providing results that can be not only useful for individual
investors for selecting mutual funds, and for investors as guidelines for
applying performance measures in fund evaluation, but are also important for
them who rank funds based on their past performance to make investment
decisions.
REFERENCES
Agrawal, D. (2007), “Measuring performance of Indian mutual funds,”
Prabhandan Tanikniqui, 1 (1), LNCT-MER, Indore, India.
Does the Choice of Performance Measure Matter … 89
Zheng, L. (1999), “Is money smart? A study of mutual fund investors’ fund
selection ability,” Journal of Finance,” (54), (3), 901-933.
In: Mutual Funds ISBN: 978-1-53610-633-6
Editor: Donald Edwards © 2017 Nova Science Publishers, Inc.
Chapter 3
ABSTRACT
In this paper, an artificial neural network (ANN) and a genetic
programming (GP) approach are both applied in order to predict Greek
equity mutual funds’ performance and net asset value. The back
propagation algorithm is used to train the weights of ANNs while
jGPModeling environment is used to implement the GP approach. The
prediction of both the performance and net asset value of mutual funds is
accomplished through historical economic information and fund-specific
historical operating characteristics. Our study is the first one to compare
the forecasting results of the ANN approach with the results obtained
through GP approach on mutual fund performance prediction. The main
conclusion of our work is that ANN’s results outperforms the GP’s
results in the prediction of mutual funds’ net asset value, while GP’s
*
Corresponding author: University of Patras, School of Business Administration, Department of
Business Administration of Food and Agricultural Enterprises, 2 George Seferis Str.,
Agrinio, 30 100, Greece. Email: dpendara@upatras.gr.
94 K. Pendaraki, G. Ν. Beligiannis and A. Lappa
1. INTRODUCTION
The prediction of mutual funds’ performance is a crucial issue for
investors and financial institutions. In international literature, there is a series
of empirical studies in support to the efficient markets hypothesis that past
performance is no guide to future performance, even though a series of
empirical studies reveal that the relative performance of equity mutual funds
persists from period to period. Grinblatt and Titman (1992), Goetzmann and
Ibbotson (1994), Gruber (1996), Blake and Morey (2000), Carhart et al.
(2002), Jan and Hung (2004), Bollen and Busse (2005), Ferruz et al. (2007),
Kacperczyk et al. (2008), Cremers and Petajisto (2009) and Vidal-García,
(2013), found evidence of performance persistence. On the other hand, Jensen
(1968), Kahn and Rudd (1995), Carhart (1997), Porter and Trifts (1998),
Fletcher (1999), Jain and Wu (2000), Philpot et al. (2000), Hallahan and Faff
(2001), Fletcer and Forbes (2002), Prather et al. (2004), Bilson et al. (2005),
Christensen (2005) and Morey (2005) found only slight or no evidence of
performance persistence. In Greek scholar, there have been conducted a few
studies (eg. Babalos et al., 2007; Drakos and Zachouris, 2007; Giamouridis
and Sakellariou, 2008) regarding the evaluation of the persistence in Greek
mutual fund performance, using parametric and non-parametric tests, also with
controversial results.
Most of the aforementioned conflicting studies use linear models into their
methodological framework and they do not capture the complexity presented
in the data. They used statistical methods which represent some limiting
assumptions such as the linearity, normality and independence among
predictor or input variables which influence the effectiveness and validity of
their results. However, artificial intelligence techniques are less open to the
Mutual Fund Prediction Models … 95
ANNs tried to predict mutual fund’s net asset value, while Indro et al. (1999)
and Roodposhti, et. al (2016) used fund specific operating characteristics in
order to predict mutual fund performance. In the present study, we proceed a
step further from these works by designing GP models in mutual fund
prediction and compared their results with the results obtained from ANN
models, instead of traditional forecasting techniques as the previous ones,
using both economic variables and operating characteristics of mutual funds.
In Greek scholar, many studies on Greek mutual funds’ performance
evaluation based on traditional fund performance measures have been applied.
See for example Handjinicolaou, (1980), Milonas (1999), Philipas (2001),
Sorros (2001), Artikis (2004), Thanou (2008), Koulis et al. (2011), etc.
Moreover, Pendaraki et al. (2003; 2005) and Babalos et al. (2012) evaluate
Greek mutual funds’ performance through multicriteria analysis, Pendaraki
and Spanoudakis (2012) through argumentation-based decision making theory,
Alexakis and Tsolas (2011), Babalos et al., (2012), Pendaraki (2012; 2015)
through data envelopment analysis and Pendaraki and Tsagarakis (2016)
through fuzzy linear regression.
Our paper contributes to the literature on mutual fund performance in two
ways. Our first contribution is that we study for the first time to our knowledge
the prediction of the Greek domestic equity mutual funds’ performance
through ANNs and GP technology. Secondly, our study refers to a
comprehensive analysis of mutual fund performance prediction, using for the
first time both operating characteristics of mutual funds and economic
variables in order to predict funds’ net asset value and performance through
ANNs and GP technology. We investigate which fund-specific characteristics
are important performance predictors and examine the case in which all
available information of macro environment is reflected in mutual funds’
prices.
The rest of the paper is organized as follows: In Section 2, the sampling
and grouping of input data as well as the definition of all input variables used
are presented. In Section 3, the ANN approach concerning the application of
multi layer perceptrons methodology for predicting the performance and the
net asset value of Greek mutual funds is presented. In section 4, the GP
methodology for predicting the performance and the net asset value of Greek
mutual funds is described. Experimental results are presented in Section 5.
Finally, conclusions and future perspectives are discussed in Section 6.
Mutual Fund Prediction Models … 97
2. DATA SET
2.1. Sample
The sample used in this study is provided from the Association of Greek
Institutional Investors and consists of daily data of domestic equity mutual
funds (MFs) over a seven year period. The first six years are used for ANN
and GP training while the last year for model evaluation. Daily returns for all
domestic equity MFs are examined for this seven-year period and are
restricted to only observations with non-missing values. Further information is
derived from the Athens Stock Exchange (ASE) and the Bank of Greece,
regarding the return of the market portfolio and the return of the three-month
Treasury bill, respectively. The economic variables used in the analysis are
derived from International Monetary Fund (World Economic Outlook
Database) and the National Statistical Office of Greece.
For the construction of the predictive models, the examined funds are
classified in three homogeneous groups according to the value of their
performance (return of the fund) and risk (beta coefficient) variables.
Performance, measures the expected outcome of the investment, while risk,
measures the uncertainty about the outcome of the investment. Thus, instead
of treating the examined mutual funds as a homogeneous group, we have
funds with high, medium and low performance (return patterns), and funds
with high, medium and low risk (risk patterns). The purpose of this
discrimination was to establish the differences in the MFs’ behaviour (shifts in
returns and betas), and to investigate how these differences influence the
results of our predicting models.
Figure 1. The general structure of the MLP used in the proposed methodology.
Mutual Fund Prediction Models … 101
Figure 2. The basic steps of the proposed neural network optimization method.
𝑣(𝑢) = 𝑔2 (∑𝐾 𝐿
𝑘=1 𝑤𝑘 ∙ 𝑔1 (∑𝑙=1 𝑤𝑘𝑙 ∙ 𝑢𝑙 + 𝑤0 ) + 𝑤0 ) (1)
change (Haykin 2008). The basic reason why we decided to use the error back
propagation algorithm to train our MLPs is that it is one of the most commonly
used algorithms for ANNs’ modelling applications, as presented in the
literature (Adamopoulos et al. 2001; Vassilopoulos et al. 2007; Huitao et al.
2002; Garcia et al. 2004). So, the main steps of the proposed neuaral network
optimization method are the following, as presented in Figure 2.
Except for that, we have changed the crossover operator used by the
jGPModeling environment. Specifically, we decided to repeat for a number of
times the crossover operation between two parents in order to create more than
two offsprings. In our case, the number of offsprings created by the application
of the crossover operator to a specific pair of parents equals 8. In order to
select the offsprings which are going to be included in the next population we
used the following procedure: At each generation, we create a intermediate
population of models consisting of all trees created by the application of the
crossover operator. The offsprings selected from this intermediate population
in order to be included in the population of the next generation are the ones
having the best fitness function values. In order this procedure to be effective,
this intermediate population has been implemented using a heap. The
flowchart of the proposed GP algorithm used in order to predict the
performance and the net asset value of Greek equity mutual funds is presented
in Figure 3.
Mutual Fund Prediction Models … 105
5. RESULTS
5.1. The Application of the Proposed Techniques
different error measures and success rates are computed for the high, medium
and low return and risk mutual funds’ categories.
Tables 5.2 and 5.3 presents the mean absolute percentage error (MAPE),
the median absolute percentage error (MeAPE), the geometric mean absolute
percentage error (GeMAPE), the maximum mean absolute percentage error
(MaxMAPE) and the standard deviation of mean absolute percentage error
(SdMAPE) of the GEOMEAN and NAV predictions when the explanatory
variables are the operating characteristics of mutual funds and when the
explanatory variables are the economic variables respectively using both ANN
and GP for all risk and return patterns. All these measures refer to the risk of
our models, thus the most preferred models are the ones with the lowest values
on the aforementioned measures.
Tables 5.4 and 5.5 present the percentage in which the estimated values of
mutual funds approximate the actual values over 80%, and the percentage that
the estimated values predict the increase or the decrease of funds’ values for
the next period. The second success rate is developed in order to provide
efficient predictions of the up and down movements of the target values for the
next time period. In other words with this measure it is monitoring if fund
prices are trending upwards or downwards.
Table 5.1. The values of the parameters used for the training and testing
phases of both MLPs
Table 5.5. Success rate when the explanatory variables are economic variables
extent this is consistent with the work of Indro et al., (1999), where the
reduced ANN models show better results when comparisons are made on
value funds taking into account the mean absolute deviation, the standard
deviation of the error and the mean absolute percentage error. According to the
results of Table 5.2 when the explanatory variables are the operating
characteristics of mutual funds, ANN generates better forecasting results than
GP for funds of all return and risk patterns in the prediction of NAV. On the
other hand, GP models are better for the prediction of GEOMEAN in most of
the cases. The results of Table 5.3 showed that when the explanatory variables
are economic variables, GP models are better for the prediction of GEOMEAN
in all the cases. ANN outperforms to GP for the prediction of NAV according
to all error measures except the MaxMAPE and SdMAPE. The same
conclusions stand up according to the results of the success rates of Tables 5.4
and 5.5.
As far as the reduced models are concerned (Table 5.6), the GP is better
for the prediction of GEOMEAN and ANN for the prediction of NAV in
almost all of the cases when the explanatory variable is the operating
characteristics of mutual funds. GP outperforms in all the cases (Table 5.7) for
the prediction of GEOMEAN when the explanatory variables are the economic
ones, while ANN is better for the prediction of NAV according to all error
measures except in a few cases regarding the MaxMAPE and SdMAPE. The
same picture is presenting according to the results of Tables 5.8 and 5.9.
Furthermore, an overall result (Tables 5.2-5.5) is that the operating
characteristics of the examined funds compared to macroeconomic variables
give the most reliable forecasting models using ANN and GP for both
GEOMEAN and NAV. Precisely, they present lower percentage errors and
higher success rates. This is an expected outcome as the mutual funds’
performance is more direct affected by the way that there are managed rather
than the macro conditions of the market. The operating variables cover all
aspects of mutual funds performance and referred to return criteria which
measure the expected outcome of the investment in the mutual funds, and risk
criteria which measure the uncertainty about the outcome of the investment.
Furthermore, this result is not surprising from engineering perspective. The
economic variables which are the 17 inputs in the prediction models are the
same for all the examined funds for a specific year. Thus, the prediction
techniques are trained taking into account variables which have the exact same
values, so that the developed models give the same price to the evaluation of
the examined data set. The same conclusion stands for the reduced models
(Tables 5.6-5.9).
114 K. Pendaraki, G. Ν. Beligiannis and A. Lappa
Table 5.8. Success rate of reduced models when the explanatory variables
are the operating characteristics of MFs
Table 5.9. Success rate of reduced models when the explanatory variables
are economic variables
for mutual fund prediction and can be used successfully for the prediction of
the up and down movements of fund values. It is also clear that the results
depend heavily both on the methodology chosen, and the parameters used.
CONCLUSION
The novelty of our work is two-fold. First, we propose both ANN and GP
for the prediction of Greek mutual fund performance. Second, we adopt in
mutual fund prediction both economic information of the macro economic
environment that influence the fund industry and operating characteristics
which drive performance of mutual funds in order to identify factors which
better influence of fund managers under different economic environments.
Thus, discriminating among various macroeconomic variables that
financial markets could be characterized through specific situations (up and
down market trends) requires a sharper picture of the actual behaviour of
mutual fund managers. Furthermore, models of mutual fund prediction will
help a manager or investor to keep track of a fund’s performance over a
number of years, identify important trends through the examination of
different return and risk patterns, and make safer investment decisions. All
these statements were a strong motivation of the present study.
The present study is beyond the two early works of Chiang et al. (1996),
where a back-propagation neural network was applied to forecast only the
NAV for US mutual funds and of Indro et al. (1999), where an ANN approach
was used to forecast only the performance of three different investment styles
of US equity mutual funds, and the later work of Roodposhti, et al. (2016),
where it was examined the factors that affect mutual fund returns of Tehran
Stock Exchange and compared the predictive power of panel data regression
and ANNs. All three studies concluded that ANN generates better forecasting
results than traditional models. In addition, our study outperforms the previous
ones in comparing the forecasting results of the ANN approach with the results
obtained through GP approach.
The main conclusion of our work is that ANN’s result outperforms the
GP’s result in the prediction of NAV, while GP’s result’s outperforms ANN’s
results in the prediction of GEOMEAN. Furthermore, the operating
characteristics of the examined funds compared to macroeconomic variables
give the most reliable forecasting models for both GEOMEAN and NAV.
Overall, our research study indicate that ANN and GP are useful tools for
mutual fund prediction in emerging markets, like Greece. The direction of
Mutual Fund Prediction Models … 117
APPENDIX
The two prediction variables (dependent variables) of the analysis are
described as follows:
The prediction of mutual fund performance is measured by the geometric
mean of excess return over benchmark (%) and is defined as the geometric
mean of excess return over the return R f of a risk free asset (GEOMEAN).
The geometric mean of fund’s excess return over a benchmark shows how
well the manager of a fund was able to pick stocks. In this analysis, the month
Treasury bill rate is used as a proxy for R f The geometric mean is calculated
t 1
t
The Net Asset Value (NAV) of mutual fund is referred to total property of
the fund in current prices. The prices of bonds, interest, cash, stocks, foreign
exchange, etc. are calculated on a daily basis and then are added up in order to
give the assets.
A brief description of the variables regarding the operating characteristics
of mutual funds (independent variables), used in the analysis follows:
matching the portfolio’s risk to that of the market. Given the standard
deviation of a MF’s excess return over the index I the Modigliani
measure is defined as the ratio R I . The fund with the highest
Modigliani measure presents the highest return for any level of risk.
xv. Another performance measure that is derived from comparing a fund
to its benchmark is the information ratio calculated as the ratio
R R f , where is the standard deviation of the MF’s excess
return over the market portfolio.
REFERENCES
Adamopoulos, A. V., Anninos, P. A., Likothanassis, S. D., Georgopoulos, E.
F. & Beligiannis, G. N. (2001). Genetically Optimized Multi-Layered
Perceptrons for the Prediction of Biomagnetic Signals. Neural Networks
and Expert Systems in Medicine and Healthcare (NNESMED 2001),
Milos Island, Greece, June 20–22, 81–83.
Alexakis, P. & Tsolas, I. (2011). Appraisal of mutual equity fund performance
using Data Envelopment Analysis, Multinational Finance Journal,
15(3/4), 273-296.
Artikis, G. P. (2004). Performance evaluation of the bond mutual funds
operating in Greece, Managerial Finance, 30(10), (2004), 1-13.
Babalos, V., Kostakis, A. & Philippas, N. (2007). Spurious results in testing
mutual fund performance persistence: evidence from the Greek market.
Applied Financial Economics Letters, 3, 103-108.
Babalos, V., Caporale G. M. & Philippas, N. (2012). Efficiency evaluation of
Greek equity funds, Research in International Business and Finance, 26,
317-333.
Babalos, V., Philippas, N., Doumpos, M. & Zopounidis, C. (2012) Mutual
fund performance appraisal using stochastic multicriteria acceptability
analysis, Applied Mathematics and Computation, 218, 5693-5703.
122 K. Pendaraki, G. Ν. Beligiannis and A. Lappa
A C
age, 18, 26, 41, 43, 44, 51, 53, 54, 77 Cairo, 61
algorithm, viii, 93, 99, 100, 101, 102, candidates, 41, 52
103, 104, 107, 108, 126 capillary, 123, 124
arbitrage, 9, 12, 60, 70 capital asset pricing model (CAPM), 6,
Arbitrage Pricing Theory, 12 7, 8, 11, 12, 16, 28, 29, 30, 31, 34,
arithmetic, 118 118
artificial intelligence, 94, 95 capital gains, 118
artificial neural network, v, vii, viii, 93, capital markets, 3, 4, 23, 37, 47, 62
94, 95, 123, 124, 126, 127 cash, 3, 10, 22, 38, 39, 49, 57, 67, 70,
Asia, 3, 27, 34, 76 82, 117
assets, 2, 3, 6, 9, 13, 15, 18, 24, 43, 45, cash flow, 22, 57, 67
54, 55, 69, 74, 117, 118 coefficient of variation, 33, 119
comparative analysis, 125
complexity, 94, 106, 107
B compliance, 32, 36, 64
composition, 78, 126
bankruptcy, 25, 95 constant prices, 98, 120
base, 15, 62, 64 construction, 6, 97, 107, 121
bear market., 9 Consumer Price Index, 98, 121
benchmarking, 125 controversial, 77, 94
benchmarks, 6, 9, 11, 12, 13, 17, 26, 27, conventional mutual funds, vii, 1, 3, 27,
29, 30, 31, 33, 34, 43, 54, 57, 61, 63, 57, 68
67, 81 corporate governance, 24, 32, 58
benefits, 5, 29, 76 correlation, 24, 76, 78, 82, 83, 84, 85,
bias, 9, 16, 19, 26, 27, 59, 91, 123 88, 115
bond market, 66 correlation coefficient, 76, 83
bonds, 2, 14, 62, 79, 117 cost, 11, 13, 17, 23, 24, 28, 38, 40, 41,
bull market, 9, 29 48, 50, 51, 58
130 Index
forecasting model, 113, 116 industry, 2, 3, 4, 16, 18, 20, 24, 58, 65,
foreign exchange, 117 66, 74, 77, 89, 90, 95, 116
formula, 83, 118 inefficiency, 8, 16
fraud, 22 inflation, 120, 121
fund performance, vii, 1, 4, 5, 6, 12, 13, ingredients, 102
14, 15, 16, 17, 18, 19, 20, 23, 26, 27, institutions, 2, 77, 94
32, 36, 37, 38, 39, 43, 44, 47, 48, 50, intelligence, 94, 95, 122
53, 54, 55, 58, 59, 60, 61, 62, 63, 65, interest rates, 14
67, 68, 69, 70, 71, 75, 76, 77, 78, 80, international investment, 64
88, 89, 90, 91, 94, 95, 96, 116, 117, International Monetary Fund (IMF), 4,
121, 122, 123, 124, 125, 126, 127 30, 33, 34, 36, 37, 40, 41, 47, 51, 52,
funds, vii, viii, 1, 2, 3, 4, 5, 6, 8, 10, 11, 57, 97, 121
13, 14, 15, 16, 17, 18, 19, 20, 21, 22, investment, vii, viii, 1, 2, 3, 4, 5, 9, 12,
23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 15, 17, 18, 21, 22, 23, 24, 25, 26, 27,
33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 28, 32, 35, 41, 42, 43, 51, 52, 53, 57,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67,
53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 68, 69, 70, 73, 74, 75, 76, 78, 79, 80,
63, 64, 65, 66, 67, 68, 69, 70, 71, 73, 83, 84, 87, 88, 97, 113, 115, 116, 118,
74, 75, 76, 77, 78, 79, 81, 82, 83, 84, 126, 127
85, 86, 87, 88, 89, 90, 91, 93, 94, 95, investors, vii, 1, 4, 5, 7, 17, 19, 20, 21,
96, 97, 99, 100, 101, 103, 104, 106, 23, 24, 25, 28, 30, 35, 36, 37, 38, 39,
107, 108, 112, 113, 115, 116, 117, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49,
118, 121, 123, 124, 125, 126, 127 50, 51, 52, 53, 54, 55, 56, 57, 58, 59,
63, 64, 66, 67, 68, 71, 74, 75, 76, 77,
78, 88, 92, 94
G Islamic finance, 2, 32
Islamic law, vii, 1, 2
gambling, vii, 1, 24 Islamic mutual funds, vii, 1, 2, 61, 62,
GDP, 98, 120, 121 64, 68, 69
genetic programming, vii, viii, 93, 94, Islamic values, vii, 1
122 issues, 17, 41, 51, 63, 64, 69, 83
gharar, vii, 1
governance, 22, 24, 32, 58, 60, 63, 69
Greece, 93, 97, 116, 121, 126 L
grouping, 96
growth, 3, 12, 14, 15, 21, 22, 38, 43, 46, labor force, 121
48, 49, 54, 56, 60, 63, 70, 74, 89, 90, Latin America, 90
121, 123 law enforcement, 77
guidelines, 24, 78, 88 learning, 22, 27, 101, 102, 106, 107, 126
linear model, 38, 48, 94
liquidity, 11, 20, 60
I litigation, 25
Luo, 46, 56, 67
idiosyncratic, 35
income, 3, 39, 49, 74, 118
independent variable, 117, 120
132 Index
M N
magazines, 20, 43, 54 negative relation, 18, 19, 35, 45, 55
magnitude, 38, 44, 48, 54, 125 Netherlands, 27, 57, 64, 122
Malaysia, 1, 2, 3, 4, 16, 23, 29, 33, 34, neural network, vii, viii, 93, 94, 101,
57, 58, 68, 91 115, 116, 122, 123, 124, 125, 126,
management, viii, 2, 3, 4, 5, 16, 17, 18, 127
25, 35, 38, 42, 43, 44, 45, 48, 52, 54, neurons, 98, 99, 101, 106
55, 56, 60, 62, 64, 66, 69, 70, 73, 74,
78, 79, 83, 86, 89, 90, 95, 115, 127
manipulation, 80 O
manufacturing, 121
market capitalization, 36 omission, 9
market segment, 70 operations, 2, 102
market share, 37, 47 opportunities, 23, 74
market timing, 7, 9, 10, 11, 30, 31, 34, optimization, 61, 101, 102, 105
64, 66, 67, 122, 124 optimization method, 101, 102, 105
marketing, 18, 37, 41, 44, 45, 47, 51, 54,
55 P
materials, 127
matter, 65, 88, 89, 91 Pacific, 3, 27, 34, 59, 70
maysir, vii, 1 Pakistan, 29, 31, 33, 34, 67, 69
measurement, vii, 2, 4, 5, 6, 7, 8, 9, 14, passive benchmark, 6
15, 16, 29, 38, 48, 60, 61, 63, 70, 75, performance appraisal, 121
79, 80, 89, 90 performance benchmarking, 125
media, 43, 44, 53, 54 performance measurement, vii, 2, 4, 5, 9,
meta-analysis, 22, 69 14, 15, 16, 29, 60, 61, 70, 75, 80, 90
methodology, 96, 100, 103, 107, 116, performers, 38, 39, 49, 83, 84, 87
117, 125 poor performance, 16, 37, 38, 39, 40, 41,
Middle East, 3, 34, 61 42, 47, 48, 49, 50, 51, 52, 103
model trees, 104 portfolio, 4, 5, 6, 7, 8, 9, 11, 12, 13, 14,
modelling, 102 15, 20, 21, 24, 28, 29, 32, 36, 43, 53,
models, vii, ix, 6, 7, 8, 9, 11, 13, 15, 16, 57, 61, 63, 69, 74, 77, 78, 79, 80, 81,
29, 30, 31, 38, 48, 80, 94, 96, 97, 99, 82, 83, 85, 90, 95, 97, 119, 120, 126
100, 102, 103, 104, 106, 107, 108, positive relationship, 19, 22, 24, 37, 44,
111, 112, 113, 114, 115, 116, 117, 47, 54
123, 125, 126 prediction models, vii, ix, 94, 97, 107,
Modern Portfolio Theory, 6 112, 113, 115, 126
momentum, 13, 19, 20, 76, 106 predictive accuracy, 125
multidimensional, 95 principles, 102
Multifactor Models, 11 private information, 9
mutation, 102, 103 productive efficiency, 89
mutual fund performance prediction, professional management, 16
viii, 93, 96 profit, 22, 25, 42, 52
Index 133
programming, vii, viii, 93, 94, 102, 103, shareholders, 22, 25, 41, 52, 82
122 Shariah, vii, 1, 2, 4, 23, 28, 29, 32, 36,
propagation, viii, 93, 95, 100, 101, 108, 61
116, 127 Sharpe ratio, viii, 7, 28, 29, 33, 34, 35,
pruning, 104 73, 75, 76, 77, 78, 79, 80, 84, 85, 87,
92
Smart Money, 47
Q social norms, 65
social responsibility, 22, 58, 62
Quran, vii, 1 Social Security, 120
socially responsible investment, viii, 2,
R 3, 61, 66, 70
stakeholders, 21, 22, 24, 25
rate of return, 8, 13, 82 standard deviation, 6, 7, 28, 29, 75, 79,
rational expectations, 58 83, 108, 113, 118, 119, 120
reality, 43, 53 standard error, 100, 101
recombination, 102 stock, vii, 6, 8, 9, 10, 11, 12, 13, 14, 24,
regression, 8, 15, 31, 36, 38, 44, 48, 54, 25, 28, 58, 59, 60, 62, 64, 65, 66, 67,
66, 95, 96, 116 71, 74, 89, 95, 115, 117, 122, 123,
reputation, 18, 22, 41, 43, 44, 51, 53, 54 124, 127
response, 14, 39, 49, 50, 60, 89 stock markets, 74
retail, 39, 43, 49, 53, 122 stock picking skill, 6
revenue, 45, 55 stock price, 12, 25, 95, 117, 122
riba, vii, 1 strategy use, 100
risk, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16, structure, 44, 55, 77, 100, 101, 102, 106,
17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 107, 117
27, 28, 29, 34, 35, 36, 38, 39, 40, 41, style, 12, 15, 27, 43, 53, 59, 60, 61, 64,
42, 45, 48, 49, 51, 52, 55, 57, 62, 63, 70, 71, 77, 116, 125
65, 66, 67, 74, 75, 76, 79, 80, 81, 82, success rate, 108, 112, 113, 115
83, 85, 89, 91, 97, 98, 108, 112, 113, Sunnah, vii, 1
115, 116, 117, 118, 119
risk aversion, 75
T
risk factors, 12, 13, 14, 62
techniques, vii, 2, 4, 7, 29, 63, 78, 94,
S 96, 102, 107, 112, 113, 115
testing, 99, 106, 107, 108, 121
Saudi Arabia, 4, 31, 33, 34, 59, 68 Thailand, 73, 74, 75, 76, 77, 78, 79, 81,
screening strategies, vii, 1, 4, 20, 21, 32, 83, 88, 91
36, 57 time periods, vii, 77
securities, 6, 7, 9, 11, 17, 18, 21, 24, 36, time series, 103, 124, 125
61, 66, 70, 71, 118 training, 97, 99, 101, 106, 107, 108
security selection, 9 Treasury, 16, 97, 117
selectivity, 15, 28, 34, 67, 122, 124 Treynor ratio, viii, 7, 16, 28, 29, 33, 73,
sensitivity, 44, 54, 65 75, 76, 77, 78, 79, 80, 84, 85, 87
services, 2, 3, 18, 80 trust fund, 58, 64, 67
134 Index
turnover, 18
V