Mutual Funds - Donald Edwards PDF

FINANCIAL INSTITUTIONS AND SERVICES
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
Additional books in this series can be found on Nova’s website

under the Series tab.
Additional e-books in this series can be found on Nova’s website

under the eBooks tab.
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.
NOTICE TO THE READER

The Publisher has taken reasonable care in the preparation of this book, 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 in this book. The Publisher shall not be liable for any special, consequential, or exemplary
damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any
parts of this book based on government reports are so indicated and copyright is claimed for those parts
to the extent applicable to compilations of such works.
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.
Library of Congress Cataloging-in-Publication Data
ISBN: 978-1-53610-659-6
Published by Nova Science Publishers, Inc. † New York

CONTENTS
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
The authors of this book provide and discuss new research on

performance measurements, types, and impacts on stock returns of mutual
funds. Chapter One reviews the theoretical and empirical literature relating to
mutual fund performance, screening, and fund flows. Chapter Two examines
performance of Thai equity mutual funds over 5-year time periods of
investment. Chapter Three provides mutual fund prediction models using
artificial neural networks and genetic programming.
Chapter 1 – 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 mutual fund performance evaluation techniques and the
underlying theory. This provides a general understanding of the importance of
performance measurement and various ways to measure mutual fund
viii Donald Edwards
performance. In addition, this chapter also reviews the literature on fund

attributes and their influence on mutual fund performance. In the screening
literature, the review includes the impact of screening and differences in
screening strategies to firm and mutual fund performance. Finally, the chapter
reviews 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 Islamic mutual funds (IMFs) 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 is still scarce and lags behind compared with the literature on the
conventional mutual funds (CMFs) and socially responsible investment (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.
Chapter 2 – 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% 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.
Chapter 3 – 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
Preface ix
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
A REVIEW OF PERFORMANCE, SCREENING

AND FLOWS IN SCREENED MUTUAL FUNDS
Ainulashikin Marzuki1,* and Andrew C. Worthington2

1
Universiti Sains Islam Malaysia, Nilai, Malaysia
2
Griffith University, Queensland, Australia
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
mutual fund performance evaluation techniques and the underlying

theory. This provides a general understanding of the importance of
performance measurement and various ways to measure mutual fund
performance. In addition, this chapter also reviews the literature on fund
attributes and their influence on mutual fund performance. In the
screening literature, the review includes the impact of screening and
differences in screening strategies to firm and mutual fund performance.
Finally, the chapter reviews 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 Islamic
mutual funds (IMFs) 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 is still scarce
and lags behind compared with the literature on the conventional mutual
funds (CMFs) and socially responsible investment (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.
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
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
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.
2. MUTUAL FUND PERFORMANCE

The mutual funds literature is quite extensive and highly concentrated in
the areas of performance and performance persistence. Another extensive
strand of literature discusses the issue of benchmark specification for
comparing mutual fund performance. The concentration of previous studies in
these areas implies the importance of performance measurement to a number
of parties, including investors, fund managers, regulators, policy makers, and
academicians.
Performance measurement in mutual funds is important for several
reasons. First, there is the need to find the most appropriate method for
evaluating the performance of mutual funds and fund managers. This is to
assess whether fund managers have any special ability or skill in providing
superior risk-adjusted returns to investors. Second, performance measurement
is required to justify the high fees paid to active fund managers. Investors pay
a certain amount of fees to gain access to the financial skills of fund managers.
In return, they expect the fund manager to obtain abnormal returns higher than
a passively managed portfolio or index fund. For active fund managers,
performance evaluation is then important to justify their presence and sell their
skills in bringing superior returns to investors for a given level of risk.
For investors, the advantages of investing in mutual funds are the benefits
of diversification, cheaper access to professional investment management,
lower initial investment, and convenience. Generally, the evidence indeed
suggests that fund managers are able to provide diversification benefits to
investors (Jensen, 1968), though, the empirical findings relating to the value
that professional fund managers provide are inconsistent. Indeed, the finding
that actively managed mutual funds underperform relative to the benchmark
has been consistently found since the seminal work of Sharpe (1966) and
Jensen (1968). Even though earlier studies found that there is persistence in
performance [see, for example, Brown and Goetzmann (1995), Grinblatt and
Titman (1992), Hendricks et al. (1993), and Elton et al. (1996)], the recent
literature fails to support superior performance persistence [see, for example,
Carhart (1997), Christopherson et al. (1998), Goetzmann and Ibbotson (1994)

and Gruber (1996)]. The following subsection discusses the development of
asset pricing models supporting the advancement of mutual fund performance
evaluation models.
2.1. Mutual Fund Performance Evaluation
Early literature relied on raw returns to measure security or portfolio

performance. Even though practitioners and academic researchers had
discussed the concept of risk, there was no specific measurement used to
quantify risk. For example in a pioneering study, Cowles (1933) compared
fund performance using raw returns with the returns of general market of
common stocks as a passive benchmark. He found that mutual funds did not
perform better than the benchmark. About thirty years later, Friend et al.
(1962) still used only mean raw returns to measure fund performance. They
compared the annual mean returns of 152 U.S mutual funds from 1953 to 1958
with the mean return of market benchmarks. Similarly, they found
underperformance of mutual funds relative to the passive portfolio. Both
studies examined the performance of mutual funds by considering their returns
but ignore their potential risk.
However, soon after, Harry Markowitz published his seminal work
“Portfolio Selection” and provided the foundation of the so-called the Modern
Portfolio Theory (MPT), the literature began to account for both risk and
return dimensions in their portfolio construction. Markowitz (1952) introduced
a measure of risk and provided insights into the concept of diversification as a
means of minimizing risk for a given level of return or equivalently
maximizing return for a given level of risk. According to Markowitz (1952),
risk is defined as the variability in or the standard deviation of returns and so
adding more assets that are perfectly (negatively) correlated among each other
in a portfolio will progressively diversify away unsystematic risk.
Later, in the 1960s, theories of asset pricing were developed based on the
MPT conceptual framework. Chief among these was the independent
development of the capital asset pricing model (CAPM) by Sharpe (1966),
Lintner (1965) and Mossin (1966). The CAPM framework predicts the
equilibrium expected return of risky and riskless assets, where assets are
priced not only based on expected return but also risk. The performance of an
asset portfolio then depends on its exposure to the market (as a risk factor).
Thus, the selection of assets in the portfolio, also known as stock picking skill,
is equally important. This theoretical model assumes that investors hold a

diversified portfolio, thus, systematic risk or beta is the only relevant
benchmark in measuring portfolio performance. In other words, financial
markets reward investors for assuming only systematic (non-diversifiable or
market) risk, but do not reward investors for unsystematic (diversifiable or
firm-specific) risk. The key assumptions under CAPM are that investors are
risk averse, possess homogeneous expectations about future portfolio
performance, there are no transaction costs and the market is in equilibrium
(Jensen, 1968). With this development, there are a variety of performance
measures and techniques constructed throughout the years. For simplicity, this
study divides these models into four main categories, namely single factor,
market timing, multifactor and other related performance models.
2.1.1. Single Factor Models

The CAPM is a single factor model as it only considers the market as a
proxy for risk. There are three main portfolio performance models developed
from the CAPM framework. These are the Sharpe ratio (Sharpe, 1966),
Treynor ratio (Treynor, 1965) and Jensen’s alpha (Jensen, 1968). The first
performance measure to consider both risk and return was the Sharpe ratio
(Sharpe, 1966). The Sharpe ratio measures portfolio return relative to risk
represented by standard deviation. The model is based on Markowitz’s mean
variance portfolio theory comparing portfolios to the capital market line
(CML). The model is as follows.
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
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)
Both Treynor (1965) and Sharpe (1966) developed relative performance

measures that help to compare the performance of a mutual fund to other
different mutual funds as well as to the market benchmark. Since it is not an
absolute measure, there is no indication as to whether the difference in
performance between two portfolios is statistically significant (Reiley and
Brown, 2006, p. 1049).
The third model and most widely used is Jensen alpha, developed by
Jensen (1968). This model improves on the previous ones as the ability of the
fund manager is captured by the intercept (alpha), which is included in the
traditional equation as follows.
𝑹𝒊,𝒕 − 𝑹𝒇,𝒕 = 𝜶𝒊 + 𝜷𝒊 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 ) + 𝜺𝒊 (4)
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
of the market portfolio as a proxy to represent a portfolio benchmark and

omission of the fund managers’ market timing ability. Due to these
weaknesses, various extensions have been included in Jensen’s alpha. For
example, the development of arbitrage asset pricing theory (APT) (to
overcome the issue of mean variance inefficient of benchmarks), decomposing
CAPM model into security selection and market timing and also incorporating
predetermined information variables into the model (to encounter the critics on
the assumptions of stationarity of expected returns and risk of assets to the
market factor). Next, the study reviews the development and the application of
these models in portfolio performance measurement.
2.1.2. Market Timing Models

According to Fama (1972) and Jensen (1972), investment performance
measurement is about evaluating the fund manager forecasting ability, which
involves stock selection and market timing. Jensen (1972) states that security
selection ability is the ability of a fund manager to select mispriced securities
while market timing ability implies the ability of a fund manager to predict the
general market movement represented by a broad based index. The manager
will react accordingly by increasing the relative volatility of their portfolio in
anticipation of a bull market and reducing volatility in anticipation of a bear
market.
One criticism of the Jensen measure is that it suffers from statistical bias
for a market-timing investment strategy. Empirical findings indicate that
failure to account for the market timing variable in the Jensen model would
cause the measurement to suffer from statistical bias when a fund manager
successfully times the market [see Jensen (1972), Admati and Ross (1985) and
Dybvig and Ross (1985)]. The consequences are that the beta estimation is
biased upwards while the estimate of alpha is biased downwards (Grinblatt
and Titman, 1989a). This implies that managers who are successful at market
timing may obtain negative performance.
Thus, many studies attempt to distinguish the security selection from
market timing in the performance measurement model. Treynor and Mazuy
(1966) enhanced Jensen’s (1968) performance measure to include the market
timing ability of a fund manager. They suggested that if a fund manager has
the ability to time the market, there should be a nonlinear relationship between
fund returns and market returns. The assumption in the market timing models
is that, the manager has private information about future market movements
and that this information will lead the manager to revise his or her portfolio
allocation. Thus, Treynor and Mazuy (1966) proposed a quadratic term to

Jensen’s (1968) model and is written as follows:
𝑹𝒊,𝒕 − 𝑹𝒇,𝒕 = 𝜶𝒊 + 𝜷𝒊 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 ) + 𝜸𝒊,𝒕 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 )𝟐 + 𝜺𝒊,𝒕 (5)
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.
𝑅𝑖,𝑡 − 𝑅𝑓,𝑡 = 𝛼𝑖 + 𝛽𝑖 (𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 ) + 𝛿𝑖 (𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 )𝐷𝑡 + 𝜀𝑖,𝑡 (6)
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
data used by previous studies. However, the duration of the outperformance

was only for a short-term.
Grinblatt and Titman (1989b) proposed another model to account for
market timing ability, namely Positive Period Weighting (PPW). This measure
evaluates the performance of a portfolio by calculating the weighted sum of
the period-by-period excess return. Employing PPW, their findings revealed
no significant difference to the Jensen method, irrespective of the benchmarks
used (Grinblatt and Titman, 1994). On this basis, they concluded that almost
all fund managers fail to exercise a successful market timing strategy.
In case of the Malaysian mutual funds, previous studies failed to find
timing ability among fund managers [see, for example, Kok et al. (2004),
Nassir et al. (1997), and Rozali and Abdullah (2006)], which is similar to the
findings in the developed market discussed above. All of these studies
employed either the Treynor–Mazuy and Henriksson–Merton models. The
PPW model however, is almost impossible to employ for Malaysian managed
funds, as data on the portfolio weights are not easily available or are very
costly to collect.
Nevertheless, Edelen (1999) highlighted that the issue of liquidity
motivated trading faced by fund managers might affect their ability to time the
market successfully. He argued that some active fund managers are genuinely
able to time the market. However, since fund managers have to react to
money-flows in and out from mutual funds, performance appears to be
negative. Accordingly, he proposed modifying the market timing models to
account for liquidity cost. Similarly, Ferson and Schadt (1996) and Ferson and
Warther (1996) documented the effect of fund flows on beta and its movement
to market return. They concluded that heavy fund flows into mutual funds
forces fund managers to trade, and this may result in negative market timing
ability.
2.1.3. Multifactor Models

In view of the weaknesses found in Jensen alpha (arising from CAPM
model), many studies attempted to improve upon the model of securities
returns with the aim of controlling and adjusting better for the risk of the
funds. Roll (1978) fundamentally argued that the Jensen measure is sensitive
to the type of benchmark used as a reference to the market portfolio. Later, it
was demonstrated empirically by Roll (1977, 1978), Jensen (1972), and
Dybvig and Ross (1985) that the performance of the same fund or portfolio
varies according to the benchmarks used. In addition, Roll (1977, 1978)
argued about the inappropriateness of broad-based stock indices to represent
diversified or benchmark portfolios. They asserted that there is no appropriate

benchmark to represent market portfolio to calculate the systematic risk. He
suggested that there may be other factors as well that possibly explain stock
price behavior. These factors include macroeconomic variables (Chen et al.,
1986; Ross, 1976). Momentum (Carhart, 1997; Jegadeesh and Titman, 1993)
and firm size and book-to-market value (Fama and French, 1992, 1993) have
also been proven to help explain cross-sectional variation that affect asset
pricing and thereby risk-adjusted measures of performance.
With the aim of overcoming the weaknesses of CAPM in this respect,
Ross (1976) formulated the Arbitrage Pricing Theory (APT). Drawing on this
basic asset-pricing model, several managed fund performance measures were
subsequently developed. These included work by Connor and Korajczyk
(1986), Lehmann and Modest (1987) and Chang and Lewellen (1984). The
advantages of APT are on its empirical simplicity (simplicity in the
assumptions), no reliance on market portfolio as benchmarks and its equation
is open to other factors that affect portfolio performance (as long as the factor
is significant in explaining the cross sectional differences in asset return).
Unfortunately, the model is difficult to follow due to its arbitrage nature, and it
does not specify a clear rule in identification of the underlying measures to
factor or price risk. Thus, this model is not widely used among academicians
or practitioners. In general, the APT model is as follows:
𝑅𝑖 = 𝐸𝑖 + 𝑏𝑖,1 𝛿1 + 𝑏𝑖,2 𝛿2 +. . . +𝑏𝑖,𝑘 𝛿𝑘 + 𝜀𝑖 (7)
where 𝑅𝑖 is the return on asset i, 𝐸𝑖 is expected return for asset i, 𝑏𝑖,𝑘 is

reaction from asset i’s returns movement in common factor 𝛿𝑘 and 𝜀𝑖 is unique
effect on asset i’s return (Ross, 1976).
While the APT model considers macroeconomic variables to account for
risk factors, Fama and French (1992, 1993, 1996) consider firm specific
factors to account for different investment styles employed by fund managers
in their asset-pricing model. The variables are firm size (a small-minus-big
factor) and book-to-market (a high-minus-low factor) factors. The model
explains whether fund managers are more inclined to invest in small cap to big
cap stocks or value to growth stocks. Literature suggests that small cap stocks
are able to generate a higher expected return compared to big stocks and value
stocks outperform growth stocks. Empirically, both risk factors in addition to
market factors are able to further increase the R squared and explain the cross
sectional variation in stock returns. The model is as follows.
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)
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
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.
2.1.4. Other Related Performance Models

Ferson and Schadt (1996) proposed to extend the unconditional model to
include predetermined information variables with changing economic
conditions. This model assumes that the market is semi strong efficient and
allows for time-varying expected returns and risk on predetermined publicly
available information including interest rates, term spread, default spread and
dividend yields. For example, expected returns of stocks and bonds fluctuate
(Keim and Stambaugh, 1986) and economic information that is publicly
available such as interest rates and stock dividend yields are able to predict
changes in expected returns over time. Ferson and Schadt (1996) argue that
active fund managers do change their trading strategies to take advantage of
any market information by modifying the exposure of the portfolio’s alpha and
beta. This conditional model is as follows:
𝑅𝑖,𝑡 − 𝑅𝑓,𝑡 = 𝛼𝑖 + 𝛽𝑖,0 (𝑅𝑚,𝑡 − 𝑅,𝑡 ) + 𝛿𝑖′ [(𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 )𝑍𝑡−1 ] + 𝜀𝑖𝑡 (10)
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.
Return-based style analysis was introduced by Sharpe (1992) to measure

performance of funds based on the asset allocation comparative to its
benchmark or index. Ample evidence reveals that fund managers’ actual asset
allocation strategies do not match with what they reported or disclosed in the
prospectus (Brown and Goetzmann, 1997). Due to this misclassification,
Sharpe (1992) introduced a performance measure that incorporated the mutual
fund’s investment style using a time-series of historical fund returns. This
technique involves a constrained regression that uses several asset classes to
replicate the historical return pattern of a portfolio. However, the limitation of
this method is that it cannot adequately explain the returns of under-diversified
portfolios such as sector funds (Sharpe, 1992). Empirical works have reported
that aggressive growth funds are able to produce higher returns compared to
other investment styles despite higher expenses [see, for example, Grinblatt
and Titman (1989a, 1993), Grinblatt et al. (1995), Daniel et al. (1997), Davis
(2001) and Chen et al. (2000)].
Other models include stochastic discount factor (SDF) (Chen and Knez,
1996), the inter-temporal marginal rates of substitution-based measure
(Glosten and Jagannathan, 1994) and characteristics based performance
methodologies (Daniel et al., 1997; Pastor and Stambaugh, 2002b; Wermers,
2000). Chen and Knez (1996) proposed stochastic discount factors (SDF) as an
alternative measure for mutual fund performance. Dahlquist and Soderlind
(1999) tested the method using Sweden’s sample of equity mutual funds
market and they found that this method was superior in minimizing errors in
benchmark specification, issue of time variation and pre-determined
information. In another study, Daniel et al. (1997) developed a new
performance measurement innovation in the performance literature, namely,
the characteristic selectivity (CS), characteristic timing (CT) and average style
(AS) measures whereas Pastor and Stambaugh (2002a) introduced another
method, namely, seemingly unrelated assets. Despite various performance
methodologies presented above, Kothari and Warner (1997) argue that many
performance measures are misspecified. Thus, including basic mean returns as
well as mean excess returns together with other risk-adjusted performance
measures ensures robustness of results.
Literature on asset pricing is extensive and evolving. Different measures
have been used to evaluate fund performance over the years. Even though this
development has been encouraging and provides better evaluation methods
compared to traditional measures, this evolution may implicate fair
comparison in fund performance. This is due to a variety of methods and the
different samples used. Despite the many models adopted, the findings
consistently indicate that, on average, fund managers’ ability to add value

through their professional management service is questionable. This is true
either before or after adjustment for expenses.
In relation to the Malaysian mutual fund industry, majority of the studies
adopt relative performance measures (for example Sharpe and Treynor ratios)
and Jensen’s alpha [see, for example, Kok et al. (2004), Nassir et al. (1997),
Abdullah and Abdullah (2009), Low (2007), and Rozali and Abdullah (2006)].
There are only one study recently adopt the Fama-French three-factor and the
Carhart four-factor models [see Hassan et al. (2010).and Lai and Lau (2010)].
Majority of these studies adopt the three-month Malaysian Treasury Bill rate
to represent the risk free rate as this is the shortest Treasury Bill rate available
in Malaysia for evaluating monthly performance of mutual funds in Malaysia
[see, for example, Abdullah et al. (2007), Abdullah and Abdullah (2009), Lai
and Lau (2010), Low (2007), and Rozali and Abdullah (2006)]. This study
extends the literature of Islamic mutual fund performance using the most
widely used performance models – CAPM and Fama–French three-factor
model.
2.2. Criticisms of Mutual Fund Performance

Measurement Practice
There are two main criticisms of mutual fund performance measurement

practice, which are survivorship bias and benchmark efficiency. Survivorship
bias means failure to include dead funds in the sample of observation when
measuring stocks or mutual fund performance. Brown et al. (1992), Brown and
Goetzmann (1995) and Malkiel (1995) emphasize that the results of many
performance evaluation studies are biased upward because dead funds are not
present in the sample chosen as dead funds tend to have poor performance.
The results are even more likely to be affected by survivorship bias if the
period of observation is longer (Elton et al., 1996). In addition, data providers
such as Morningstar Inc. and Lipper Inc. are only interested in reporting
existing funds, thus, they do not report dead funds in the database.
The issue of benchmark inefficiency in Jensen’s performance measure has
attracted arguments and criticisms from academicians [see, for example, Roll
(1978) and Elton et al. (1993)]. For example, Ippolito (1989) using a single
factor model found abnormal performance in managed funds. Later, Elton et
al. (1993) challenged this finding when he found that the result reversed after
adding more factors into the model.
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
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.
2.3. Fund Specific Characteristics and Performance
This section reviews literature on the determinants of mutual fund

performance. Besides fund risk and return characteristics, many other studies
attempt to determine whether fund specific characteristics are able to explain
fund performance. The types of fund characteristics discussed in the literature
include fund age, size, fees, past returns and fund flows.
Theoretically, older funds that have been in existence in the market for a
longer period outperform younger funds. The rationale behind this proposition
is that experienced fund managers with superior ability may manage older
funds. However, many studies consistently found that younger funds perform
better than older funds. For example, Malhotra and McLeod (1997) found that
age has a negative relationship with fund returns. Otten and Bams (2002), and
Heaney (2008) supported the negative relationship between these variables.
One possible justification is that a younger manager who is relatively new in
the investment industry is assigned to manage younger funds. As reputation is
important, younger managers will strive for abnormal returns compared to
older fund managers who have reached relative stability in the career.
The number of assets under management reflects the size of funds and
may influence fund performance. Small funds are ‘easy’ to manage compared
to larger funds. As the size of mutual funds becomes larger, fund managers
may have difficulty in moving their assets in and out from securities. On the
other hand, larger funds may be more efficient and enjoy economies of scale.
Empirically, several authors found no relationship between fund size and
performance [see, for example, Carhart (1997), Ciccotello and Grant (1996),
Droms and Walker (1994), Grinblatt and Titman (1994), Bird et al. (1983),
Gallagher (2003) and Dahlquist et al. (2000)]. However, Chen et al. (2004),
and Indro et al. (1999) found that fund performance deteriorated as the funds
become larger.
Fund managers sell their skills and services for fees. The fees are included
in the management expenses and charged annually. In addition, funds incur
marketing and distribution costs, which are charged once as initial charges or
annually as ongoing charges as part of the management expenses. Other
expenses in running the funds include turnover, taxation, brokerage fees, and
trustee fees. Recently, studies have focused on the relationship between

expenses and performance to ascertain whether fund managers add value.
Earlier studies did not find any relationship between fund performance and
expense ratio (Carlson, 1970; Droms and Walker, 1994; Ippolito, 1989).
Recently, Carhart (1997) reported that there is a negative relationship between
these variables where funds with low expense ratios exhibit higher abnormal
returns compared to those with a high expense ratio. Elton et al. (1996), Golec
(1996), Prather et al. (2004) and Gruber (1996) argued that high expense ratios
cause underperforming funds. For example, Prather et al. (2004) investigated
the relationship between expenses and fund performance in the US between
1996 and 2000. Using a multi-factor model, they found that expenses drive
down mutual fund performance. Gruber (1996) found that the fund expense
ratio negatively affects risk adjusted return. He argued that active management
adds value to investors. However, the expenses charged are more than the
value they add. Interestingly, Klapper et al. (2004, p. 4) and Gruber (1996)
argued that while performing funds might imply superior managerial skill,
funds managed by these managers do not increase revenues by charging higher
fees. Instead, they benefit from increased fees, which result from increased
fund size. On the other hand, there are studies that reported a positive
relationship between fund expenses and fund performance (Chen et al., 2004;
Droms and Walker, 1996). The rationale is that fund managers have superior
ability to obtain a higher return and, thus, charge higher fees.
Despite quotes in mutual fund prospectuses that past returns do not
indicate future returns, most investors make asset allocation decisions based
on a fund’s past return information. Whether fund performance persists
(predictor of future performance) is debatable as earlier literature reported
persistence is present in mutual funds [see, for example, Brown and
Goetzmann (1995), Carhart Carhart (1997), Jain and Wu (2000), Grinblatt and
Titman (1992), Hendricks et al. (1993)]. Brown and Goetzmann (1995)
examined the US equity funds from 1976 to 1988 using raw and risk-adjusted
returns. They found that performance persistence existed, but only among poor
performing funds not outperforming funds. Carhart (1997) investigated a free
survivor bias sample of US diversified equity funds from 1962 to 1993. He
found that evidence of performance persistence might be due to the
momentum effect. He demonstrated that previous performing funds exhibit
higher raw and risk-adjusted returns compared to the previous year’s
underperforming funds. However, it was only for a short period of one year.
Further analysis revealed that the persistence of superior performance
among funds is attributed to size and momentum factors where previous
winner funds tend to hold in the portfolio. In addition, he found persistence

exists for poor performing funds. Funds that were last year losers would
continue to be losers this year. By examining advertised funds in the business
magazines, Jain and Wu (2000) provided evidence that there was no
persistence in fund performance. Advertisements highlighted funds with past
abnormal performance. However, the abnormal performance disappeared after
the advertising period. These studies conclude that past good performance
does not predict future good performance but persistence exists for previous
year losing funds.
In addition, fund flows may also contain some information that predicts
future fund performance. This may be true for sophisticated investors as they
are more likely to trade heavily mutual funds. Gruber (1996) and Zheng
(1999) found that funds that received more money-flow would subsequently
outperform funds that were losing money. They term this as the ‘smart money’
effect, as new investors are smart and are able to predict future fund returns.
However, the effect is temporary and largely due to momentum (Zheng, 1999).
These studies concluded that past performance could somehow predict future
performance. Only new investors are aware of this and enjoy net positive
returns while existing unit holders continue to earn lower returns. However,
there are studies that found funds that received high money-flow would
subsequently suffer underperformance. Berk and Green (2004) suggested a
theoretical explanation that this phenomenon may be due to liquidity
motivated trading and decreasing return to scale.
In summary, the researcher found that fund specific characteristics are
important to explain the differential in performance across mutual funds.
Investors may make fund selection decision based on these characteristics.
While evidence from the developed countries and for CMFs is substantial,
little is known whether these findings are consistent in the emerging markets
or in the case of IMFs. This warrants further empirical works.
While the fund managers’ ability to bring above average risk-adjusted
returns remains in question, it does not stop investors from putting money into
mutual funds. The emergence of new product innovations such as SRI funds,
ethical funds and Islamic funds provide more flavor to the mutual fund
industry. These products even limit the diversification potential of funds with
screening strategies that theoretically may affect the performance. It is
interesting to investigate their relative performance to the conventional funds.
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.
3.1. Screening, Firm Performance and Behavior
This section reviews several theoretical and empirical papers relating to

the impact of screening on firm performance. With the growth of SRI and
ethical funds, it is interesting to ask whether these screening strategies are able
to influence firm behavior. This fits with two essential objectives for SRI and
screening more generally. According to de Colle and York (2009), the main
objectives of SRI movements are: i) to align investor’s ethical or moral value
with their financial or investment decisions, and ii) to encourage companies to
act in accordance with these values so as to deviate from the modern financial
theory of shareholder wealth maximization. Negative screening screens out
unethical securities from the portfolio formation. It is interesting to consider
whether this screening is able to influence or promote ethical corporate
behavior. Do firms really want to be included in the SRI or ethical portfolio?

Alternatively, could it be that the benefit of SRI is only a ‘feel good’ sentiment
created by not being involved in unethical corporate behavior?
Theoretically, under neoclassical economic theory, the main objective of
the firm is to maximize shareholder wealth. Specifically, Friedman (1970, p.
32) asserted that:
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.
This implies that economic agents concerns about maximizing their

wealth and investment decisions are mainly on risk return characteristics.
However, with the increasing growth of screened funds it implies that
managers and business entities hold responsibilities beyond the financial
objective. This new responsibility supports the stakeholder theory, which
explains the importance of pursuing both financial and social objectives and
which opposes the stockholder theory by Friedman (1970).
There are several theoretical arguments that support the positive
relationship between corporate financial performance (CFP) and corporate
social performance (CSP). Specifically, Orlitzky et al. (2003) identified four
possible theories to validate this notion. First, lies in the stakeholder theory
that business should not only consider shareholders, but also needs to manage
all stakeholders within the environment in which they operate. Having a
favourable relationship with all stakeholders may drive positive financial
performance. Second, the slack resources theory predicts that companies with
good financial performance subsequently practice high corporate social
responsibility. The justification is that companies with good financial
performance generate rich cash flows, which the companies use to implement
corporate social strategies. Third, the internal resources/learning perspective,
which predicts that companies practicing corporate social performance, may
operate efficiently. Finally, the reputation perspective predicts that exercising
good governance and ethical values generates good reputation for the company
and this leads to greater goodwill as well as improved financial performance.
Based on the meta-analysis conducted by Orlitzky et al. (2003), they found
that there is a positive relationship between CFP and CSP.
The answer to whether ethical screenings are able to influence firms’
behavior lies in several theoretical arguments. First, based on the work of
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
year to year. However, it is interesting to note that the list of Shariah

compliant companies on the Malaysian stock market is quite liberal. This is
because it is based on the SAC guidelines issued by the Securities Commission
of Malaysian (SCM) for which the quantitative screening is lenient.
Many studies in the US have investigated whether SRI screens have a
positive impact on firm value. They examine the implications of applying
social screens such as a corporate governance screen [see, for example,
Gompers et al. (2003), Cremers and Nair (2005), Bauer et al. (2004) and
Claessens (1997)], environmental screen [see, for example, Klassen and
McLaughlin (1996), Dowell et al. (2000) and Konar and Cohen (2001)] and
stakeholder relation screen [see, for example, Hillman and Keim (2001),
Orlitzky et al. (2003) and Renneboog et al. (2008a)]. In general, all the studies
above conclude that maximising stakeholders’ value or being socially
responsible adds value to the firms. This implies that there is positive link
between social and financial performance. While there are studies that
reported positive relationships between these variables, several other studies
found a negative linkage. The following section reviews the impact of
screening at the portfolio level.
3.2. Screening and Portfolio Performance
There are several reasons to believe that screened funds may

underperform conventional funds. First, screened fund portfolios deviate from
the MPT developed by Markowitz (1952), which objectively selects funds on
the correlation of risk-return and assets. Ethical investors consider social and
ethical value as important as financial objectives. In doing so, they employ
negative screening, which limits the number of securities available to form
screened portfolios and, consequently, severely affects its potential
diversification. The screened portfolio may carry high risk, as unsystematic
risk or diversifiable is not diversified away completely (examples include,
Knoll, 2002; Langbein and Posner, 1980). Specifically, screened funds tend to
eliminate or exclude stocks of specific industries from the investment
portfolio. Additionally, screened funds avoid ‘sinful’ industries from their
portfolio such as gambling, liquor, tobacco, and armaments. Empirical
evidence shows that these industries provide higher return and that these
stocks performed across economic cycles (Hong and Kacperczyk, 2009;
Visaltanachoti et al., 2009).
Second, a screened portfolio selects companies that are potentially

expensive as these companies incur unnecessary cost to please stakeholders
other than shareholders. These companies may be attractive because of their
social performance but they have potentially inferior financial returns
compared to other companies in the sector. Unlike screened portfolios,
unscreened ones may invest in any company that performs well financially.
Hence, unscreened funds would generate higher returns than screened funds.
Finally, screened funds require additional costs to search for companies that
comply with both financial and social criteria. As searching for this
information is costly, this brings additional expenses for screened funds. As
unscreened funds do not incur this cost, they tend to have better returns than
screened funds.
On the other hand, the opponents of SRI or screened funds argue that SRI
or screened funds may outperform unscreened funds. Richardson (2007)
provides several justifications for this argument. First, SRI or ethical
investment requires more in depth analysis, which may filter for management
quality. This ensures that socially screened portfolios are able to provide
investors with better stock selection and offer higher risk adjusted return in
comparison to unscreened funds. Companies with management that are
concerned with social and ethical values also have good management and
entrepreneurial skills to generate profit and return to shareholders. This makes
the company both financially as well as socially attractive for investment.
According to Alexander and Buchholz (1978), the management of SRI firms
are not only concerned about financial return but also social performance
because they believe in the stakeholder theory in which their responsibility is
not only to their shareholders but also to other stakeholders around them.
Other authors that share similar thoughts are Dillenburg et al. (2003),
Renneboog et al. (2008b).
Second, companies that avoid good social and environmental practices
will affect its social and environmental performance negatively. As the market
is efficient, any announcement on this issue will reflect in its stock price and
finally the investors share drops. Having a SRI policy in place could identify
any potential risk such as future litigation and potential bankruptcy or
liquidation. Recent studies indicate that SRI portfolios benefit in risk reduction
through careful stock selection where companies not only need to perform
well financially but are also ethically and socially sound (Richardson, 2007).
3.2.1. Empirical Evidence in SRI

The above arguments are supported by the mounting evidence of
empirical research findings in SRI which are mixed and inconclusive [see, for
example, Hamilton et al. (1993), Luther et al. (1992), Luther and Matatko
(1994), Geczy et al. (2005), Gregory et al. (1997), Mallin et al. (1995),
Kreander et al. (2005), Schroder (2004), Bauer et al. (2006), and Bauer et al.
(2007)]. For example in the U.S, Hamilton et al. (1993) examined 17 SRI
funds established prior to 1985 that outperformed traditional mutual funds of
similar risk for the period 1986-1990. Findings indicated, however, that there
is no statistical difference in terms of performance between socially screened
and unscreened mutual funds.
The first evidence of SRI fund performance in the UK was documented by
Luther et al. (1992). They studied 15 ethical funds for the period from 1984 to
1990 and found no strong evidence that ethical funds outperform the market
benchmarks, FT All Share Price Index or MSCI World Index. In addition, they
provided some evidence that ethical fund portfolios tend to concentrate on
small capitalization and low dividend yield companies compared to
conventional funds.
Luther and Matatko (1994) conducted a study on nine ethical funds
against two benchmarks – FT All Share Price Index and Small-Cap Index from
1985 to 1992. The findings confirm the previous conjecture that the asset
allocation of SRI funds is skewed to the small-cap companies. Thus,
comparing ethical funds with a small company index as a market benchmark
provides better results compared to a wide market benchmark. The results
show that SRI funds underperform the market benchmark as during the period
the small cap companies widely underperform the market benchmark.
However, when comparing between conventional and SRI funds,
performances of SRI funds are similar to the conventional funds despite the
constraints imposed on it.
Mallin et al. (1995) used a slightly larger sample from 1986 to 1993. Their
study improved the problem of benchmark misspecification by the earlier
studies by comparing ethical and non-ethical funds adopting matched-pair
analysis. In matching these groups of funds based on the fund size and fund
age, they found there was weak evidence that SRI funds underperformed
conventional funds. Nonetheless, when compared against market benchmarks,
both funds underperformed.
To further investigate the effect of small-cap bias, Gregory et al. (1997)
examined 18 ethical mutual funds for the period of 1986 to 1994. Building on
the works of Mallin et al. (1995), they added investment objective as another
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
conventional funds. Interestingly, the underperformance of SRI funds was not

because of the investment strategies.
An example of the studies that support underperformance of SRI funds is
the one conducted by Geczy et al. (2005). They examined the diversification
cost of SRI investors by constructing a SRI portfolio using the Bayesian
framework. They assumed that the performance of SRI funds depended on the
prior belief of SRI investors about managerial ability and the use of the asset-
pricing model in fund selection. If investors do not believe in managerial stock
selection skill and rely on CAPM, the performance of SRI funds is as good as
conventional funds. However, if investors use the four-factor model and
strongly believe that fund managers have stock selection skill, the
underperformance of SRI funds is significant.
3.2.2. Empirical Evidence in Islamic Investments

Similar to SRI fund literature, literature in Islamic investment
performance also generally found mixed results while some studies reported
no significant differences between performance of Islamic and non-Islamic
indices (Albaity and Ahmad, 2008; Girard and Hassan, 2005; Hakim and
Rashidian, 2002; Hakim and Rashidian, 2004), other studies found
outperformance of Islamic over non Islamic indices (Hussein, 2005; Hussein
and Omran, 2005).
Hakim and Rashidian (2002) compared the performance of the Dow Jones
Islamic (DJI) US index with the Wilshire 5000 index from 1999 to 2002. They
found that DJI index is less risky (standard deviation) than the Wilshire 5000
index. Using the Sharpe ratio, the Islamic index outperformed the conventional
index. They suggested that the Shariah screened index presents unique risk-
return characteristics that are not affected by the broad equity market. In a later
study, Hakim and Rashidian (2004) examined the extent of Shariah compliant
(DJIMI) index correlated with the Dow Jones World (DJW) index and Dow
Jones Sustainability (DJS) index from 2000 to 2004. They found that the
standard deviation and Treynor ratio of DJIMI are similar to those of DJW, but
DJW underperformed DJS.
Girard and Hassan (2005) examined the performance of Islamic and non-
Islamic indices using a variety of measures such as Sharpe, Treynor, Jenson,
Carhart (1997) four factor model, Fama’s selectivity, net selectivity and
diversification for the sample period from January 1996 to November 2005.
They found that there is no difference between Islamic and non-Islamic
indices. They further found that the Islamic indices outperformed from 1996 to
2000 and underperformed from 2001 to 2005 their conventional counterparts.
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)
investigated the performance of 59 IMFs globally from 2001 to 2006 and he

found that IMFs do not significantly under or outperform their Islamic and
conventional market benchmarks under normal market conditions. However,
IMFs outperformed both market benchmarks in bear markets (2002).
Abderrezak (2008) investigated the performance of IMFs by comparing them
to three types of benchmarks mainly Islamic, conventional and ethical
benchmarks. Using CAPM and three factor models, he found that IMFs
underperform conventional and Islamic indices. Using matched-pair analysis,
there was no significant difference between Islamic and ethical funds.
However, both studies used old data sets with Abderrezak, using a data set
from 1997 to 2002.
Hayat and Kraeussl (2011) examined a sample of 145 IMFs for the period
from 2000 to 2009. They found that IMFs underperformed Islamic and
conventional market benchmarks. The underperformance of IMFs became
worse during the recent global financial crisis in 2008/09, which in contrast to
the earlier study by Abdullah et al. (2007). They suggested that investors are
better off to buy index-tracking funds.
Abdullah et al. (2007) investigated the performance of IMFs and
compared it to CMFs. Using a data set from 14 IMFs and 51 CMFs between
1995 and 2001, they found that both IMFs and CMFs suffer from lower
diversification, which affect their performance. Even though, IMFs exhibit
lower diversification, the difference is not significant. They used seven
performance models including CAPM, Modigliani and Miller (MM) and
Information ratio (IR), they found that Islamic funds performed better than
conventional funds when the market was down, and performed worse, when
the market was in uptrend. This unique feature of Islamic funds is good news
to investors as it may act as a hedging mechanism. In relation to market timing
ability, both funds are poor in timing the market. They also found that both
IMFs and CMFs managers do not exhibit any market timing ability.
There are studies by Mansor and Bhatti (2011a, 2011b, 2011c) and
Mansor et al. (2012)], on the performance of Malaysian IMFs. They
investigated 128 IMFs and 350 CMFs from January 1990 to April 2009 and
consistently found that there is no significant difference in performance
between IMFs and CMFs with both funds outperform market benchmark
represented by KLCI. They also found that both IMF and CMF managers have
similar market timing ability where both are able to time the market
successfully (Mansor and Bhatti, 2011a). Their study also found that both
IMFs and CMFs behave similarly during bearish and bullish market where
these funds outperform the KLCI during bullish and underperform market
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
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.
3.3. Screening Strategies and Portfolio Performance
In view of various types of screening strategies available for SRI fund

managers to apply, some studies investigate how these differences of
screening strategies affect performance across screened funds [see, for
example, Goldreyer et al. (1999), Barnett and Salomon (2006), Renneboog et
al. (2008a), Lee et al. (2010) and Humphrey and Lee (2011)]. Examples of
these screening strategies are environmental, social, shareholder activism and
corporate governance screening. As fund managers may also apply more than
one type of screening strategies in constructing their portfolios, recent
literature has investigated how the number of screening strategies (screening
intensity) influence performance within screened funds [examples include
Renneboog et al. (2008a), Lee et al. (2010) and Humphrey and Lee (2011)].
In relation to Islamic investments, the most obvious screening strategy is
exclusionary screening. As this exclusionary screening fundamentally derives
from Shariah, fund managers have no option to be selective in including or
excluding types of companies from the investment universe. In Islamic finance
literature, there are only two studies that investigated the impact of different
screenings on portfolio performance [see, for examples, Derigs and Marzban
(2008) and Nainggolan (2011)]. Derigs and Marzban (2008) investigated the
impact of different Islamic financial screening on the investment universe of
Islamic funds and Nainggolan (2011) investigated how Shariah compliance
influences IMFs performance.
Goldreyer et al. (1999) compared the performance of 49 SRI funds to 20
randomly selected samples of unscreened funds from 1981 to 1997. Using
Sharpe, Treynor and Jensen alpha, they found that there was no significant
difference in performance between the two types of funds. However, they
found that SRI funds that adopted positive or inclusionary screening
performed better than SRI funds that only employed negative screening. Even
though the study suffered from a small sample size, it provided evidence that
types of investment screening affect the performance of screened funds.
Table 1. Selected IMF performance studies
Author Sample Period No. of funds Performance measures Main findings

(Year)
Abdullah Malaysia 1992– Unmatched Sharpe ratio, Treynor ratio, IMFs significantly underperform CMFs during the pre-Asian
et al. 2001 sample of 14 Jensen’s alpha, Modigliani financial crisis period (1992 to 1996). During the crisis period
(2007) IMFs and 51 measure, and information (1997-1998), IMFs outperform CMFs. The results in the pre-and-
CMFs ratio during the crisis periods are statistically significant, as measured
by the adjusted Sharpe ratio and the Modigliani measure. For the
post-crisis period (1999-2001) and the overall period (1992-2001),
the IMFs do not perform significantly different from CMFs. No
statistically significant positive timing ability of both IMFs and
CMFs is found.
Merdad Saudi 2003- Unmatched Sharpe ratio, Treynor ratio, IMFs perform better than CMFs during bearish (March 2006-
et al. Arabia 2010 sample of 12 Jensen’s alpha, Modigliani January 2010) and financial crisis (September 2008-January 2010)
(2010) IMFs and 16 measure, a measure of the period but not during bullish (January 2003-February 2006) and
CMFs difference between the full (January 2003 to January 2010) period. However, these results
Treynor ratio of the fund are not robust across various benchmarks such as GCC Islamic
and the market index, the Index, MSCI World Islamic Index, Tadawul All Share Index
coefficient of variation, and (TASI), and MSCI World Index. There is no test for statistical
Treynor & Mazuy (1966) difference in the alpha performance of IMFs and CMFs during
timing ability model) those four periods. There is little evidence that HSBC fund
manager has selection and timing ability for conventional funds
during bullish period and for IMFs during bearish period
Mahmud Pakistan 2006– Unmatched 20 Annual returns, Sharpe All mutual funds including IMFs underperform the market
& Mirza 2010 IMFs and 66 ratio, modified Sharpe ratio benchmarks. Their performance is even worse during bearish
(2011) CMFs and Jensen’s alpha (1968) market condition.
Table 1. (Continued)
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.
In addition to comparison between positive and negative screening,

another study examined the difference among wider types of investment
screens on funds’ risk adjusted returns and whether applying an additional
number of SRI screening (screening intensity) progressively provides better or
worse risk adjusted returns. Barnett and Salomon (2006) investigated a panel
of 67 SRI funds from 1972 to 2000 in the US and found that there was a non-
linear relationship between the number of screenings used and the risk
adjusted return of the SRI funds. They found that there is negative relationship
between number of social screens and performance at first, then this
relationship become positive as the number of screens increase. In addition,
their study found that different types of social screens have different impact to
financial performance of managed funds. This implies that socially responsible
or ethical investors are not homogeneous, and they have different social or
ethical tastes. Some may prefer to have their investment support shareholder
activism while others may prefer environmental screening.
Renneboog et al. (2008a) used SRI fund data from countries all over the
world and investigated the impact of screening on the expected return of SRI
funds. Using a sample of 440 SRI funds all over the world, they found that
there was an inverse relationship between screening intensity and SRI
performance. They found that there is a positive relation between the number
of SRI screens and performance. Their results showed that SRI funds with at
least eight screens are more likely to generate higher returns compared with
funds with fewer screens. This implies that SRI screens are able to signal good
management skills and avoid future risk.
Lee et al. (2010) find that there is no effect between screening intensity
and SRI performance using the total return, Sharpe ratio, Information ratio,
Modigliani and Modigliani and Jensen alpha. However, they find there is a
significant reduction in the Carhart four-factor alpha for every additional
screen imposed. Lee et al. (2010) investigated the relationship between
screening intensity and risk. They defined risks as total, systematic and
unsystematic risk. They found that screening intensity has no effect on
idiosyncratic risk but increased screening reduced systematic risk and total
risk. They explained that reduction in systematic risk was in line with
managers selecting stocks with lower beta to minimise overall risk in SRI
funds.
Humphrey and Lee (2011) investigate the performance, risk and screening
intensity using SRI funds in Australia. They find weak evidence that positive
or negative screening impact total return, and find weak evidence that
screening intensity has positive relation with risk-adjusted performance. In the
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.
4. FUND FLOWS AND PERFORMANCE

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’).
4.1. Fund Flows and Past Performance
Investors consider past fund performance as the most important factor in

fund selection decision [see Capon et al. (1994) and Wilcox (2003)], even
though past studies consistently reveal that performance persists among poor
performance [examples include Carhart (1997) Christopherson et al. (1998),
Goetzmann and Ibbotson (1994) and Gruber (1996)]. This behavior implies
that past performance may influence mutual funds market share [see, for
example, Ippolito (1992) and Lynch and Musto (2003)]. Consequently, as
investors may act irrationally by chasing past returns, fund sponsors may take
advantage of this market imperfection (anomalies) by aggressively marketing
top (past) performing funds resulting in a convex flow performance
relationship.
Spitz (1970) was the first to examine the performance flow relationship,
and he found no relationship between past performance and money-flows.
Later, studies found a positive relationship between recent past performance
and money-flow to mutual funds where there is a positive relation between
recent good (worse) performing funds and net money inflows (outflows) [see
Chevalier and Ellison (1997), Del Guercio and Tkac (2008), Goetzmann and
Peles (1997), and Patel et al. (1994)]. This implies that investors reward recent
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
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
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.
This asymmetric relationship may also be due to costly information, high

redemption or switching fees, tax consideration and search costs due to market
imperfection [see, for example, Sirri and Tufano (1998) and Huang et al.
(2007)]. Retail investors have to pay a high search cost to find the best and
most suitable funds for them. Thus, once they have bought the funds, it is quite
difficult for them to exit, as they may have to bear additional costs such as
redemption or switching fees. If they choose to switch, they have to incur
another search cost.
Lynch and Musto (2003) and Berk and Green (2004) provide analytical
studies to explain why investors chase past performance. In a theoretical
paper, Lynch and Musto (2003) argued that this is because performance
persistence exists among superior funds, not inferior funds. As poor
performance does not predict future performance, investors do not respond as
strongly or as pronounced as past good performance. This leads to the convex
curve in the fund flow-performance relationship.
Berk and Green (2004) used a Bayesian model to explain why investors
chase past performance. According to this model, rational investors use past
performance to signal fund manager skill to produce superior performance.
This explains why investors still hold the funds even though those funds
perform poorly. Investors expect that fund managers will revise their strategy
and anticipate future better performance. This however, is contradictory to
Carhart (1997), who found that performance persistence only exists in poor
performing funds but not superior funds. If this holds true, funds with poor
performance will experience significant outflow. Lynch and Musto (2003),
and Berk and Green (2004) exerted that the fund flow-performance
asymmetrical relationship exists even though the performance persistence is
not present.
While there is sufficient evidence in relation to conventional managed
funds, not many studies have investigated the behavior of socially responsible
or ethical investors towards past performance [see, for example, Bollen
(2007), Renneboog et al. (2008a) and Benson and Humphrey (2008)] and,
surprisingly, almost none on IMF investors. The behavior of ethical or
religious investors is interesting to investigate. This is because they deviate
from the standard finance theory that assumes rational investors only consider
risk and return when making financial decisions. Rational investors are those
who only wish to maximize return or utility for a given level of risk and
minimize risk for a given level of return. As evidenced in past literature,
screened funds investors such as SRI, ethical or faith based investors care
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.
4.2. Fund Flow Volatility
Another strand of literature concerning fund flow investigates the

volatility of fund flow into mutual funds. The objective of looking at fund flow
volatility is to identify whether investors are loyal to funds even though the
fund experiences poor performance. High flow volatility may also indicate that
investors are myopic. Active investors are myopic as they invest for short-term
returns. Myopic investors affect long-term investors and fund managers. Fund
managers and fund companies would be better off having investors that are
loyal, as this will affect the fund manager incentives. In general, the size of
asset under management (AUM), which partly depends on the amount of
money-flows from investors, determines the amount of incentives a manager
gets. Myopic investors who actively switch between funds, consequently, give
a negative impact on the fund’s performance as well as fund flows.
Bollen (2007), Balios and Livieratos (2007) and Angel and Rivoli (1997)
argued that ethical investors are more loyal than non-ethical investors are.
Non-ethical investors may not be loyal to their funds because they are merely
interested in risk and returns as rational profit maximising investors. Ethical
investors are more loyal as they integrate social and religious values in their
investment along with risk returns criteria. Even if their investments do not
provide superior return, ethical or screened investors may hold on to the stocks
due to their ethical values. Meaning they consume the nonfinancial attributes
and receive feel good sentiment that their investments are in line with the
moral or ethical values they hold.
Only a few empirical studies support the notion that ethical investors are
loyal. These studies are Bollen (2007) and Peifer (2011). Bollen (2007)
investigated the behavior between SRI and conventional investors by using
205 US SRI equity funds matched with similar non-SRI funds from 1980 to
2002. He found that SRI funds experience lower volatility than conventional
funds. Furthermore, SRI investors are less sensitive to past poor performance
but surprisingly more sensitive to past good performance than conventional
investors are. They explained that SRI investors consumed the SRI attributes.
This in return makes them more committed and loyal to their funds. In another
study, Peifer (2011) extends the study of Bollen (2007) by incorporating
another type of investor. Rather than a simple comparison between
conventional and SRI, he further divided SRI investors into religious SRI and
non-religious SRI investors. Similarly, he used US open-end equity funds with
recent data from 1992 to 2007. He found that religious SRI investors were
more loyal than non-religious SRI as well as conventional investors. He
argued that the findings of Bollen (2007), those SRI investors were more loyal,
could be attributed to religious SRI investors.
4.3. Other Determinants of Fund Flows
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,
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 realised 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.
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
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
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.
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’).
5.1. Fund Flows and Past Performance
Investors consider past fund performance as the most important factor in

fund selection decision [see Capon et al. (1994) and Wilcox (2003)], even
though past studies consistently reveal that performance persists among poor
performance [examples include Carhart (1997) Christopherson et al. (1998),
Goetzmann and Ibbotson (1994) and Gruber (1996)]. This behavior implies
that past performance may influence mutual funds market share [see, for
example, Ippolito (1992) and Lynch and Musto (2003)]. Consequently, as
investors may act irrationally by chasing past returns, fund sponsors may take
advantage of this market imperfection (anomalies) by aggressively marketing
top (past) performing funds resulting in a convex flow performance
relationship.
Spitz (1970) was the first to examine the performance flow relationship,
and he found no relationship between past performance and money-flows.
Later, studies found a positive relationship between recent past performance
and money-flow to mutual funds where there is a positive relation between
recent good (worse) performing funds and net money inflows (outflows) [see
Chevalier and Ellison (1997), Del Guercio and Tkac (2008), Goetzmann and
Peles (1997), and Patel et al. (1994)]. This implies that investors reward recent
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
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
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 realis 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.
This asymmetric relationship may also be due to costly information, high
redemption or switching fees, tax consideration and search costs due to market
imperfection [see, for example, Sirri and Tufano (1998) and Huang et al.
(2007)]. Retail investors have to pay a high search cost to find the best and
most suitable funds for them. Thus, once they have bought the funds, it is quite
difficult for them to exit, as they may have to bear additional costs such as
redemption or switching fees. If they choose to switch, they have to incur
another search cost.
Lynch and Musto (2003) and Berk and Green (2004) provide analytical
studies to explain why investors chase past performance. In a theoretical
paper, Lynch and Musto (2003) argued that this is because performance
persistence exists among superior funds, not inferior funds. As poor
performance does not predict future performance, investors do not respond as
strongly or as pronounced as past good performance. This leads to the convex
curve in the fund flow-performance relationship.
Berk and Green (2004) used a Bayesian model to explain why investors
chase past performance. According to this model, rational investors use past
performance to signal fund manager skill to produce superior performance.
This explains why investors still hold the funds even though those funds
perform poorly. Investors expect that fund managers will revise their strategy
and anticipate future better performance. This however, is contradictory to
Carhart (1997), who found that performance persistence only exists in poor
performing funds but not superior funds. If this holds true, funds with poor
performance will experience significant outflow. Lynch and Musto (2003),
and Berk and Green (2004) exerted that the fund flow-performance
asymmetrical relationship exists even though the performance persistence is
not present.
While there is sufficient evidence in relation to conventional managed
funds, not many studies have investigated the behavior of socially responsible
or ethical investors towards past performance [see, for example, Bollen
(2007), Renneboog et al. (2008a) and Benson and Humphrey (2008)] and,
surprisingly, almost none on IMF investors. The behavior of ethical or

religious investors is interesting to investigate. This is because they deviate
from the standard finance theory that assumes rational investors only consider
risk and return when making financial decisions. Rational investors are those
who only wish to maximize return or utility for a given level of risk and
minimize risk for a given level of return. As evidenced in past literature,
screened funds investors such as SRI, ethical or faith based investors care
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.
5.2. Fund Flow Volatility
Another strand of literature concerning fund flow investigates the

volatility of fund flow into mutual funds. The objective of looking at fund flow
volatility is to identify whether investors are loyal to funds even though the
fund experiences poor performance. High flow volatility may also indicate that
investors are myopic. Active investors are myopic as they invest for short-term
returns. Myopic investors affect long-term investors and fund managers. Fund
managers and fund companies would be better off having investors that are
loyal, as this will affect the fund manager incentives. In general, the size of
asset under management (AUM), which partly depends on the amount of
money-flows from investors, determines the amount of incentives a manager
gets. Myopic investors who actively switch between funds, consequently, give
a negative impact on the fund’s performance as well as fund flows.
Bollen (2007), Balios and Livieratos (2007) and Angel and Rivoli (1997)
argued that ethical investors are more loyal than non-ethical investors are.
Non-ethical investors may not be loyal to their funds because they are merely
interested in risk and returns as rational profit maximizing investors. Ethical
investors are more loyal as they integrate social and religious values in their
investment along with risk returns criteria. Even if their investments do not
provide superior return, ethical or screened investors may hold on to the stocks
due to their ethical values. Meaning they consume the nonfinancial attributes
and receive feel good sentiment that their investments are in line with the
moral or ethical values they hold.
Only a few empirical studies support the notion that ethical investors are
loyal. These studies are Bollen (2007) and Peifer (2011). Bollen (2007)
investigated the behavior between SRI and conventional investors by using
205 US SRI equity funds matched with similar non-SRI funds from 1980 to
2002. He found that SRI funds experience lower volatility than conventional
funds. Furthermore, SRI investors are less sensitive to past poor performance
but surprisingly more sensitive to past good performance than conventional

investors are. They explained that SRI investors consumed the SRI attributes.
This in return makes them more committed and loyal to their funds. In another
study, Peifer (2011) extends the study of Bollen (2007) by incorporating
another type of investor. Rather than a simple comparison between
conventional and SRI, he further divided SRI investors into religious SRI and
non-religious SRI investors. Similarly, he used US open-end equity funds with
recent data from 1992 to 2007. He found that religious SRI investors were
more loyal than non-religious SRI as well as conventional investors. He
argued that the findings of Bollen (2007), those SRI investors were more loyal,
could be attributed to religious SRI investors.
5.3. Other Determinants of Fund Flows
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,
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.
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
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
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.
Abdullah, F., Hassan, T. and Mohamad, S. (2007). Investigation of

performance of Malaysian Islamic unit trust funds: Comparison with
conventional unit trust funds. Managerial Finance, 33(2), 142–153.
Abdullah, N. A. and Abdullah, N. A. H. (2009). The performance of
Malaysian unit trusts investing in domestic versus international markets.
Asian Academy of Management Journal of Accounting and Finance, 5(2),
77–100.
Admati, A. R. and Ross, S. A. (1985). Measuring investment performance in a
rational expectations equilibrium model. The Journal of Business, 58(1),
1–26.
Agarwal, V., Daniel, N. D. and Naik, N. Y. (2003). Flows, performance and
managerial incentives in the hedge fund industry. Working Paper. Georgia
State University. Georgia, United States.
Albaity, M. and Ahmad, R. (2008). Performance of Syariah and composite
indices: Evidence from Bursa Malaysia. Asian Academy of Management
Journal of Accounting and Finance, 4(1), 23–43.
Alexander, G. J. and Buchholz, R. A. (1978). Corporate social responsibility
and stock market performance. The Academy of Management Journal,
21(3), 479–486.
Angel, J. J. and Rivoli, P. (1997). Does ethical investing impose a cost upon
the firm? A theoretical perspective. The Journal of Investing (Winter
1997), 57–61.
Balios, D. and Livieratos, A. (2007). Ethical investing and volatility: A
theoretical perspective. International Research Journal of Finance and
Economics(9), 105–110.
Barber, B. M., Odean, T. and Zheng, L. (2005). Out of sight, out of mind: The
effects of expenses on mutual fund flows. Journal of Business and
Accounting, 78(6), 2095–2119.
Barnea, A., Heinkel, R. and Kraus, A. (2005). Green investors and corporate
investment. Structural Change and Economic Dynamics, 16(3), 332–346.
Barnett, M. L. and Salomon, R. M. (2006). Beyond dichotomy: The
curvilinear relationship between social responsibility and financial
performance. Strategic Management Journal, 27(11), 1101–1122.
Bauer, R., Derwall, J. and Otten, R. (2007). The ethical mutual fund
performance debate: New evidence from Canada. Journal of Business
Ethics, 70, 111–124.
Bauer, R., Guenster, N. and Otten, R. (2004). Empirical evidence on corporate
governance in Europe: The effect on stock returns, firm value and
performance. Journal of Asset Management, 5(2), 91–104.
Bauer, R., Koedijk, K. and Otten, R. (2005). International evidence on ethical

mutual fund performance and investment style. Journal of Banking &
Finance, 29(7), 1751–1767.
Bauer, R., Otten, R. and Rad, A. T. (2006). Ethical investing in Australia: Is
there a financial penalty? Pacific-Basin Finance Journal, 14(1), 33–48.
Benson, K. L. and Humphrey, J. E. (2008). Socially responsible investment
funds: Investor reaction to current and past returns. Journal of Banking &
Finance, 32, 1850–1859.
Berk, J. B. and Green, R. C. (2004). Mutual fund flows and performance in
rational markets. Journal of Political Economy, 112(6), 1269–1295.
Berk, J. B. and Tonks, I. (2007). Return persistence and fund ﬂows in the
worst performing mutual funds. NBER working paper. Retrieved from
http://www.nber.org on 2 October 2009.
BinMahfouz, S. and Hassan, K. M. (2012). A comparative study between the
investment characteristics of Islamic and conventional equity mutual
funds in Saudi Arabia. The Journal of Investing, Winter 2012, 128–143.
Bird, E., Chin, H. and McCrae, M. (1983). The performance of Australian
superannuation funds. Australian Journal of Management, 8(1), 49–69.
Bollen, N. P. B. (2007). Mutual fund attributes and investor behavior. Journal
of Financial and Quantitative Analysis, 42(3), 683–708.
Bollen, N. P. B. and Busse, J. A. (2001). On the timing ability of mutual fund
managers. The Journal of Finance, 56(3), 1075–1094.
Bondt, W. F. M. D. and Thaler, R. (1985). Does the stock market overreact?
The Journal of Finance, 40(3), 793–805.
Brown, S. J., Goetzmann, W., Ibbotson, R. G. and Ross, S. A. (1992).
Survivorship bias in performance studies. The Review of Financial
Studies, 5(4), 553–580.
Brown, S. J. and Goetzmann, W. N. (1995). Performance persistence. The
Journal of Finance, 50, 679–698.
Brown, S. J. and Goetzmann, W. N. (1997). Mutual fund styles. Journal of
Financial Economics, 43(3), 373–399.
Capon, N., Fitzsimons, G. J. and Weingarten, R. (1994). Affluent investors
and mutual fund purchases. International Journal of Bank Marketing,
12(3), 17–25.
Carhart, M. M. (1997). On persistence in mutual fund performance. The
Journal of Finance, 52(1), 57–82.
Carlson, R. S. (1970). Aggregate performance of mutual funds, 1948-1967.
Journal of Financial and Quantitative Analysis, 5(1), 1–32.
Chang, E. C. and Lewellen, W. G. (1984). Market timing and mutual fund

investment performance. The Journal of Business, 57(1), 57–72.
Chen, H.-L., Jegadeesh, N. and Wermers, R. (2000). The value of active
mutual fund management: An examination of the stockholdings and trades
of fund managers. Journal of Financial and Quantitative Analysis, 35(3),
343–368.
Chen, J., Hong, H., Huang, M. and Kubik, J. D. (2004). Does fund size erode
mutual fund performance? The role of liquidity and organization. The
American Economic Review, 94(5), 1276–1302.
Chen, N.-F., Roll, R. and Ross, S. A. (1986). Economic forces and the stock
market. The Journal of Business, 59(3), 383–403.
Chen, Z. and Knez, P. J. (1996). Portfolio performance measurement: Theory
and applications. The Review of Financial Studies, 9(2), 511–555.
Chevalier, J. and Ellison, G. (1997). Risk taking by mutual funds as a response
to incentives. Journal of Political Economy, 105(6), 1167–1200.
Christopherson, J. A., Ferson, W. E. and Glassman, D. A. (1998).
Conditioning manager alphas on economic information: Another look at
the persistence of performance. The Review of Financial Studies, 11(1),
111–142.
Ciccotello, C. S. and Grant, C. T. (1996). Equity fund size and growth:
Implications for performance and selection. Financial Services Review,
5(1), 1–12.
Claessens, S. (1997). Corporate governance and equity prices: Evidence from
the Czech and Slovak Republics. The Journal of Finance, 52(4), 1641–
1658.
Connor, G. and Korajczyk, R. A. (1986). Performance measurement with the
arbitrage pricing theory: A new framework for analysis. Journal of
Cooper, M. J., Gulen, H. and Rau, P. R. (2005). Changing names with style:
Mutual fund name changes and their effects on fund flows. The Journal of
Finance, 60(6), 2825–2858.
Cowles, A. (1933). Can stock market forecasters forecast? Econometrica, 1(3),
309–324.
Cremers, K. J. M. and Nair, V. B. (2005). Governance mechanisms and equity
prices. The Journal of Finance, 60(6), 2859–2894.
Dahlquist, M., Engstrom, S. and Soderlind, P. (2000). Performance and
characteristics of Swedish mutual funds. Journal of Financial and
Quantitative Analysis, 35(3), 409–423.
Dahlquist, M. and Soderlind, P. (1999). Evaluating portfolio performance with

stochastic discount factors. The Journal of Business, 72(3), 347–383.
Daniel, K., Grinblatt, M., Titman, S. and Wermers, R. (1997). Measuring
mutual fund performance with characteristic-based benchmarks. The
Davis, J. L. (2001). Mutual fund performance and manager style. Financial
Analysts Journal, 57(1), 19–27.
de Colle, S. and York, J. G. (2009). Why wine is not glue? The unresolved
problem of negative screening in socially responsible investing. Journal of
Business Ethics, 85(1), 83–95.
Del Guercio, D. and Tkac, P. A. (2008). Star power: The effect of Morningstar
ratings on mutual fund flow. Journal of Financial and Quantitative
Analysis, 43(4), 907–936.
Derigs, U. and Marzban, S. (2008). Review and analysis of currrent Shariah-
compliant equity screening practices. International Journal of Islamic and
Middle Eastern Finance and Management, 1(4), 285–303.
Derigs, U. and Marzban, S. (2009). New strategies and a new paradigm for
Shariah-compliant portfolio optimization. Journal of Banking & Finance,
33(6), 1166–1176.
Dillenburg, S., Greene, T. and Erekson, H. (2003). Approaching socially
responsible investment with a comprehensive ratings scheme: Total social
impact. Journal of Business Ethics, 43(3), 167–177.
Dowell, G., Hart, S. and Yeung, B. (2000). Do corporate global environmental
standards create or destroy market value? Management Science, 46(8),
1059–1074.
Droms, W. G. and Walker, D. A. (1994). Investment performance of
international mutual funds. The Journal of Financial Research, 18(1), 1–
14.
Droms, W. G. and Walker, D. A. (1996). Mutual fund investment
performance. Quarterly Review of Economics and Finance, 36(3), 347.
Dybvig, P. H. and Ross, S. A. (1985). Differential information and
performance measurement using a security market line. The Journal of
Finance, 40(2), 383–399.
Edelen, R. M. (1999). Investor flows and the assessed performance of open-
end mutual funds. Journal of Financial Economics, 53(3), 439–466.
Elfakhani, S. M. and Hassan, M. K. (2005). Performance of Islamic mutual
funds. Paper presented at the 12th Economic Research Forum Conference
Cairo, Egypt.
Elton, E. J., Gruber, M. J. and Blake, C. R. (1996). The persistence of risk-

adjusted mutual fund performance. The Journal of Business, 69(2), 133–
157.
Elton, E. J., Gruber, M. J., Das, S. and Hlavka, M. (1993). Efficiency with
costly information: A reinterpretation of evidence from managed
portfolios. The Review of Financial Studies, 6(1), 1–22.
Fama, E. F. (1970). Efficient capital markets: A review of theory and
empirical work. The Journal of Finance, 25(2), 383–417.
Fama, E. F. (1972). Components of investment performance. The Journal of
Finance, 27(3), 551–567.
Fama, E. F. and French, K. R. (1992). The cross-section of expected stock
returns. The Journal of Finance, 47(2), 427–465.
Fama, E. F. and French, K. R. (1993). Common risk factors in the returns on
stocks and bonds. Journal of Financial Economics, 33(1), 3–56.
Fama, E. F. and French, K. R. (1996). Multifactor explanations of asset pricing
anomalies. The Journal of Finance, 51(1), 55–84.
Fant, L. F. and O'Neal, E. S. (2000). Temporal changes in the determinants of
mutual fund flows. The Journal of Financial Research, 23(3), 353–371.
Ferson, W. E. and Schadt, R. W. (1996). Measuring fund strategy and
performance in changing economic conditions. The Journal of Finance,
51(2), 425–461.
Ferson, W. E. and Warther, V. A. (1996). Evaluating fund performance in a
dynamic market. Financial Analysts Journal, 52(6), 20–28.
Forte, G. and Miglietta, F. (2007). Islamic mutual funds as faith-based funds in
a socially responsible context. Social Science Research Network.
Retrieved from http://ssrn.com on 31 January 2011.
Friedman, M. (1970). The social responsibility of business is to increase its
profits. The New York Times Magazines, 13, 32.
Friend, I., Brown, F. E., Herman, E. S. and Vickers, D. (1962). A study of
mutual funds. Washington D.C., US: Government Printing Office.
Gallagher, D. R. (2003). Investment manager characteristics, strategy, top
management changes and fund performance. Accounting & Finance,
43(2003), 283–309.
Geczy, C. C., Stambaugh, R. F. and Levin, D. (2005). Investing in socially
responsible mutual funds. Social Science Research Network. Retrieved
from http://ssrn.com on 30 August 2009.
Girard, E. and Hassan, M. K. (2005). Faith-based ethical investing: The case
of Dow Jones Islamic indexes FMA Papers. Retrieved from
http://www.fma.org on 22 July 2009.
Glosten, L. R. and Jagannathan, R. (1994). A contingent claim approach to

performance evaluation. Journal of Empirical Finance, 1(2), 133–160.
Goetzmann, W. N. and Ibbotson, R. (1994). Do winners repeat? Journal of
Portfolio Management Science, 20(2), 9–18.
Goetzmann, W. N. and Peles, N. (1997). Cognitive dissonance and mutual
fund investors. The Journal of Financial Research, 20(2), 145.
Goldreyer, E. F., Ahmed, P. and Diltz, J. D. (1999). The performance of
socially responsible mutual funds: Incorporating sociopolitical
information in portfolio selection. Managerial Finance, 25(1), 23.
Golec, J. H. (1996). The effects of mutual fund managers' characteristics on
their portfolio performance, risk and fees. Financial Services Review, 5(2),
133–148.
Gompers, P., Ishii, J. and Metrick, A. (2003). Corporate governance and
equity prices. The Quarterly Journal of Economics, 118(1), 107–155.
Gregory, A., Matatko, J. and Luther, R. (1997). Ethical unit trust financial
performance: Small company effects and fund size effects. Journal of
Business and Accounting, 24(5), 705–725.
Grinblatt, M. and Titman, S. (1989a). Mutual fund performance: An analysis
of quarterly portfolio holdings. The Journal of Business, 62(3), 393–416.
Grinblatt, M. and Titman, S. (1989b). Portfolio performance evaluation: Old
issues and new insights. The Review of Financial Studies, 2(3), 393–421.
Grinblatt, M. and Titman, S. (1992). The persistence of mutual fund
performance. The Journal of Finance, 47(5), 1977–1984.
Grinblatt, M. and Titman, S. (1993). Performance measurement without
benchmarks: An examination of mutual fund returns. The Journal of
Business, 66(1), 47–68.
Grinblatt, M. and Titman, S. (1994). A study of monthly mutual fund returns
and performance evaluation techniques. Journal of Financial and
Quantitative Analysis, 29(3), 419–444.
Grinblatt, M., Titman, S. and Wermers, R. (1995). Momentum investment
strategies, portfolio performance and herding: A study of mutual fund
behavior. The American Economic Review, 85(5), 1088–1105.
Grossman, S. J. and Stiglitz, J. E. (1980). On the impossibility of
informationally efficient markets. The American Economic Review, 70(3),
393–408.
Gruber, M. J. (1996). Another puzzle: The growth in actively managed mutual
funds. The Journal of Finance, 51(3), 783–810.
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.
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.
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.
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
2011 International Conference on Economics and Finance Research

Singapore.
Mansor, F. and Bhatti, M. I. (2011b). Islamic mutual funds performance for
emerging market, during bullish and bearish: The case of Malaysia. Paper
presented at the 2nd International Conference on Business and Economic
Research, Kedah, Malaysia.
Mansor, F. and Bhatti, M. I. (2011c). Risk and return analysis on performance
of the Islamic mutual funds: evidence from Malaysia. Global Economy
and Finance Journal, 4(1), 19-31.
Mansor, F., Bhatti, M. I. and Khan, H. (2012). Islamic mutual funds
performance: A panel analysis. Paper presented at the 2nd Malaysian
Postgraduate Conference, Bond University, Queensland, Australia.
Markowitz, H. (1952). Portfolio selection. The Journal of Finance, 7(1), 77–
91.
Massa, M. (2003). How do family strategies affect fund performance? When
performance-maximization is not the only game in town. Journal of
Merdad, H., Hassan, M. K. and Alhenawi, Y. (2010). Islamic versus
conventional mutual funds performance in Saudi Arabia: A case study.
J.KAU: Islamic Economics, 23(2), 157–193.
Merton, R. C. (1987). A simple model of capital market equilibrium with
incomplete information. The Journal of Finance, 42(3), 483–150.
Mossin, J. (1966). Equilibrium in a capital asset market. Econometrica, 34(4),
768–783.
Nainggolan, Y. (2011). Taking a leap of faith: are investors left short changed.
Doctor of Philosophy, Queensland University of Technology, Brisbane,
Australia.
Nanda, V., Wang, Z. J. and Zheng, L. (2004). Family values and the star
phenomenon: Strategies of mutual fund families. The Review of Financial
Studies, 17(3), 667–698.
Nassir, A. M., Mohamad, S. and Hua, N. M. (1997). Selectivity and timing:
evidence from the performance of Malaysian unit trusts. Pertanika
Journal of Social Science & Humanity, 5(1), 45–57.
Nathie, M. (2009). Islamic equity investments: Determinants of investment
behaviour of Malaysian Islamic equity investors under conditions of
competing alternatives. Degree of Doctor Philosophy, Griffith University,
Brisbane, Australia.
Oh, N. Y. (2005). Essays on the dynamic relationship between different types

of Investment flow and prices. Degree of Doctor of Philosophy, University
of New South Wales, Sydney, Australia.
Orlitzky, M., Schmidt, F. L. and Rynes, S. L. (2003). Corporate social and
financial performance: A meta-analysis. Organization Studies, 24(3), 403–
441.
Otten, R. and Bams, D. (2002). European mutual fund performance. European
Financial Management, 8(1), 75.
Pastor, L. and Stambaugh, R. F. (2002a). Investing in equity mutual funds.
Journal of Financial Economics, 63(3), 351–380.
Pastor, L. and Stambaugh, R. F. (2002b). Mutual fund performance and
seemingly unrelated assets. Journal of Financial Economics, 63(3), 315–
349.
Patel, J., Zeckhauser, R. and Hendricks, D. (1994). Investment flow and
performance: evidence from mutual funds, cross-border investments and
new issues. In S. Ryuzo, R. M. Levich and R. V. Ramachandran (Eds.),
Japan, Europe, and International Financial Markets: Cambridge
University Press, Cambridge.
Peifer, J. L. (2011). Morality in the financial market? A look at religiously
affiliated mutual funds in the USA. Socio-Economic Review, 9(2), 235–
259.
Potter, M. E. (2000). Determinants of aggregate mutual fund flows. The
Journal of Business and Economic Studies, 6(2), 55–55–73.
Prather, L., Bertin, W. J. and Henker, T. (2004). Mutual fund characteristics,
managerial attributes and fund performance. Review of Financial
Economics, 13(4), 305–326.
Razzaq, N., Gul, S., Sajid, M., Mughal, S. and Bukhari, S. A. (2012).
Performance of Islamic mutual funds in Pakistan. Economics and Finance
Review, 2(3), 16–25.
Reiley, F. K. and Brown, K. C. (2006). Investment analysis and portfolio
management (Eight ed.). Canada: Thomson South-Western.
Remolona, E. M., Kleiman, P. and Gruenstein, D. (1997). Market returns and
mutual fund flows. FRBNY Economic Policy Review, 3, 33–52.
Renneboog, L., Ter Horst, J. R. and Zhang, C. (2008a). The price of ethics and
stakeholder governance: The performance of socially responsible mutual
funds. Journal of Corporate Finance, 14(3), 302–322.
Renneboog, L., Ter Horst, J. R. and Zhang, C. (2008b). Socially responsible
investments: Institutional aspects, performance and investor behaviour.
Journal of Banking & Finance, 32(2008), 1723–1742.
Renneboog, L., Ter Horst, J. R. and Zhang, C. (2011). Is ethical money

financially smart? Nonfinancial attributes and money flows of socially
responsible investment funds. Journal of Financial Intermediation, 20(4),
562–588.
Richardson, B. (2007). Do the fiduciary duties of pension funds hinder socially
responsible investment? Banking and Finance Law Review, 22(2), 145–
201.
Roll, R. (1977). A critique of the asset pricing theory's tests Part I: On past and
potential testability of the theory. Journal of Financial Economics, 4(2),
129–176.
Roll, R. (1978). Ambiguity when performance is measured by the securities
market line. The Journal of Finance, 33(4), 1051–1069.
Ross, S. A. (1976). The arbitrage theory of capital asset pricing. Journal of
Economic Theory, 13(3), 341–360.
Rozali, M. B. and Abdullah, F. (2006). The performance of Malaysian equity
funds. The Business Review, 5(2), 301–306.
Ruenzi, S. (2005). Mutual fund growth in standard and specialist market
segments. Financial Markets and Portfolio Management, 19(2), 153–167.
Sawicki, J. and Ong, F. (2000). Evaluating managed fund performance using
conditional measures: Australian evidence. Pacific-Basin Finance
Journal, 8(3-4), 505–528.
Schroder, M. (2004). The performance of socially responsible investments:
Investment funds and indices. Financial Markets and Portfolio
Management, 18(2), 122–142.
Sharpe, W. F. (1966). Mutual fund performance. The Journal of Business,
39(1), 119–138.
Sharpe, W. F. (1992). Asset allocation: Management style and performance
measurement. Journal of Portfolio Management, 18(2), 7–19.
Shefrin, H. and Statman, M. (1985). The disposition to sell winners too early
and ride losers too long: Theory and evidence. The Journal of Finance,
40(3), 777–790.
Sinclair, N. (1990). Market timing ability of pooled superannuation funds
January 1981 to December 1987. Accounting & Finance, 30, 51–65.
Sirri, E. R. and Tufano, P. (1998). Costly search and mutual fund flows. The
Spitz, A. E. (1970). Mutual fund performance and cash inflows. Applied
Economics, 2, 141–145.
Treynor, J. L. (1965). How to rate management of investment funds. Harvard
Business Review, 43(1), 63–75.
Treynor, J. L. and Mazuy, K. K. (1966). Can mutual funds outguess the

market? Harvard Business Review, 44(4), 131–136.
Visaltanachoti, N., Zou, L. and Zheng, Q. (2009). The performance of "sin"
stocks in China. Working paper. Massey University.
Warther, V. A. (1995). Aggregate mutual fund flows and security returns.
Journal of Financial Economics, 39(2–3), 209–235.
Wermers, R. (2000). Mutual fund performance: An empirical decomposition
into stock-picking talent, style, transactions costs and expenses. The
Wilcox, R. T. (2003). Bargain hunting or star gazing? Investors' preferences
for stock mutual funds. The Journal of Business, 76(4), 645–663.
Zheng, L. (1999). Is money smart? A study of mutual fund investors' fund
selection ability. The Journal of Finance, 54(3), 901–933.
Chapter 2
DOES THE CHOICE OF PERFORMANCE

MEASURE MATTER FOR RANKING
OF MUTUAL FUNDS?
Amporn Soongswang1, and Yosawee Sanohdontree2

1
Graduate College of Management, Sripatum University, Jatujak,
Bangkok, Thailand
2
Electricity Generating Authoroty of Thailand, Bang Kruai,
Nonthaburi, Thailand
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.
Keywords: fund ranking, equity fund, mutual fund, performance measure,

Thailand
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
measure because it is meaningful when either risk perceived by investors can

be expressed by standard deviation or when returns are normally distributed.
Other similar measures, such as the Treynor ratio and Jensen’s alpha are also
used in studies. However, more recent studies have emphasized on preferences
and characteristics of returns’ distribution that go beyond the mean and
variance, such as the Sortino ratio, Calmar ratio, Omega ratio, MPPM
(Manipulation-Proof Performance Measure), Upside Potential ratio and
Appraisal ratio (see Ornelas, Junior & Fernandes, 2009 and Zakamouline &
Koekebakker, 2009).
Another view is that the need to consider simultaneously multi-criteria
incorporating investors’ own preferences is natural since they not always share
the same financial objective, risk aversion and investment horizon. This is
consistent with suggestion by Ornelas et al. (2009) in that the fund
performance measure should depend on the investors’ preferences and asset’s
behavior. From this perspective, the Data Envelopment Analysis (DEA)
technique seems predominantly appealing as it provides the possibility of
incorporating many criteria at the same time. The DEA technique has been
applied in various fields including public administration, engineering and
commerce and finance. Nguyen-Thi-Thanh (2006) asserts that the DEA
technique is an efficient tool to assist investors in multi-criteria decision-
making tasks. Also, it is suggested as a mathematical optimizing technique,
which is applicable to equity mutual funds’ performance evaluation (also see
Charnes, Cooper & Rhodes, 1978).
Eling and Schuhmacher (2007) use several performance measures to
evaluate hedge funds and conclude that rankings are virtually identical. This is
similar to those reported by Eling (2008), who analyzes a sample of mutual
funds’ performance across the world using the Sharpe ratio compared with
some other measures. In contrast, Zakamouline (2010) uses the same hedge
fund data used in Eling and Schuhmacher (2007), and suggest that the usage of
different performance measures leads to different rankings. This is in line with
Ornelas et al. (2009). Thus, the issue of selection of evaluation methods for
funds’ performance measurement and ranking is inconclusive.
Even though the number of open-ended funds has been increasing, studies
related to these topics on emerging markets, particularly Thailand, have been
limited. Most past studies had emphasized on closed-ended funds. To
understand more about these funds’ behavior and answer the question whether
performance measure is important for fund ranking, it is justified to carry out a
comprehensive study on the funds’ performance in Thailand. The funds’
performance over 5-year time-periods was evaluated using several measures:
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.
(2009). Apparently, the evidence on funds’ performance is mixed. However,

the more recent findings cast doubts on the efficient market hypothesis and
rekindle investors’ hope of earning abnormal returns by plowing through
historic performance records (also see Zheng, 1999). Thus, if mutual fund
performance is predictable, using performance evaluating can help investors
select funds that will continue to out-perform in the future.
Notice that US funds are much larger than elsewhere in the world, and
domestic funds are larger than international funds, on average. There are
reasons to believe that results of studies may be different as there are
significantly different characteristics between the US mutual fund industry and
the rest of the world. These factors include fund size, style, age and fees,
economic development, financial development, quality of legal institutions and
law enforcement, mutual fund industry structure and others such as ability to
select funds (see Chen, Hong, Huang & Kubik, 2004; Gehin, 2004; Khorana et
al., 2005; Khorana, Servaes & Tufano, 2009; Zheng, 1999). For example,
Muga, Rodriguez, and Santamaria (2007), a Mexico study, find persistence in
mutual fund performance both over consecutive time periods and in the multi-
period setting. Noulas, Papanastasiou, and Lazaridis (2005), a Greek study,
analyze the behavior of 23 mutual funds for the period 1997-2000 and
conclude that the mutual fund industry is relatively young resulting in no
definite conclusion. Agrawal (2007), a study on Indian mutual funds, reveals
that performance of the fund managers affects the returns of the firm.
Moreover, mutual fund is not a widely discussed subject in developing
markets including Thailand, when compared to others. Among few studies that
have focused on Thai mutual funds, Nitibhon (2004) employs the Jensen’s
alpha, condition model, factor model and portfolio holding model to measure
performances of 114 equity funds, and reports statistically insignificant
positive returns. Tirapat (2004b) examines 222 funds’ performance using the
Treynor ratio, Sharpe ratio and Jensen’s alpha, and suggests funds out-perform
the market, but there is no persistency in performance during the periods of
study, which are consistent with those of Detzler (1999), but are inconsistent
with Ferreira et al. (2009) and Muga, Rodriguez, and Santamaria (2007).
The used performance evaluation method is another controversial issue,
whether the choice of measures influences the funds’ evaluation and whether
the measure matters for ranking of mutual funds. Eling (2008) analyzes a
dataset of 38,954 mutual funds then compared the Sharpe ratio with other
performance measures and finds virtually identical ranking ordering. He
concludes that choosing a performance measure is not critical to fund
evaluation and the Sharpe ratio is generally adequate. Meanwhile, Ornelas et
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.
DATA AND METHODOLOGY

This study uses monthly and longer time-period of data covering net asset
values and dividends for the five-year time-periods (May 1, 2002 - April 30,
2007). A larger sample consisting of the returns on the portfolios of 138 open-
ended equity mutual funds was examined. There are four significant sources of
data used for analyses in this study set out as follows: the AIMC (Association
of Investment Management Companies), asset management companies, SET
and Bank of Thailand (BOT) is another source providing 91-day coupon rate
of the Thai government bonds.
Returns on portfolios that belong to the same risk class can be compared
using the three different approaches: the Treynor ratio, Sharpe ratio and
Jensen’s alpha. This is in accordance with the suggestions by Rao, Srivastava,
and Ramachandra (2006) in that these are absolute measures that can be used
to evaluate different portfolios.
Treynor (1965) argues that a portfolio manager should be able to diversify
and eliminate all the unsystematic risk; and thus the Treynor indicator is the
ratio of excess return or risk premium divided by the systematic risk.
Sharpe (1966) uses a ratio to evaluate performance of mutual funds. The
ratio is similar to that of the Treynor ratio with the difference in that the
Sharpe ratio uses the total risk of a portfolio and not just the systematic risk. It
is excessive risk-free return per unit of total risk, and the total risk is measured
by standard deviation of excessive risk-free return.
Jensen (1968) suggests a risk-adjusted measure of portfolio performances
known as “Jensen’s alpha.” This technique considers excessive returns on
mutual funds from expected returns based on level of systematic risk of each
portfolio.
Farrell (1957) addresses that there is existence of multiple inputs and
outputs, the usual approach to calculating efficiency, which is output/input, is
inadequate. The measurement of relative efficiency, which is weighted sum of
outputs/weighted sum of inputs, is more appealing. The DEA technique, which
was initiated by Murthi, Choi, and Desai (1997), can be applied to assess
mutual funds’ performance (Rao et al., 2006).
Even though Eling (2008), Ornelas et al. (2009) and Zakamouline (2010)
have different views on the choice of performance measures, eventually,
Ornelas et al. (2009) suggest that the evaluation of mutual funds should not
rely on a single performance measure. Data with different periodicity and
investment horizons should be used to give robustness. Also, the use of
different performance measures can bring robustness to results, and possibly
avoid manipulation strategies with focus on specific measures. Similarly,

Eling (2008) suggests that a sample that is both large and covers an extensive
period of time is needed to verify statistically whether the results are genuine
or spurious. Wermers (2003) documents that many investments advisory
services advocate the use of a 3-or 5-year evaluating period for consumers to
judge the relative merits of funds (also see Zakamouline, 2010). Finally,
Scholz and Schnusenberg (2008) confirm that evaluating non-survivors and
surviving funds in a joint sample impacts resulting ranking severely.
Thus, the study is distinguish from many past studies that use only the
traditional measures or different/partly different methods, such as Ferreira et
al. (2009), Gupta and Aggarwal (2007) and Rao et al. (2006); or is
complementary to more recent studies, such as Eling (2008), Ornelas et al.
(2009) and Zakamouline (2010) that use the Sharpe ratio compared to those
measures known as “the ones go beyond the mean-variance approach.”
Specifically, in this study, the data were analyzed using four different
performance measurement models: the Treynor ratio, Sharpe ratio, Jensen’s
alpha and the DEA technique. The different investment horizons of the
analyses of fund performances consisting of long time-periods: 1-year (May 1,
2006 – April 30, 2007); 3-year (May 1, 2004 – April 30, 2007) and 5-year
(May 1, 2002 – April 30, 2007).
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.
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.
Data Envelopment Analysis (DEA)

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
where E k is the DEA score of k DMU, y ok the amount of the o output

th th
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
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.
Manager’s investment skill = (rp-rf) - (σ p /σ m ) (rm-rf) (9)
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)
where σ X is the standard deviation of X, σ Y the standard deviation of Y,

ρ X,Y > 0 the values of data set X increase or decrease in the same direction of
set Y, ρ X,Y < 0 the values of data set X increase or decrease in opposite
direction of set Y and ρ X,Y = 0 there is no correlation between data set X
and Y.
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.
Tables 1-4 present that on average, the performances of open-ended equity

mutual funds in the sample of this study experience significantly positive
excess returns for most time-periods of investment. Between 87-100% and 76-
83% of the numbers of total funds evaluated using the Treynor ratio, Sharpe
ratio and Jensen’s alpha methods, and the DEA technique are out-performers,
respectively. Specifically, all funds perform significantly better than the
market for all time-periods of investment, when the traditional performance
measures are employed. However, they behave worse than the market, as the
DEA technique is applied. Thus, the results differ depending on used
performance evaluation measures.
Table 5 shows the relationships between the results estimated applying the
Treynor ratio and Sharpe ratio; Treynor ratio and Jensen’s alpha; Sharpe ratio
and Jensen’s alpha are all significantly positive and have very high correlation.
Nevertheless, there is very low correlation between the results estimated using
the DEA technique and each of the three traditional measures, for all time-
periods of investment.
Table 1. Performance of Thai Open-ended Equity Mutual Funds

Evaluated Using the Treynor Ratio
Time % Out Mean Market Std. Std. t-stat Sig

period perform deviation error
1-Year 96 -0.0027 -0.0062 0.0026 0.0002 16.11 0.000
3-Year 91 0.0042 0.0015 0.0027 0.0003 9.90 0.000
5-Year 100 0.0167 0.0085 0.0036 0.0004 20.07 0.000

Evaluated Using the Sharpe Ratio
Time % Out Mean Market Std. Std. error t-stat Sig

period perform deviation
1-Year 96 -0.1709 -0.4013 0.1577 0.0134 17.16 0.000
3-Year 90 0.3139 0.1198 0.1916 0.0193 10.07 0.000
5-Year 87 0.8046 0.6521 0.1497 0.0173 8.82 0.000

Evaluated Using the Jensen’s Alpha
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

Evaluated Using the DEA Technique
Time % Out Mean Market Std. Std. t-stat Sig

period perform deviation error
1-Year 78 0.9251 0.9411 0.0846 0.0072 -2.23 0.028
3-Year 83 0.9432 0.9556 0.0625 0.0063 -1.97 0.052
5-Year 76 0.9686 0.9742 0.0460 0.0053 -1.05 0.296
Table 5. Relationships between Performance Measures: the Treynor

Ratio, Sharpe Ratio, Jensen’s Alpha and DEA Score for Different Time-
periods of Investment
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
Table 6 suggests that for 1-year time-period of investment, ABSM,

ABSM, BTP and ABSM are ranked No. 1 amongst the top ten best performers
measured by each performance evaluation method. These open-ended equity
mutual funds were managed by the Aberdeen Asset Management Co., Ltd. and
BBL Asset Management Co., Ltd. In the aspect of ranking evaluation, the
evidence indicates that the DEA technique has a closer relationship with the
Theynor and Sharpe ratios than the Jensen’s alpha. However, 40 % of the top
ten best funds ranked using the DEA technique, which are B-INFRA, BKA2,
B-LTF and BTP, are the same top ten best funds ranked based on the
traditional measures.
The results also present that for 3-year time-period of investment, ABG
and BTP are ranked No. 1 amongst the top ten best performers analyzed by
each performance evaluation measure. These open-ended funds were managed
by the Aberdeen Asset Management Co., Ltd. and BBL Asset Management
Co., Ltd consecutively. The relationships between the DEA method and the
Jensen’s alpha and Theynor ratio are closer, compared to that between the
DEA technique and the Sharpe ratio. Nevertheless, up to 80% of the top ten
best funds ranked based on the DEA technique are the same as those ranked
using the traditional measures. These are ABG, ABSC-RMF, BERMF, B-
INFRA, BKA, BKA2, B-SUB and BTP.
Table 6 also suggests that for 5-year time-period of investment, ABG and
B-INFRA are ranked No. 1 amongst the top ten best performers, as evaluated
by each performance assessment measure. These open-ended equity mutual
funds were managed by the Aberdeen Asset Management Co., Ltd. and BBL
Asset Management Co., Ltd. The evidence indicates that in terms of fund
rankings, the DEA technique seems closer to the Sharpe and Treynor ratios
rather than the Jensen’s alpha. However, four from the top ten best funds
ranked based on the DEA technique are the same as those ranked using the
traditional approaches. ABG, B-INFRA, KPLUS and TDF are ranked amongst
the top ten best performers based on all four different performance evaluation
methods. Only B-INFRA is ranked amongst the top ten best performers for 1-
year, 3-year and 5-year time-periods of investment, when all four diverse
performance assessment metrics are used.
Thus, the results are in line with those suggested by Ornelas et al. (2009)
and Zakamouline (2010) that the choice of performance measures matter for
ranking of mutual funds.
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.
Agrawal, R. and Mukhtar, W. (2010), “Critical analysis of stock selection

strategies of growth mutual funds in India: Application of Sharpe
Optimization Technique,” www.ssrn.com.
Bekaert, G. and Urias, S. M. (1998), “Is there a free lunch in emerging market
equities?,” www.ssrn.com.
Bergstresser, D. and Poterba, J. (2002), “Do after-tax returns affect mutual
fund inflows?” Journal of Financial Economics, (63), 381-414.
Brown, S. and Goetzmann, W. (1995), “Performance persistence,” Journal of
Finance, (50), 679-698.
Carhart, M. (1997), “On persistence in mutual fund performance,” Journal of
Finance, (52), 57-82.
Charnes, A., Cooper, W. W. and Rhodes, E. (1978), “Measuring the efficiency
of decision making units,” European Journal of Operation Research, (2),
(4), 429-444.
Chen, J., Hong, H., Huang, M. and Kubik, J. (2004), “Does fund size erode
performance? Liquidity, organizational diseconomies, and active money
management,” American Economic Review, (94), 1276-1302.
Chevalier, J. and Ellison, G. (1997), “Risk taking by mutual funds as a
response to incentives,” Journal of Political Economy, (105), 1167-1200.
Detzler, M. L. (1999) “The value of mutual fund rankings to the individual
investor,” www.ssrn.com.
Eling, M. and Schuhmacher, F. (2007), “Does the choice of performance
measure influence the evaluation of hedge funds?,” Journal of Banking
and Finance, (31), (9), 2632-2647.
Eling, M. (2008), “Does the measure matter in the mutual fund industry?,”
Financial Analysts Journal, (64), (3), 54-66.
Elton, E. J., Gruber, J. M. and Blake, R. C. (1996), “The persistence of risk-
adjusted mutual fund performance,” Journal of Business, (69), (2), 133-
157.
Farrell, M. J. (1957), “The measurement of productive efficiency,” Journal of
the Royal Statistical Society: Series A, (120), 253-281.
Ferreira, A. M., Miguel, F. A. (2009), “The determinants of mutual fund
performance: A cross-country study,” www.ssrn.com.
Gehin, W. (2004), “Survey of the Literature on Hedge Fund Performance,”
www.ssrn.com.
Goetzmann, W. N. and Ibbotson, R.G. (1994), “Do winner repeat? Patterns in
mutual fund performance,” Journal of portfolio management, (20)
(Spring), 9-18.
Grinblatt, M. and Titman, S. (1989), “Mutual fund performance: An analysis

of quarterly portfolio holdings,” Journal of Business, (62), 393-416.
Guber, M. (1996), “Another puzzle: The growth in actively managed mutual
funds,” Journal of Finance, (51), 783-807.
Gupta, M. and Aggarwal, N. (2007), “Performance of mutual funds in India:
An empirical study,” The ICFAI Journal of Applied Finance, (13), (9), 5-
16.
Hendricks, D., Patel, J. and Zeckhauser, R. (1993) “Hot hands in mutual
funds: Short-run persistence of relative performance 1974-1988,” Journal
of Finance, (48), 93-130.
Ippolito, R. (1989), “Efficiency with costly information: A study of mutual
fund performance,” Quarterly Journal of Economics, (104), 1-23.
Jensen, M. (1968), “The performance of mutual funds in the period 1945-
1964,” Journal of Finance, (23), (2), 389-416.
Kempf, A. and Ruenzi, S. (2008), “Family matters: rankings within fund
families and fund inflows,” Journal of Business Finance and Accounting,
(35), 177-199.
Khanthavit, A. (2001), “A Markov-Switching model for mutual fund
performance evaluation,” Working paper, Faculty of Commerce and
Accountancy, Thammasat University.
Khorana, A., Servaes, H. and Tufano, P. (2005), “Explaining the size of the
mutual fund industry around the world,” Journal of Financial Economics,
(78), 145-185.
Khorana, A., Servaes, H. and Tufano, P. (2009), “Mutual fund fees around the
world,” Review of Financial Studies, (22), 1279-1310.
Malkiel, B. (1995), “Returns from investing in equity mutual funds, 1971-
1991,” Journal of Finance, (50), 549-573.
Muga, L., Rodriguez, A. and Santamaria, R. (2007), “Persistence in mutual
funds in Latin American emerging markets: The case of Mexico,” Journal
of Emerging Market Finance, (6), (1), 1-37.
Murthi, B. P. S., Choi, Y. K. and Desai, P. (1997), “Efficiency of mutual funds
and portfolio performance measurement: A non-parametric approach,”
European Journal of Operational Research, (98), 408-418.
Nguyen-Thi-Thanh, H. (2006), “On the use of data envelopment analysis in
assessing local and global performances of hedge funds,” www.ssrn.com.
Nitibhon, C. (2004) “An analysis of performance, persistence and flows of
Thai equity funds,” Master of Science in Finance, Faculty of Commerce
and Accountancy, Chulalongkorn University.
Noulas, G. A., Papanastasiou, A. J. and Lazaridis, J. (2005), “Performance of

mutual funds,” Managerial Finance, (32), (2), 101-112.
Ornelas, R. H. J., Junior, F. D. A. S. A. and Fernandes, L. B. J. (2009), “Yes,
the choice of performance measure does matter for ranking of US mutual
funds,” www.ssrn.com.
Otten, R. and D, Bams. (2002), “European mutual fund performance,”
European Financial Management, (8), (1), 75-101.
Panetta, F. and Cesari, R. (2002), “The performance of Italian equity funds,”
Journal of Banking and Finance, (26), 99-126.
Plantinga, A. and Groot, D. S. (2001), “Risk-adjusted performance measures
and implied risk-attitudes,” www.ssrn.com.
Pushner, G., Rainish, R. and Coogan, D. (2001), “Performance of European
focused mutual funds,” American Business Review, (19), (1), 39-45.
Ramesh, R. and Raj, A. (1987), “Performance of US-based international
mutual funds,” Akron Business and Economic Review, (18), (4), 98-107.
Rao, S. V. D. N., Srivastava, L. and Ramachandra, V. S. (2006), “Single and
multi-criteria ranking of mutual fund schemes,” ssrn.com.
Scholz, H. and Schnusenberg, O. (2008), “Ranking of equity mutual funds:
The bias in using survivorship bias-free datasets,” Working paper.
Catholic University of Eichstaett-Ingolstadt and University of North
Florida.
Sharpe, W. F. (1966), “Mutual Fund Performance,” Journal of Business, (39),
119-138.
Sirri, E. and Tufano, P. (1998), “Costly search and mutual fund flows,”
Journal of Finance, (53), (5), 1589-1622.
Tirapat, S. (2004b), “Performance of mutual funds in Thailand: Development
and Performance of Mutual Funds in Five ASEAN Countries: Indonesia,
Malaysia, the Philippines, Singapore, and Thailand,” Bangkok: ASEAN
University Network Press.
Treynor, J. L. (1965), “How to Rate Management of Investment Funds,”
Harvard Business Review, (43), 63-75.
Wermers, Russ. (2003), “Is money really “smart”? New evidence on the
relation between mutual fund flows, manager behavior, and performance
persistence,” www.ssrn.com.
Zakamouline, V. (2010), “The choice of performance measure does influence
the evaluation of hedge funds,” www.ssrn.com.
Zakamouline, V. and Koekebakker, S. (2009), “Portfolio performance
evaluation with generalized Sharpe ratios: Beyond the mean and
variance,” Journal of Banking and Finance, (33), 1242-1254.
Zheng, L. (1999), “Is money smart? A study of mutual fund investors’ fund
selection ability,” Journal of Finance,” (54), (3), 901-933.
Chapter 3
MUTUAL FUND PREDICTION MODELS

USING ARTIFICIAL NEURAL NETWORKS
AND GENETIC PROGRAMMING
Konstantina Pendaraki , Grigorios Ν. Beligiannis

*
and Alexandra Lappa

Department of of Business Administration of Food and Agricultural
Enterprises, University of Patras, Agrinio, Greece
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
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.
Keywords: forecasting, artificial neural networks, genetic programming,

mutual funds
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
elements of these assumptions. This study aims to resolve this troublesome

situation using Artificial Neural Networks (ANNs) and Genetic Programming
(GP) approaches which provide a high level of adaptability in the decisions of
the portfolio manager or investor, when his environment is changing and the
characteristics of the funds are multidimensional.
ANNs are capable of recognizing and presenting non-linear relations in
the data set, while with at least one hidden layer are able of using data to
develop an interior illustration of the relationships between variables without
fitting to an explicit model. GP is an inventive approach to artificial
intelligence, capable of solving complex problems with quite no human
supervision (Rafiei et al., 2011) and utilizes the survival of the fittest to select
the best solution for the examined problem (Davis, 1991; Koza, 1992; Koza,
1998).
Over the last decade, many artificial intelligence financial applications
have been presented in the literature. ANNs and GP technologies comprise
useful tools for forecasting and predicting financial units. The main
characteristics of these tools are their adaptability, robustness, effectiveness
and efficiency in solving financial problems, such as bankruptcy prediction,
loan evaluation, bond rating, stock market forecasting, portfolio management,
etc. These applications have shown significant success and very encouraging
results outperforming most of traditional statistical approaches, such as
discriminant and regression analysis. For an extensive literature on ANN
applications in finance see the research of Wong and Selve (1998) and Fadlalla
and Lin (2001). Applications of GP in finance can be found in Chen et al.
(2006), Chen and Navet (2006), Bhattacharyya et al. (2002) and Kaboudan
(2000).
Regarding the mutual fund industry, there are studies applying ANN
(Klein and Rossin, 1999; Wang and Huang, 2010) and GP (Pattarin, et al.
2004; Tsai, et al., 2011), but with totally different objectives compared the
present work. However, there are similar works which explored ANN (e.g.,
Kinoto et al., 1990; Huarng and Yu, 2006; Guresen, et al., 2011; Wang, et al.,
2011) and GP (e.g., Mahfound and Mani, 1996; Kaboudan, 2000; Kim and
Hun, 2000; Cheng et al., 2010) on stock price forecasting.
As far as we know, in accordance to our work, there are only three studies
using ANNs for the prediction of mutual fund performance. Specifically,
Chiang et al. (1996) used economic variables and through back-propagation
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.
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.
2.2. Variable Definitions
Based on the theoretical background and past research, in order to

evaluate the effectiveness of prediction models in forecasting the performance
and the net asset value of mutual funds, firstly, 15 variables (operating
characteristics) are identified as inputs to the models (Table 2.1) and secondly,
17 variables (economic variables) are identified as inputs to the models (Table
2.2). These variables are calculated for each one of the examined periods.
A detailed description of both the dependent variables and prediction
variables of the present analysis is given on the appendix.
Table 2.1. The variables (operating characteristics) used in the analysis
Input Variable Input Variable

1 Mean value of shares 9 Annual return of ASE (%)
2 Return (%) 10 Mean value of ASE return
3 Mean value of NAV 11 Standard deviation of ASE returns
4 Mean return 12 Coefficient of variation
5 Standard deviation of 13 Value at risk (%)
the returns
6 Sharpe index 14 Modigliani measure
7 beta coefficient 15 Information ratio
8 Treynor index
Table 2.2. The economic variables used in the analysis
Input Economic Variable Input Economic Variable

1 Gross domestic product, constant 10 Real GDP Growth
prices
2 Gross domestic product, current 11 Gross Fixed Total
prices Investment
3 Gross domestic product per 12 Industrial Production
capita, constant prices
4 Gross domestic product per 13 Unemployment rate
capita, current prices
5 Gross domestic product based on 14 Employment
purchasing-power-parity (PPP)
share of world total
6 Inflation, average consumer 15 Consumer Price Index
prices
7 General government balance 16 Credit Expansion
8 General government structural 17 Deficit-General
balance Government
9 Current account balance
3. THE ANN METHODOLOGY

3.1. Multi-Layer Perceptorns
As known, in a multilayer feed forward network (or multilayer perceptron,

MLP) the neurons are organized into layers and have only forward
connections (Haykin 2008). The first layer is called input layer and consists of
a number of neurons usually equal to the number of inputs. Each input is in

general connected to all input neurons. The last layer is called output layer and
consists of a number of computational neurons. All layers between the input
and output layers are called hidden layers and consist of many computational
neurons. Usually those networks are fully connected, which means that each
neuron in any layer of the network is connected to all neurons of the previous
layer. Every neuron in the output and hidden layers follows the general model
of the neuron (Haykin 2008).
In order to successfully apply ANNs to forecast the performance and the
net asset value of Greek equity mutual funds, the following aspects are
considered:
 Pre-processing of application data, if necessary, and selection of

training and testing data sets.
 Selection of the most appropriate ANN architecture for the problem at
hand.
 Selection of the most suitable parameters for the ANN training
algorithm.
As stated in Section 2.2, in order to apply ANNs to predict the

performance and the net asset value of mutual funds, firstly, 15 variables
(operating characteristics) are identified as inputs to the models (Table 2.1)
and secondly, 17 variables (economic variables) are identified as inputs to the
models (Table 2.2). So, the input layer of the first MLP (applied to predict the
performance of mutual funds) consists firstly of 15 input neurons and secondly
of 17 input neurons. The input layer of the second MLP (applied to predict the
net asset value of mutual funds) also consists firstly of 15 input neurons and
secondly of 17 input neurons. Since arbitrary mappings can be accomplished
with two hidden layers, more than two will not add functionality (Blum and Li
1991). In this research, a single hidden layer proved sufficient, as will be
discussed and demonstrated in Section 5. As stated above, the ANN outputs in
our approach are the performance of mutual funds for the first MLP and the
net asset value of mutual funds for the second MLP, respectively. Since the
dimension of the output vector of both MLPs used is one, one output neuron is
adequate to estimate the desired output values.
In the proposed approach, since the input data are not arbitrary large, the
ANNs literature shows that in most cases such problems of approximation by
ANNs can be solved with one hidden layer adequately (Adamopoulos et al.
2001; Vassilopoulos et al. 2007; Huitao et al. 2002; Garcia et al. 2004).
3.2. The Proposed ANN Methodology for Constructing Mutual

Funds’ Prediction Models
The proposed strategy consists of two basic procedures. The first

procedure is to identify the input to the ANN models, that is, the operating
characteristics and the economic variables in order to develop the performance
and the net asset value of mutual funds’ prediction method. This procedure is
described in detail in Section 2.
The second procedure employs two multilayer perceptrons (MLPs) trained
with the standard error back propagation algorithm. The advantage of using
MLPs is their ability to define a problem empirically by using sufficient
examples, rather than describing it analytically. MLPs can be regarded as an
effective and efficient nonparametric technique for estimating values of
unknown parameters (Adamopoulos et al. 2001; Vassilopoulos et al. 2007;
Huitao et al. 2002; Garcia et al. 2004). The theoretical framework of MLP
modeling is precisely discussed in Haykin (2008). Hence, the proposed
strategy uses input vector parameters and estimates the most suitable output
values, in order to enhance the prediction of the performance and the net asset
value of mutual funds. In the proposed approach, where the input data are not
arbitrary large, as stated in Section 3.1, one hidden layer will be adequate. In
order to verify experimentally this decision, exhaustive experiments with
many different MLP structures were carried out. All different structures were
compared with each other based on the root mean squared error (RMSE) value
measured. The structure which achieved the lowest RMSE was selected as the
most appropriate one and this proved to be the MLP having one hidden layer.
Figure 1. The general structure of the MLP used in the proposed methodology.
Figure 2. The basic steps of the proposed neural network optimization method.
So, as described in Section 2, the 15 operating characteristics and the 17

economic variables are used as inputs for the MLPs while the measured
responses, the performance and the net asset value of Greek equity mutual
funds, are used as outputs, respectively. The output of both MLPs used in the
proposed strategy, which structure is shown in Figure 1, can be formulated as
follows:
𝑣(𝑢) = 𝑔2 (∑𝐾 𝐿
𝑘=1 𝑤𝑘 ∙ 𝑔1 (∑𝑙=1 𝑤𝑘𝑙 ∙ 𝑢𝑙 + 𝑤0 ) + 𝑤0 ) (1)
is the input vector, wkl  R is

L 1
where v  R is the MLP’s output, u  R
the weight between input and hidden layers, wk  R R is the weight between
hidden and output layers, g1(·) is the activation function in the hidden layer,
g2(·) is the activation function in the output layer and w0 is the threshold used.
In our case, L equals the number of neurons in the input layer, K equals the
number of neurons in the hidden layer and w0 = 1.
A conventional choice for activation function at hidden layer is the
logistic sigmoid function (Haykin 2008). For the output layer the linear
activation function is used (Haykin 2008). The MLPs used in this contribution
are trained using the well known standard error back propagation algorithm. A
complete description of this algorithm can be found in Haykin (2008). In its
training function the amount the weight is changed in every epoch of the
learning phase is proportional to the negative gradient plus the previous weight
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.
4. THE GENETIC PROGRAMMING METHODOLOGY

4.1. Genetic Programming
Genetic programming (GP) is a domain-independent problem-solving

technique in which computer programs are evolved to solve, or approximately
solve, problems (Koza 1992). It is based on the Darwinian principle of
reproduction and survival of the fittest and is similar to the biological genetic
operations such as crossover and mutation. In GP the evolution operates on a
population of computer programs of varying sizes and shapes. GP starts with
an initial population of thousands or millions of randomly generated computer
programs, composed of the available programmatic ingredients and then
applies the principles of biological evolution to create a new (and often
improved) population of programs. The generation of this new population is
done in a domain-independent way using the Darwinian principle of survival
of the fittest, an analogue of the naturally-occurring genetic operation of
sexual recombination (crossover), and occasional mutation (Koza 1998). The
crossover operation is designed to create syntactically valid offspring
programs (given closure amongst the set of programmatic ingredients).
GP combines the expressive high-level symbolic representations of
computer programs with the near-optimal efficiency of learning of Holland’s
genetic algorithm. A computer program that solves (or approximately solves) a
given problem often emerges from this process (Koza 1998). GP for the
problem of system modeling doesn’t need an a priori knowledge of the model
structure like other techniques do. GP evolves a system model from scratch. It
doesn’t use a predefined model structure or a bank of model structures and
search just for the parameter values that fit better the data, like other data
driven methods. Instead, GP creates an initial population of models and
evolves those using genetic operators in order to find the mathematical
expression that best fit the data of the given system. That is, GP searches
simultaneously for the model structure and parameters. Genetic Programming
has been applied successfully in a number of financial time series modeling

and prediction problems (Chen 2002; Wang 2010; Wang and Kuo 2010).
4.2. The Proposed GP Methodology for Constructing Mutual

Funds’ Prediction Models
In order to implement the proposed GP methodology for predicting the

performance and the net asset value of Greek mutual funds an integrated GP
environment with a graphical user interface (GUI), called jGPModeling is
used. The jGPModeling environment is developed using the JAVA
programming language, and is an implementation of the steady-state GP
algorithm (Georgopoulos et al. 2008). That algorithm evolves tree based
structures that represent models of input – output relation of a system. In the
jGPModeling environment it is implemented a steady state GP, which is a
variation of the classic GP algorithm and in contrast with that, it doesn’t make
use of the generational evolutionary scheme. The steady state GP is selected
because the great multiprocessing capabilities that exhibits (Georgopoulos et
al. 2008). In jGPModeling every model is represented as a tree and its size is
limited to a maximum number of points (i.e., total number of functions and
terminals) or a maximum depth of the model tree. The models in the initial
population usually have very poor performance. Nevertheless, some
individuals will turn out to be somewhat more fit than others. These
differences in performance are then exploited by the GP algorithm.
The basic steps of the GP algorithm implemented in jGPModeling
environment are the following:
1 Generating at random the initial population

2. Selecting at random a subset of the population for a tournament
(tournament selection)
3. Evaluating the members of this subset
4. Selecting tournament winners
5. Applying genetic operators of mutation and crossover on the
tournament winners
6. Replacing the tournament losers by the winner offsprings in the
population
7. If the termination criterion is not fulfilled return to step 2
8. Returning the best ever found individual (model)
For a more detailed description of the jGPModeling environment the

reader can refer to Georgopoulos et al. 2008.
It has been demonstrated in the literature that the performance of a GP
algorithm remains unchanged after a specific number of generations while the
average size of the tree models continues to increase till it reaches the value of
the maximum depth of the model tree, defined at the beginning of the
execution. This phenomenon is called bloat (Banzhaf et al. 1998). In our
application, exhaustive experiments showed that after a specific number of
generations, although the size of the model trees increases their performance is
not improved. This occurs because the size of the subtrees which do not affect
the performance of the algorithm continues to increase. In order to cope with
this phenomenon, we have added a tree pruning procedure based on the
following logical rules:
 In case the expression of an “if” node is stable, the whole tree is

replaced by the active subtree
 In case the subtree of an “addition” or “subtraction” node is always
equal to zero, this subtree is replaced by its active part.
 In case the result of a subtree is equal to a constant number, this
subtree is replaced by a terminal node.
 In case a subtree has a specific function as a parent node and this
parent node has the respective inverse function as its parent node,
these nodes are deleted.
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.
Figure 3. The basic steps of the proposed of GP optimization method.

5. RESULTS
5.1. The Application of the Proposed Techniques
As we have already mentioned above, the input data used in order to

develop the ANN and GP models relate to seven years. The first six years are
used in order to train/create our models while the last year for testing our
models. The input and target values for both ANN and GP models are
normalized in the vector [0 1], so that the weights would not blow up during
the training of our models. In order to normalise our data we used a very
simply method that is frequently used in the literature and it is based on the
deviation of our variables with their maximum value.
The structure of the MLPs used, which, as described in Section 3.2, has 15
inputs for the first MLP and 17 inputs for the second MLP, one hidden layer
with K neurons for both MLPs and a single output neuron (the performance
and the net asset value of Greek mutual funds for the first and the second
MLP, respectively), may be designed as a “L-K-1” net (Figure 1). Next,
several ANN parameters have to be determined, including: (1) stopping
criterion, (2) number neurons in the hidden layer, (3) learning rate, and (4)
momentum.
As stopping criterion, the root mean squared error (RMSE) value is
selected (Huitao et al. 2002; Garcia et al. 2004). This means that the error
function, which measures the RMSE difference between calculated (predicted)
and desired (actual) output values, should be less than a predefined value,
which in our case was set to 0.1.
The number of neurons in the hidden layer provides the dimension of
hidden space. Having an appropriate hidden dimension could provide the
necessary complexity to accurately model and estimate the performance and
the net asset value of Greek mutual funds. In order to determine the number of
neurons in the hidden layer we used the theorem of Kolmogorov (Hornik et al.
1990; Kurkov 1992; Poggio and Girosi 1990; Yao 1993), which claims that
2 × N + 1 neurons in the hidden layer should suffice, where N is the number of
inputs. In order to verify experimentally this decision, exhaustive experiments
with many different MLP structures are carried out. All different structures are
compared with each other based on the RMSE value measured. The structure
which achieved the lowest RMSE is selected as the most appropriate one, that
is, 4-21-1 for the first MLP and 4-18-1 for the second MLP, respectively.
The most appropriate values of the rest MLP’s training parameters
(learning rate and momentum) are also defined by carrying out exhaustive
experiments. Many different combinations of MLP’s training parameters’

values are compared with each other based on the RMSE value measured. The
values which achieved the lowest RMSE are selected as the most appropriate
ones. The values of the parameters used for the training and testing phases of
both MLPs are shown in Table 5.1. The most appropriate MLPs found, are
afterwards used to construct ANN models for predicting the performance and
the net asset value of Greek mutual funds.
Testing the ANN with similar data as that used in the training set is one of
the few methods used to verify that the network has adequately learned the
input domain. In most instances, such traditional testing techniques prove
adequate for the acceptance of an ANN system. In ANN modeling
applications, the training set is used as the primary set of data that is applied to
the ANN for learning and adaptation. The validation set is used to further
refine the ANN construction. The testing set is then used to determine the
performance of the ANN by computation of an error metric. In the presented
approach, the testing data set which are not included in the training data set of
each MLP are used in order to check each MLP’s ability to generalize, that is,
to predict with low RMSE, values which are not used in its training phase.
This training-validating-testing approach is the first, and often the only, option
system developers consider for the assessment of an ANN (Skapura and
Gordon 1996; Ripley 1996). The assessment is accomplished by the repeated
application of ANN training data, followed by an application of ANN testing
data to determine whether the ANN is acceptable. In the presented approach
where the input data are not arbitrary large, training and testing phases suffice,
as shown by experimental results. So, we decided not to use cross-validation
because, according to experimental results, we observed that using cross-
validation just added complexity to the proposed methodology without
increasing its accuracy significantly.
In order not to affect the training and testing phases of the MLP we
decided to choose both training and testing data sets randomly. We followed
the suggestions reported in Huitao et al. (2002) and Garcia et al. (2004) and
used 85% of the input data set for training and 15% for testing. Training and
testing phases for both MLPs are conducted using the MATLAB R2010a
software.
The structure of the GP algorithm used in order to create prediction
models for the performance and the net asset value of Greek mutual funds is
the one described in Section 4.2. As stated before, the GP algorithm is
implemented using the jGPModeling environment. To evaluate the
effectiveness and the relative performance of ANN and the GP models,
different error measures and success rates are computed for the high, medium
and low return and risk mutual funds’ categories.
5.2. Comparative Results
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
Parameter First MLP Second MLP

Structure 4-21-1 4-18-1
Hidden layer’s Sigmoid Sigmoid
activation function
Output layer’s Linear Linear
activation function
Training algorithm Back propagation – Back propagation –
conjugate gradient descent conjugate gradient descent
Learning rate 0.25 0.2
Momentum 0.15 0.15
Stopping criterion RMSE<0.1 RMSE<0.1
Training epochs 130 110
Table 5.2. Error measures when the explanatory variables are

the operating characteristics of MFs
Target GEOMEAN NAV

values
Patterns Return Risk Return Risk
High ANN GP ANN GP ANN GP ANN GP
MAPE 0.23 0.18 4.00 0.59 0.28 1.23 0.77 1.64
MeAPE 0.08 0.09 0.52 0.36 0.04 1.87 0.37 2.12
GeMAPE 0.16 0.07 0.47 0.32 0.22 1.06 0.35 2.81
MaxMAPE 1.71 0.57 117.22 3.24 0.75 4.37 31.45 29.25
SdMAPE 7.64 7.06 9.12 4.99 8.69 12.59 5.57 6.83
Medium
MAPE 1.87 0.58 4.91 0.45 0.88 1.05 4.38 3.38
MeAPE 0.22 0.18 0.27 0.07 0.01 1.94 0.01 0.78
GeMAPE 0.53 0.22 0.65 0.11 0.41 2.12 0.44 1.15
MaxMAPE 16.94 8.89 98.41 17.02 6.58 9.32 77.58 36.28
SdMAPE 4.87 2.88 9.98 4.85 4.70 5.51 11.04 10.83
Low
MAPE 5.57 0.64 4.85 0.72 4.10 7.97 4.85 4.71
MeAPE 0.13 0.32 0.22 0.26 0.10 1.02 0.22 0.93
GeMAPE 0.92 0.49 0.97 0.23 1.11 1.84 0.97 1.03
MaxMAPE 66.7 2.45 53.72 3.34 38.85 34.41 43.72 38.66
SdMAPE 5.40 11.95 7.52 38.56 6.43 7.65 7.52 12.32
Table 5.3. Error measures when the explanatory variables are

economic variables
Target GEOMEAN NAV

values
MAPE 10.87 0.32 13.16 1.11 10.57 10.88 7.26 16.91
MeAPE 4.27 0.18 1.53 0.54 2.5 4.27 0.93 2.46
GeMAPE 3.46 0.11 2.3 0.67 2.58 3.46 1.52 3.2
MaxMAPE 50.45 0.94 146.91 5.12 47.83 24.15 157.23 100.67
SdMAPE 112.95 9.82 104.21 47.33 94.21 67.45 63.67 42.42
Medium
MAPE 10.23 0.76 15.13 0.83 10.71 12.08 19.69 15.13
MeAPE 1.56 0.67 3.17 0.11 1.91 2.28 2.94 3.17
GeMAPE 2.27 0.52 3.61 0.24 2.45 2.51 4.21 3.61
MaxMAPE 56.76 11.23 158.43 10.69 418.33 107.93 350.74 220.62
SdMAPE 39.78 5.04 124.36 7.17 398.45 221.65 140.22 120.48
Low
MAPE 18.98 0.92 13.13 1.56 13.5 41.19 12.1 12.1
MeAPE 1.77 0.71 2.12 0.52 0.91 3.35 2.21 2.21
GeMAPE 2.55 1.34 2.19 0.67 1.68 4.34 2.29 2.29
MaxMAPE 129.44 5.69 94.56 6.78 83.22 63.54 73.44 43.16
SdMAPE 67.34 17.39 38.65 40.35 26.12 22.16 28.32 24.83
Table 5.4. Success rate when the explanatory variables are the operating characteristics of MFs
Target values GEOMEAN NAV

Success over 80% 55.55% 62.50% 27.77% 29.41% 44.44% 24.56% 33.33% 19.12%
Up and down prediction 100% 100% 40% 100% 100% 75.00% 80% 50.00%
Medium
Success over 80% 20% 57.14 16% 71.43% 20% 23.00% 24% 19.00%
Low
Success over 80% 27.77% 28.12% 16.68% 42.10% 11.11% 22.67% 16.66% 18.34%
Table 5.5. Success rate when the explanatory variables are economic variables

Success over 80% 32.66% 50.00% 22.11% 25.11% 32.22% 16.70% 27.66% 11.11%
Up and down prediction 80% 100% 37.33% 100% 80% 60.00% 60% 33.33%
Medium
Success over 80% 16% 45.45 12% 65.33% 14% 16.00% 14% 12.00%
Up and down prediction 73.84% 100% 75% 100% 71.53% 53.85% 75% 25.00%
Low
Success over 80% 19.66% 25.24% 15.55% 33.67% 19.66% 16.67% 14.11% 11.11%
Up and down prediction 62% 80.00% 61% 78.00% 70% 40.00% 65% 11.11%
Mutual Fund Prediction Models Using Artificial Neural Networks … 111
Table 5.6. Error measures of reduced models when the explanatory

variables are the operating characteristics of MFs

MAPE 0.19 0.17 3.68 0.43 0.17 1.01 0.55 1.11
MeAPE 0.06 0.07 0.39 0.31 0.03 1.47 0.28 1.93
GeMAPE 0.12 0.06 0.32 0.28 0.18 0.95 0.26 2.25
MaxMAPE 1.48 0.38 99.86 2.35 0.61 3.7 26.23 21.58
SdMAPE 6.92 5.12 9.12 4.14 7.74 9.85 5.14 5.76
Medium
MAPE 1.61 0.45 4.13 0.37 0.69 0.95 3.91 3.15
MeAPE 0.18 0.16 0.19 0.06 0.01 1.61 0.01 0.69
GeMAPE 0.41 0.19 0.53 0.1 0.34 1.23 0.38 0.97
MaxMAPE 15.14 6.92 80.47 13.42 6.15 7.76 65.98 29.81
SdMAPE 3.91 2.14 8.79 3.87 4.12 4.78 10.28 8.32
Low
MAPE 5.19 0.57 4.18 0.65 3.72 6.58 4.23 4.24
MeAPE 0.09 0.31 0.13 0.24 0.06 0.89 0.17 0.78
GeMAPE 0.84 0.42 0.78 0.21 0.95 1.41 0.78 0.87
MaxMAPE 55.3 2.15 44.58 3.06 31.67 27.16 31.65 31.36
SdMAPE 4.73 9.93 6.89 30.65 5.87 9.53 6.94 15.27
Table 5.7. Error measures of reduced models when the explanatory

variables are economic variables

MAPE 8.33 0.29 10.11 1.03 8.21 9.83 6.22 31.14
MeAPE 3.89 0.18 1.19 0.51 2.11 4.03 0.84 2.18
GeMAPE 3.02 0.08 1.98 0.62 2.08 3.12 1.31 2.95
MaxMAPE 42.33 0.88 127.18 4.34 37.76 20.56 129.73 87.78
SdMAPE 92.34 8.25 97.55 37.85 71.99 58.52 54.29 38.21
Medium
MAPE 9.06 0.65 13.38 0.71 9.12 10.23 17.03 12.38
MeAPE 1.17 0.6 2.22 0.09 1.32 2.03 2.25 2.72
GeMAPE 1.87 0.46 3.05 0.21 2.01 2.11 3.85 2.91
MaxMAPE 47.69 10.31 123.72 10.26 378.44 97.35 321.56 197.27
SdMAPE 30.12 4.43 106.63 7.68 356.58 196.52 118.89 104.88
Low
MAPE 13.56 0.85 10.94 1.23 11.56 41.19 10.18 10.52
MeAPE 1.19 0.63 1.72 0.43 0.71 3.35 2.12 2.03
GeMAPE 2.03 1.1 1.81 0.52 1.21 4.34 2.02 2.11
MaxMAPE 108.47 5.21 78.81 5.89 71.91 63.54 60.23 36.68
SdMAPE 56.82 18.92 29.66 35.51 20.22 22.16 21.11 22.35
Initially, the objective is to develop prediction models incorporating the

initial set of the 15 fund-specific historical operating characteristics and the 17
economic variables. In an attempt to reduce dimensionality and to incorporate
the most significant volume of information, we select the most relevant criteria
which best describe the performance of mutual funds and the economic
environment based on factor analysis through the SPSS statistical software.
Tables 5.6 and 5.7 presents the error measures of both ANN and GP of a
reduced set of explanatory variables while Tables 5.8 and 5.9 presents the
success rates of the reduced models, of both the operating characteristics and
economic variables respectively.
The application of factor analysis resulted in the development of six
factors that account for 92.103% of the total variance in the data regarding the
operating characteristics. As far as the economic variables are concerned,
factor analysis resulted in the development of four factors that account for
96.299% of the total variance. The selection of the variables is performed on
the basis of their factor loadings. In particular, the variables with the highest
factor loadings are selected from each factor, thus leading to the selection of
eight variables (6, 5, 1, 15, 12, 11, 14 and 4, from Table 2.1) for operating
characteristics and five variables (2, 7, 6, 11 and 10, from Table 2.2) for the
economic variables. These variables are the ones with the highest loadings on
the developed factors (in all cases the factor loadings for the selected variables
are higher than 0.9, in absolute terms). In this respect, these criteria describe in
the best way the developed factors.
Tables 5.6 and 5.7 present the error measures 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 for the reduced models. Accordingly, in
Tables 5.8 and 5.9 are presenting the success rates for the reduced models.
To test for the statistical significance of the difference in the performance
of the ANN and GP prediction techniques, we applied the Friedman test. The
Friedman test is conducted to test for differences in the forecasting
effectiveness as measured by the MAPE (which gives the opportunity to
facilitate comparisons across variables with different scales), and indicated
significant differences between the two techniques at the 5% level of
significance.
As seen from the tables, an overall result is that neither ANN nor GP
outperforms in all the examined cases for both the full and reduced prediction
models. The reduced models present lower error measures except from four
cases, while there are few cases where we have the same results. To some
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).
Table 5.8. Success rate of reduced models when the explanatory variables
are the operating characteristics of MFs
Target GEOMEAN NAV

values
Success 59.76% 65% 31.66% 33.33% 54.47% 28.67% 42.66% 23.25%
over 80%
Up and 100% 100% 48% 100% 100% 79% 82% 57%
down
prediction
Medium
Success 28% 62.45 19% 78.38% 25% 28% 28% 24%
over 80%
Up and 100% 100% 100% 100% 100% 82% 100% 50%
down
prediction
Low
Success 32.85% 31% 19.97% 45% 14.25% 27.34% 18.23% 19.86%
over 80%
Up and 82% 100% 100% 100% 100% 69% 100% 19.12%
down
prediction
Table 5.9. Success rate of reduced models when the explanatory variables
are economic variables
Target GEOMEAN NAV

values
Success 35.55% 56% 24.28% 28% 35.67% 19.50% 30.77% 14.23%
over 80%
Up and 82% 100% 41.33% 100% 83% 66% 66% 36.67%
down
prediction
Medium
Success 18% 52% 15% 69% 17% 18% 16% 15%
over 80%
Up and 77.12 100% 78% 100% 76.56% 60.00% 79% 29%
down
prediction
Low
Success 23.33% 28.45% 18.77% 36.67% 21.33% 19% 17.22% 15.45%
over 80%
Up and 65% 84% 63% 81% 74% 44% 68% 14.85%
down
prediction
Another conclusion refers to the out performance of our forecasting results

when the predictions are based on the return category for high and medium
return pattern regarding to operating characteristics of mutual funds. The
results of the low pattern are mixed. However, this is not the case for the
economic variables where we have mixed results. The reduced models resulted
to the same conclusions. Furthermore, our results showed that the high risk
categories resulted higher success rates, because in these categories we have
the aggressive funds. It is commonly understood, that the less aggressive funds
are easier to be forecasted because of their correlation with the examined
benchmark (stock market index). This outcome is in accordance with the
forecasting results of Chiang et al., (1996), where all the applied models tend
to have higher forecast errors for the more aggressive mutual funds. In our
application, it’s observed an inconsistency for the GEOMEAN prediction
independently the explanatory variables under consideration, that the high risk
categories resulted lower percentage with success over 80% using GP. For the
reduced models the same inconsistency refers to the NAV prediction. It is also
believed that the frequent, alternating rise and fall of the market in the
examined period will significantly increase inconsistencies in the prediction
ability of our models.
Furthermore, all inconsistencies presented in our results based on the
return and risk categories may occur as a result of a change in a given fund’s
management or investment objective. It is important to understand that if a
fund changes objective, return and risk pattern, for example from high to low
during the examined years, the prediction models may forecast poorly.
According to Chiang et al., (1996), this is due to the fact that neural networks
train to find patterns and complex relationships. And he underlines that from
the detailed results, one may possible conclude that certain funds held a steady
objective during this time period. Finally, it is worth mentioning that although
both prediction models resulted high percentage of accuracy in the prediction
of the up and down movements of fund values for the next time period (Tables
5.4, 5.5, 5.8 and 5.9), the percentage in which both prediction models
approximate the actual fund values over 80% accuracy is rather poor (Tables
5.4 and 5.8). These results are even worse when the explanatory variables are
the macroeconomic characteristics of the market (Tables 5.5 and 5.9). Based
on these experimental findings we conclude that fund’s operating
characteristics significantly influence its operating performance and lead to
better forecasting results. The same stands for the reduced models.
Concluding, although none of the prediction techniques outperforms the other
in all cases, our results confirm that both ANN and GP are acceptable methods
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
forecasting results may also be due to the methodology employed, as well as

on the sample data examined. Furthermore, we believe that the architecture
structure of our models is critical for the obtained results. ANN are distinct
from GP, but may be used next to each other to deal with separate aspects of
the same problem.
Future research should be conducted in financial markets to determine if
other macroeconomic variables (such as stock prices, money, Treasury bill,
etc.) and operational characteristics (such as mutual fund’s expense ratio,
price-earnings ratio, price-book ratio, etc.) added to ANN and GP models
might further enhance their prediction performance. Also, the improvement of
existing methods and/or the creation of new ones for the prediction of Greek
mutual fund performance will be main part of our future work.
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
as follows: R  T 1  R  where R is the geometric mean of period T.

T
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:
i. Mutual funds share is referred to the capital of mutual funds which is

divided into units, not stocks or bonds. When a person is subscribed in
a mutual fund, the amount deposited is transformed into units, which
do not change, unless money is added or withdrawn. In this analysis

we take into account the mean value of shares.
ii. The return on a mutual fund investment in a given time period is
calculated by taking into account the change in a fund’s net asset
value. Precisely, the change in a fund’s net asset value refers to the
market value of securities the funds hold, divided by the number of
the fund’s shares during a given time period, with the assumption of
reinvestment of all income and capital-gains distributions, and
dividing it by the original net asset value. The fund’s return in period t
is defined as follows: Rpt  NAVt  DISTt  NAVt 1 , where R pt is the
NAVt 1
return of mutual fund in period t, NAVt is the closing net asset value
of the fund on the last trading day of the period t, NAVt 1 is the
closing net asset value of the fund on the last trading day of the period
t-1 and DISTt is the income and capital distributions (dividend of the
fund) taken during period t.
iii. Mean value of NAV
iv. The arithmetic return on a MF (mean return) investment includes both
income and capital gains or losses. It is calculated as the percentage
change in the net asset value of a MF over a given period taking also
into account the dividends paid.
v. The standard deviation is the most commonly used measure of
variability. For a MF the standard deviation s is used to measure the
variability of daily returns presenting the total risk of the fund. The
standard deviation of daily returns is transformed to refer to the
sample period (n observations) under consideration using the simple
formula ns .
vi. A traditional total performance measures, is Sharpe index (Sharpe
1966), which is used to measure the expected return of a fund per unit
of risk. This measure is defined by the following ratio  R  R f   .
The evaluation of MFs with this index shows that a MF with higher
performance per unit of risk is the best-managed fund, while a MF
with lower performance per unit of risk is the worst managed fund.
vii. The beta (β) coefficient is a measure of fund risk in relation to the
market risk. It is called systematic risk and the CAPM implies that it
is crucial in determining the prices of risky assets. For the calculation
of beta (β) coefficient the well-known capital asset pricing model is

used: R     RM   , where α is a coefficient measuring the return
of a fund when the market is constant, and  is an error term that
represents the impact of non-systematic factors that are independent
from the market fluctuations.
viii. Another traditional total performance measure, as Sharpe index is the
Treynor index (Treynor 1965) which is used to measure the expected
return of a fund per unit of risk, through the following ratio
 R  R f   (Treynor’s index). The evaluation of MFs with this indix
shows that a MF with higher performance per unit of risk is the best-
managed fund, while a MF with lower performance per unit of risk is
the worst managed fund.
ix. Annual return of Athens Stock Exchange (Annual return on ASE).
The market return which is calculated in period t is defined as
follows:
Closed end pricet  Closed end pricet 1
RMt  .
Closed end pricet 1
x. Mean value of ASE return
xi. Standard deviation of ASE returns
xii. The coefficient of variation (CV) is defined as the ratio of the
standard deviation to the absolute value of mean. It is only defined for
non-zero mean and the absolute value is taken to ensure it is always
positive.
xiii. Another well-known measure of risk is Value at Risk (VaR). VaR
measures the maximum losses that an investor can have in a certain
time period for a given confidence level. The calculation of VaR is
based on the variance-covariance (VC) approach (Jorion 2000).
Given, the daily standard deviation  and the expected daily return
R of a MF, its VaR for a time period of N days for a confidence
level α is calculated as VaR  NR  Z N , where the term Z is
obtained from the standard normal cumulative distribution as
P  Z  Za   1   . In this study, VaR is calculated for the three year
period of the analysis with  2.5% ( Z  1.96 ).
xiv. Modigliani and Modigliani (1997) proposed an alternative measure of
risk-adjusted performance according to which, each portfolio is
adjusted to the level of risk of the market benchmark, thereby
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.
The definition of the 17 economic variables (independent variables) is

given in Table 1 below:
Table 1. The economic variables used in the analysis
Input Economic Variable Definition-Notes [Units]

1 Gross domestic product, constant Gross domestic product at constant market
prices prices [Billions Euros]
2 Gross domestic product, current Gross domestic product at market prices
prices (Billions Euros)
3 Gross domestic product per See notes for: Gross domestic product,
capita, constant prices constant prices (Euros) Population
(Persons) [Euros]
4 Gross domestic product per See notes for: Gross domestic product,
capita, current prices current prices (Euros) Population
(Persons) [Euros]
5 Gross domestic product based on See notes for: Gross domestic product,
purchasing-power-parity (PPP) current prices (Euros) [Percent]
share of world total
6 Inflation, average consumer From average annual inflation, consumer
prices prices (Index, 2000=100) [Annual percent
change]
7 General government balance Includes: Central Government, State
Government, Local Government, Social
Security Funds [Billions Euros]
8 General government structural Includes: Central Government, State
balance Government, Local Government, Social
Security Funds [Billions Euros]
9 Current account balance See notes for: Gross domestic product,
current prices (Euros) Current account
balance (Euros) [Percent of GDP]
Input Economic Variable Definition-Notes [Units]

10 Real GDP Growth GDP growth on an annual basis adjusted
for inflation and expressed as a percent
[Annual percent change]
11 Gross Fixed Total Investments Includes: construction, equipment [Annual
percent change]
12 Industrial Production Includes manufacturing, mining, and
construction [Annual percent change]
13 Unemployment rate Percent of total labor force [Annual
percent change]
14 Employment Employment, national definition
[Millions]
15 Consumer Price Index Harmonized index of consumer prices
annual average [Index, 2000=100]
16 Credit Expansion Private Sector [Annual percent change]
17 Deficit-General Government Government deficit-to-GDP ratio [Percent
of GDP]
Sources: International Monetary Fund, Statistical Office of Greece, Bank of Greece.
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.
Bhattacharyya, S., Pictet, O. V. & Zumbach, G. (2002). Knowledge-Intensive

Genetic Discovery in Foreign Exchange Markets. IEEE Transactions on
Evolutionary Computation, 6(2), 169–181.
Bilson, C., Frino, A. & Healy, R. (2005). Australian retail fund performance
persistence, Accounting and Finance, 45, 25-42.
Blake, C. R. & Morey, M. R. (2000). Morningstar Ratings and Mutual Fund
Performance. Journal of Financial and Quantitative Analysis, 35, 451-
483.
Blum, E. K. & Li, L. K. (1991). Approximation theory and feedforward
networks. Neural Networks, 4(4), 511–515.
Bollen, N. P. & Busse, J. A. (2005). Short-term persistence in mutual fund
performance. Review of Financial Studies, 18(2), 569-597.
Carhart, M. (1997). On the persistence in mutual fund performance. The
Carhart, M. M., Carpenter, J. N., Lynch, A. W. & Musto, D. K. (2002). Mutual
fund Survivorship. Review of Financial Studies, 15, 1439-1463.
Chen, S. H. (2002). Genetic algorithms and genetic programming in
computational finance. Dordrecht, The Netherlands: Kluwer academic
publishers group.
Chen S. H., Kuo T. W. & Hoi, K. M. (2008). Genetic Programming and
Financial Trading: How Much about “What we Know”. In C. Zopounidis,
M. Doumpos, & P. M. Pardlos (Eds.), Handbook of Financial
Engineeering, (pp. 99–54). New York, USA: Springer.
Chen, S. H. & Navet, N. (2006). Pretests for genetic-programming evolved
trading programs: “zero-intelligence” strategies and lottery trading, 13th
International Conference, ICONIP 2006, Hong Kong, China, October 3-6,
450–460.
Cheng, C. H., Chen, T. L. & Wei, L. Y. (2010). A hybrid model based on
rough sets theory and genetic algorithms for stock price forecasting.
Information Sciences, 180, 1610-1629.
Chiang, W. C. & Urban, T. L. (1996). A neural network approach to mutual
fund net asset value forecasting. Omega, 24(2), 205–215.
Cristensen, M. (2005). Danish mutual fund performance, selectivity, market
timing and persistance, Working paper.
Drakos, K. & Zachouris, P. (2007). On persistence in the Greek equity fund
market. Global Business and Economics Review, 9(1), 75-91.
Davis, L. (1991). Handbook of genetic algorithms. New York: Van Nostrand
Reinhold.
Ferruz, L., Sarto, J. L. & Andreu, L. (2007). A comparison between German

and Spanish equity funds markets. Journal of Asset Management, 8, 147-
151.
Fadlalla, A. & Lin, C. H. (2001). An analysis of the applications of neural
networks in finance. Interfaces, 31(4), 112–122.
Fletcher, J. (1999). The evaluation of the performance of UK American Unit
Trusts. International Review of Economics and Finance, 8, 455-466.
Fletcher, J. & Forbes, D. (2002). An exploration of the persistence of UK unit
trust performance. Journal of Empirical Finance, 9, 475-493.
Garcia, S. F., Sanchez, M. J. & Rodriguez-Delgado, M. A. (2004).
Optimization of the separation of a group of triazine herbicides by
micellar capillary electrophoresis using experimental design and artificial
neural networks, Electrophoresis, 25(7-8), 1042–1050.
Georgopoulos, E. F., Zarogiannis, G. P., Adamopoulos, Vassilopoulos, A. V.
& Likothanassis, S. D. (2008). A Genetic Programming Environment for
System Modeling, Artificial Intelligence: Theories, Models and
Applications, Lecture Notes in Computer Science, 5138, 85–96.
Giamouridis, D. & Sakellariou, K. (2008). Short-Term Persistence in Greek
Mutual Fund Performance, (January 2, 2008). Available at SSRN: http://
ssrn.com/abstract=1080912 or http://dx.doi.org/10.2139/ssrn.1080912.
Goetzmann, W. N. & Ibbotson, R. (1994). Do winners repeat? Patterns in
mutual fund return behavior. The Journal of Portfolio Management, 20(2),
9–18.
Cremers, M. & Petajisto, A. (2009). How active is your fund managers ? A
new measure that predicts performance. Review of Financial Studis, 22,
3329-3365.
Grinblatt, M. & Titman, S. (1992). The persistence of mutual fund
performance. The Journal of Finance, 47(5), 1977–1984.
Gruber, M. J. (1996). Another puzzle: The growth in actively managed mutual
funds. Journal of Finance., 51(3), 783–810.
Guresen, E., Kayakutlu, G. & Daim, T. U. (2011). Using artificial neural
network models in stock market index prediction. Experts Systems with
Applications, 38, 10389-10397.
Hallahan, T. A. & Faff, R. W. (2001). Induced persistence or reversals in fund
performance ? The effect of survivorship bias. Applied Financial
Economics, 11, 119-126.
Handjinicolaou, G. (1980). The performance of Greek mutual funds in the
period 1973-76: A case of internationally diversified portfolios, Spoudai,
11(3/4), 381-391.
Haykin, S. (2008). Neural Networks and Learning Machines. 3rd edition.

Prentice Hall International.
Hornik, K., Stinchcombe, M. & White, H. (1990). Universal approximation of
an unknown mapping and its derivatives using multilayer feed forward
networks. Neural Networks, 3(5), 551–560.
Huarng, K. & Yu, H. K. (2006). The application of neural networks to forecast
fuzzy time series. Physica A, 363, 481-491.
Huitao, L., Ketai, W., Hongping, X., Xingguo, C. & Zhicte, H. (2002).
Application of experimental design and artificial neural networks to
separation and determination of active components in traditional Chinese
medicinal preparations by capillary electrophoresis. Chromatographia,
55(9-10), 579–583.
Indro, D. C., Jiang, C. X., Patuwo, B. E. & Zhang, G. P. (1999). Predicting
mutual fund performance using artificial neural networks, Omega, 27(3),
373–380.
Jan, Y. C. & Hung, M. W. (2004). Short-run and long-run persistence in
mutual funds. Journal of Investing, 13, 67-71.
Jain, P. C. & Wu, J. S. (2000). Truth in mutual fund advertising: Evidence on
future performance and fund flows. Journal of Finance, 55, 937-958.
Jensen, C. M. (1968). The performance of mutual funds in the period 1945-
1964. Journal of Finance, 23(2), 389–416.
Jorion, P. (2000). Value at Risk: The New Benchmark for Managing Financial
Risk. 2nd edition, New York, USA: McGraw-Hill.
Kaboudan, M. (2000). Genetic Programming Prediction of Stock Prices.
Computational Economics, 16(3), 207–236.
Kacperczyk, M., Sialm, C. & Zheng, L. (2008). Unobserved Actions of
Mutual Funds, Reviwe of Financial Studies, 21, 2379-2416.
Kahn, R. N. & Rudd, A. (1995). Does historical performance predicts future
performance? Financial Analysts Journal, 51(6), 43–52.
Kim, K. & Han, I. (2000). Genetic algorithms approach to feature
discretization in artificial neural networks for prediction of stock index.
Expert System with Application, 19, 125-132.
Kimoto, T., Asaakawa, K., Yoda, M. & Takeoka, M. (1990). Stock market
prediction system with modular neural network, in: Proceedings of the
International Joint Conference on Neural Networks, San Diego,
California, 1-6.
Koulis, A., Beneki, C., Adam, M. & Botsaris, C. (2011). An assessment of the
performance of Greek mutual equity funds selectivity and market timing,
Applied Mathematical Sciences, 5(4), 159-171.
Klein, B. D. & Rossin, D. F. (1999). Data quality in neural network models:

effect of error rate and magnitude of error on predictive accuracy, Omega,
27(5), 569–582.
Koza, J. R. (1992). Genetic Programming: On the Programming of Computers
by Means of Natural Selection. Cambridge, MA, USA: The MIT Press.
Koza, J. R. (1998). Genetic Algorithms and Genetic Programming at Stanford
1998. Stanford, CA, USA: Stanford University Bookstore.
Kurkov, V. (1992). Kolmogorov’s theorem and multilayer neural networks.
Neural Networks, 5(3), 501–506.
Mahfound, S. & Mani, G. (1996). Financial forecasting using genetic
algorithms. Applied Artificial Intelligence, 10, 543-565.
Milonas, N. (1999) Greek mutual funds-Theory and practice, Greek Bank
Association, Sakkoulas Publishing, Athens.
Modigliani, F. & Modigliani L. (1997). Risk-adjusted performance. Journal of
Portfolio Management, 23(2), 45–54.
Morey, M. R. (2005). The kiss of death: A 5-star Morningstar mutual fund
rating? Journal of Investment Management, 3, 41-52.
Pattarin, F., Paterlini, S. & Minerva, T. (2004). Clustering financial time
series: an application to mutual funds style analysis, Computational
Statistics & Data Analysis, 47(2), 353–372.
Philippas, N. (2001). An empirical evaluation of the performance of the Greek
fund managers, Spoudai, University of Piraeous, 51(1/2), 142-161.
Philpot, J., Hearth, D. & Rimbey, J. (2000). The performance persistence and
manager skill in non-conventional bond mutual funds. Financial Services
Review, 9, 247-258.
Pendaraki, K., Doumpos M. & Zopounidis, C. (2003). Assessing equity funds’
performance using a multicriteria methodology: A comparative analysis,
South Eastern Europe Journal of Economics, 1, 85-104.
Pendaraki, K., Zopounidis C. & Doumpos, M. (2005). A multicriteria
methodology and an application to the Greek market of equity mutual
funds, European Journal of Operational Research, 163, 462-481.
Pendaraki, K. (2012). Mutual fund performance evaluation using data
envelopment analysis with higher moments, J. Appl. Finance Bank, 2, 97-
112.
Pendaraki, K. (2015). Mutual fund performance benchmarking using a
quadratic directional distance function approach, International Journal of
Financial Engineering and Risk Management, 2(1), 30-47.
Pendaraki, K. & Spanoudakis, N. (2012). An interactive tool for mutual funds

portfolio composition using argumentation, Journal of Business,
Economics and Finance, 1(3), 33-51.
Pendaraki, K. & Tsagarakis, K. (2016). Linear Regression versus Fuzzy Linear
Regression: Does it Make a Difference in the Evaluation of the
Performance of Mutual Fund Managers?, (Chapter 11) in C.L. Dunis et al.
(eds.), Artificial Intelligence in Financial Markets, New Developments in
Quantitative Trading and Asset Management, Palgrave Macmillan
DOI:10. 057/978-1-137-48880-0.
Poggio, R. T. & Girosi, F. (1990). Networks for approximation and learning.
Proceedings of the IEEE, 78(9), 1481–1497.
Porter, G. E. & Trifts, J. W. (1998). The performance persistence of
experienced mutual fund managers. Financial Services Review, 7, 57-68.
Prather, L., Bertin, W. & Henker, T. (2004). Mutual fund characteristics,
managerial attributes, and fund performance. Review of Financial
Economics, 13, 315-349.
Rafiei, F. M., Manzari, S. M. & Bostanian, B. (2011). Financial health
prediction models using artificial neural networks, genetic algorithm and
multivariate discriminant analysis: Iranian evidence. Experts Systems with
Ripley, B. D. (1996). Pattern Recognition and Neural Networks. Cambridge,
UK: Cambridge University Press.
Roodposhti, F. R., Chavoshi, K., Saber, E. & Ali, Bashirpour. (2016).
Optimization of the Mutual-Fund Portfolio of Tehran Stock Exchange
Using Artificial Neural Networks and Genetic Algorithm. International
Business Management, 10, 2249-2256.
Sharpe, W. F. (1996). Mutual fund performance. Journal of Business, 39(1),
119–138.
Skapura, D. M. & Gordon, P. S. (1996). Building Neural Networks. New
York, USA: Addison-Wesley.
Sorros, J. (2001). Equity mutual fund managers’ performance in Greece,
Managerial Finance, 27(6), 68-74.
Thanou, E. (2008). Mutual fund evaluation during up and down market
conditions: The case of Greek equity mutual funds, International
Research Journal of Finance and Economics, 13, 84-93.
Tsai, T. J., Yang, C. B. & Peng, Y. H. (2011). Genetic algorithms for the
investment of the mutual fund with global trend indicator, Expert Systems
with Applications, 38(3), 1697–1701.
Treynor, J. L. (1965). How to rate management of investment funds. Harvard

Business Review, 43(1), 63–75.
Vassilopoulos, A. P., Georgopoulos, E. F. & Dionysopoulos, V. (2007).
Artificial neural networks in spectrum fatigue life prediction of composite
materials, International Journal of Fatigue, 29(1), 20–29.
Vidal-García, J. (2013). The Persistence of European Mutual Fund
Performance. Research in International Business and Finance, 28, 45–67.
Yao, X. (1993). Evolutionary artificial neural networks, International Journal
of Neural Systems, 4(3), 203–222.
Wang, P. P. (2010). Computational Intelligence in Economics and Finance,
Volume I. Berlin Heidelberg: Springer-Verlag.
Wang, K. & Huang, S. (2010). Using fast adaptive neural network classifier
for mutual fund performance evaluation, Expert Systems with
Applications, 37(8), 6007–6011.
Wang, P. P. & Kuo, T. W. (2010). Computational Intelligence in Economics
and Finance: Volume II, Berlin Heidelberg: Springer-Verlag.
Wong, B. K. & Selvi, Y. (1998). Neural network applications in finance: A
review and analysis of literature (1990-1996). Information and
Management, 34(3), 129–139.
Wang, J. Z., Wang, J. J., Zhang, Z. G. & Guo, S. P. (2011). Forecasting stock
indices with back propagation neural network. Experts Systems with
INDEX
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
Council of the Islamic Fiqh Academy, 2 environmental standards, 61

cross-border investment, 69 environments, 116
cross-validation, 107 equilibrium, 6, 23, 58, 68
current prices, 98, 117, 120 equity(ies), vii, viii, 2, 3, 10, 14, 15, 17,
19, 23, 28, 31, 42, 46, 52, 56, 57, 59,
60, 61, 63, 64, 67, 68, 69, 70, 73, 74,
D 75, 76, 77, 78, 79, 81, 83, 84, 87, 88,
89, 90, 91, 93, 94, 96, 97, 99, 101,
DEA, viii, 73, 75, 76, 78, 79, 80, 81, 84, 104, 116, 121, 122, 123, 124, 125,
85, 86, 87, 88 126
deficit, 121 equity investment, 2, 68
dependent variable, 44, 54, 97, 117 equity market, 3, 23, 28, 31
depth, 25, 103, 104 ethical issues, 41, 51
derivatives, 124 Europe, 3, 58, 69, 125
developed countries, 20, 27 European market, 27
developing countries, 3 evidence, 5, 10, 15, 17, 19, 20, 24, 26,
deviation, 6, 7, 28, 29, 75, 79, 83, 84, 85, 27, 32, 33, 35, 37, 40, 46, 47, 50, 56,
98, 106, 108, 113, 118, 119, 120 57, 58, 59, 62, 67, 68, 69, 70, 76, 87,
dichotomy, 58 91, 94, 121, 126
dimensionality, 112 exclusionary screening, vii, 1, 4, 32
discretization, 124 experimental design, 123, 124
discriminant analysis, 126 exposure, 6, 10, 14, 21
disposition, 39, 49, 70
dissonance, 39, 49, 63
distribution, 18, 75, 78, 119 F
diversification, 5, 6, 20, 24, 28, 30, 36,
67, 76, 81, 82 factor analysis, 112
faith, 40, 51, 62, 64, 68
family characteristics, 43, 53
E financial, vii, 1, 2, 3, 4, 5, 7, 21, 22, 23,
24, 25, 30, 31, 32, 33, 34, 35, 36, 37,
economic cycle, 24 40, 41, 43, 46, 47, 51, 52, 54, 56, 58,
economic development, 77 59, 63, 64, 65, 66, 67, 69, 74, 75, 77,
economic theory, 22 88, 94, 95, 103, 116, 117, 125
economics, 65, 88 financial crisis, 30, 33, 34
economies of scale, 18 financial development, 77
Efficient market hypothesis, 17 financial institutions, 2, 94
election, 8, 9, 10, 25, 28, 63, 89 financial markets, 7, 66, 74, 116, 117
electrophoresis, 123, 124 financial performance, 22, 23, 24, 35,
emerging markets, 20, 39, 50, 75, 76, 90, 36, 41, 46, 51, 56, 58, 63, 64, 67, 69
116 firm size, 12
empirical studies, viii, 2, 4, 36, 42, 47, firm value, 23, 24, 36, 58
52, 94 flow relationship, 37, 47
environment, viii, 22, 93, 95, 96, 103, fluctuations, 119
104, 107, 112, 116 forecasting, viii, 9, 64, 93, 94, 95, 96,
environmental management, 66 97, 112, 113, 115, 116, 122, 125
Index 131
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
U variables, 9, 12, 14, 18, 19, 24, 94, 95,

96, 97, 98, 99, 100, 101, 106, 108,
universe, 32, 36 109, 110, 111, 112, 113, 114, 115,
unsystematic risk, 6, 21, 24, 35, 79 116, 117, 120
vector, 99, 100, 101, 106
volatility, 5, 9, 41, 42, 45, 51, 52, 55, 58

Mutual Funds - Donald Edwards PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mutual Funds - Donald Edwards PDF

Uploaded by

Copyright:

Available Formats

FINANCIAL INSTITUTIONS AND SERVICES

Additional books in this series can be found on Nova’s website

Additional e-books in this series can be found on Nova’s website

NOTICE TO THE READER

Library of Congress Cataloging-in-Publication Data

Published by Nova Science Publishers, Inc. † New York

The authors of this book provide and discuss new research on

performance. In addition, this chapter also reviews the literature on fund

A REVIEW OF PERFORMANCE, SCREENING

Ainulashikin Marzuki1,* and Andrew C. Worthington2

mutual fund performance evaluation techniques and the underlying

2. MUTUAL FUND PERFORMANCE

Carhart (1997), Christopherson et al. (1998), Goetzmann and Ibbotson (1994)

2.1. Mutual Fund Performance Evaluation

Early literature relied on raw returns to measure security or portfolio

is equally important. This theoretical model assumes that investors hold a

2.1.1. Single Factor Models

Both Treynor (1965) and Sharpe (1966) developed relative performance

𝑹𝒊,𝒕 − 𝑹𝒇,𝒕 = 𝜶𝒊 + 𝜷𝒊 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 ) + 𝜺𝒊 (4)

of the market portfolio as a proxy to represent a portfolio benchmark and

2.1.2. Market Timing Models

allocation. Thus, Treynor and Mazuy (1966) proposed a quadratic term to

𝑹𝒊,𝒕 − 𝑹𝒇,𝒕 = 𝜶𝒊 + 𝜷𝒊 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 ) + 𝜸𝒊,𝒕 (𝑹𝒎,𝒕 − 𝑹𝒇,𝒕 )𝟐 + 𝜺𝒊,𝒕 (5)

𝑅𝑖,𝑡 − 𝑅𝑓,𝑡 = 𝛼𝑖 + 𝛽𝑖 (𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 ) + 𝛿𝑖 (𝑅𝑚,𝑡 − 𝑅𝑓,𝑡 )𝐷𝑡 + 𝜀𝑖,𝑡 (6)

data used by previous studies. However, the duration of the outperformance

2.1.3. Multifactor Models

diversified or benchmark portfolios. They asserted that there is no appropriate

𝑅𝑖 = 𝐸𝑖 + 𝑏𝑖,1 𝛿1 + 𝑏𝑖,2 𝛿2 +. . . +𝑏𝑖,𝑘 𝛿𝑘 + 𝜀𝑖 (7)

where 𝑅𝑖 is the return on asset i, 𝐸𝑖 is expected return for asset i, 𝑏𝑖,𝑘 is

2.1.4. Other Related Performance Models

Return-based style analysis was introduced by Sharpe (1992) to measure

consistently indicate that, on average, fund managers’ ability to add value

2.2. Criticisms of Mutual Fund Performance

There are two main criticisms of mutual fund performance measurement

2.3. Fund Specific Characteristics and Performance

This section reviews literature on the determinants of mutual fund

trustee fees. Recently, studies have focused on the relationship between

winner funds tend to hold in the portfolio. In addition, he found persistence

3.1. Screening, Firm Performance and Behavior

This section reviews several theoretical and empirical papers relating to

behavior. Do firms really want to be included in the SRI or ethical portfolio?

This implies that economic agents concerns about maximizing their

year to year. However, it is interesting to note that the list of Shariah

3.2. Screening and Portfolio Performance

There are several reasons to believe that screened funds may

Second, a screened portfolio selects companies that are potentially

3.2.1. Empirical Evidence in SRI

conventional funds. Interestingly, the underperformance of SRI funds was not

3.2.2. Empirical Evidence in Islamic Investments

investigated the performance of 59 IMFs globally from 2001 to 2006 and he

3.3. Screening Strategies and Portfolio Performance

In view of various types of screening strategies available for SRI fund

Author Sample Period No. of funds Performance measures Main findings

In addition to comparison between positive and negative screening,

4. FUND FLOWS AND PERFORMANCE

4.1. Fund Flows and Past Performance

Investors consider past fund performance as the most important factor in

This asymmetric relationship may also be due to costly information, high

4.2. Fund Flow Volatility

Another strand of literature concerning fund flow investigates the

4.3. Other Determinants of Fund Flows

subsequently, their incentive. Agarwal et al. (2003) extended the study on

fund management companies have an incentive to introduce many new