An Analysis Of Companies Portfolio Performance Using Sharpe Ratio: -

A Study On The Differences Of Performance Between Malaysian Conventional And Islamic ­­­­­­­­­­­­­­Equity Mutual Fund In 2007

1.0 Introduction

1.0.1 Chapter Description

In this chapter, explaining the background of the study, problem statement, objectives of the study, hypotheses, significance of this study, as well as the scope and limitations during the process of completing this study.

1.0.2 Background of the Study

Portfolio evaluation is on the time before 1960. Investors evaluated portfolio performance almost entirely on the basis of the rate of return. They were aware of the concept of risk but did not know how to quantify or measure it, so they could consider it explicitly. Developments in portfolio theory in the early 1960s showed investors on how to quantify and measure risk in terms of the variability of returns. Still, because no single measure combined both return and risk, the two factors had to be considered separately as researchers such as Friend, Blume, and Crockett (1970).

Specifically, the investigators grouped portfolios into similar risk classes based on a measure of risk (such as the variance of return) and compared the rates of return for alternative portfolios directly within these risk classes. Before 1960, investors evaluated portfolio performance almost entirely on the rate of return, although they knew that risk was a very important variable in determining investment success. The reason for omitting risk was the lack of knowledge on how to measure and quantify it.

After the development of portfolio theory in early 60s, and CAPM in subsequent years, risk, measured as either by standard deviation or beta, was included in evaluation process. However, since there was not a single measure combining return and risk, two factors were to be considered separately that were researchers grouped portfolios into similar risk classes and compared rates of return of portfolios in the same risk class.

There are many kinds of measurement such as Jensen, Treynor and also Sharpe to evaluate the company's portfolio performance. Jensen's alpha has been a popular performance measure because it is a return concept. Related to Dr. William F. Sharpe's contribution to style analysis of investment performance, the Sharpe's alpha is related to the Jensen's alpha in the sense that both measures excess returns. They differ, however, in the selection and construction of benchmarks.

Sharpe (1966) developed a composite index which was very similar to the Treynor measure, the only difference was that it was being used as standard deviation, instead of beta. To measure the portfolio risk, the researcher needs the average rate of return for Portfolio during a specified time period, the average rate of return on risk-free rate during the same period, Sharpe performance index and the standard deviation of the rate of return for Portfolio during the time period. Sharpe preferred to compare portfolios to the capital market line (CML) rather than the security market line (SML). Sharpe index, therefore, evaluated funds performance based on both rate of return and diversification (Sharpe 1967). For a completely diversified portfolio Treynor and Sharpe indices would give identical rankings.

Although the mutual fund industry in Malaysia started as far back as 1959 with the establishment of the Malayan Unit Trust Ltd, the development of the industry did not take-off until the 1980s with the launching of the Amanah Saham Nasional (ASN).

In 2004, the Commission approved 17 new Syariah-based unit trust funds, bringing the total number of such funds to 71 or 24.4% of the total 291 approved funds in the industry as at the end of 2004 (2003: 55 funds or 24.3% of the total industry). Of the 71 Syariah-based unit trust funds, 14 were balanced funds, 14 were bond funds, 39 were equity funds, 2 two were fixed income funds and two were money market funds. The number of units in circulation for Syariah-based unit trust funds also increased from 8.59 billion units as at the end of 2003 to 13.16 billion units in 2004. The number of accounts registered an increase of 23.4% or 80,848 accounts, with a total of 427,000 accounts in 2004.

One conventional fund made changes to its investment objectives and operations which enabled it to comply with the requirements of Syariah-based unit trust funds. In terms of value, the NAV of Syariah-based unit trust funds grew to RM6.76 billion representing 7.7% of the industry, an increase of 0.9% from the previous year. Over a 10-year period (1995-2004), the NAV of Syariah-based unit trust funds grew at a compounded annual growth rate (CAGR) of 26.18% while the overall industry CAGR was 7.89%.

The recognition of the increasing dominance and importance of unit trusts as an investment instrument has spurred researchers to devise appropriate techniques to assess portfolio performance. The earlier works by Sharpe (1966), Treynor (1965) and Jensen (1968) represented significant contributions to the evaluation of portfolio performance.

Therefore, the primary aim of this paper is to present new evidences for the analysis of companies' portfolio performance using Sharpe ratio by studying the differences the performance between Malaysian conventional and Islamic Equity Mutual Fund in 2007.

1.0.3 Overview of Conventional and Islamic Mutual Fund

Mutual fund or better known as unit trust in Malaysia is an investment vehicle created by asset management companies specializing in pooling savings from both retail and institutional investors. Individual investors seeking liquidity, portfolio diversification and investment expertise are increasingly choosing unit trust funds as their investment vehicle. However, these investors do differ in their preferences based on their risk threshold, liquidity needs and their needs to comply with religious requirement.

In the Malaysian context, the performance of mutual funds or more popularly known as unit trust funds as reported by Shamsher and Annuar (1995), Tan (1995), Leong and Aw (1997), Annuar et al,. (1997) and Low and Noor A. Ghazali (2005) concluded that on average, funds were unable to beat the market.

The number of unit trust has grown dramatically in recent years. Unit trusts are now the preferred way for individual investors and many institutions to participate in the capital markets, and their popularity has increased demand for evaluations of fund performance.

Muslims are not allowed to invest in standard mutual funds since their religion prohibits them from investing in certain equities, like those of banks or companies that deal in pork, alcohol, pornography and certain entertainment related products.

An Islamic mutual fund is similar to a “conventional” mutual fund in many ways; however, unlike its “conventional” counterpart, an Islamic mutual fund must conform to the Sharia (Islamic Law) investment precepts. The Sharia encourages the use of profit sharing and partnership schemes, and forbids riba (interest), maysir (gambling and pure games of chance), and gharar (selling something that is not owned or that cannot be described in accurate detail; i.e., in terms of type, size and amount) (El-Gamal 2000).

The Sharia guidelines and principles govern several aspects of an Islamic mutual fund, including its asset allocation (portfolio screening), investment and trading practices, and income distribution (purification). When selecting investments for their portfolio (asset allocation), conventional mutual funds can freely choose between debt-bearing investments and profit-bearing investment, and invest across the spectrum of all available industries. An Islamic mutual fund, however, must set up screens in order to select those companies that meet its qualitative and quantitative criteria set by Sharia guidelines.

1.1 Problem Statement

At some levels, people are always interested in evaluating the performance of their investments. Having to spend the time and incurred the expense to design an asset allocation strategy and select the specific set of securities to form their portfolios, investors - whether they are individuals, corporations, or financial institutions. It must be periodically determined whether this effort is worthwhile. Investors in managing their own portfolios should evaluate their performance, as should those who pay one or several professional money managers to make these decisions for them. It is imperative to determine the realized investment performance which justifies the additional costs of engaging professional management.

Comparing a portfolio's historical returns to those produced by other managers or indexes can be instructive; such comparisons do not produce a complete picture of the portfolio's performance. Indeed, the central tenet of the modern approach to performance measurement is that it is impossible to make a thorough evaluation of an investment without explicitly control the risk of the portfolio. Given the complexity and importance of the issues involved, it is not a surprise to learn that there is not a single universally accepted procedure for risk-adjusting portfolio returns. Nevertheless, there are several techniques that are commonly employed.

Some previous studies found results that are inconsistent with Chua's findings. These studies include Ewe (1994), Shamsher and Annuar (1995), and Tan (1995). Shamsher and Annuar (1995), focused their study on the performance of 54 unit trusts covering the period of late 80s to early 90s. They found out that the returns on investment in unit trust were well below the risk free and market returns. Furthermore, the results indicated that not only the degree of portfolios diversification was below expectation but the actual returns and risk characteristics of funds were also inconsistent with their stated objectives. Tan (1995) analyzed performance of 12 unit trusts over a 10-year period, 1984-1993. He concluded that unit trusts in general perform worse than the market portfolio. Consistent with Chua's findings, Tan also concluded that government sponsored funds performed better than private funds.

As we can see, there are three portfolio performance evaluation techniques that comprise the basic ‘toolkit' for measuring risk-adjusted performance. Although some redundancy exists among these measures, each of them provides unique perspectives, so that best viewed as complementary measures. In particular, examining the controversy surrounding the selection of the proper benchmark to use in the risk-adjustment process and discussing why these benchmark problems become larger when beginning to invest globally.

From here, how to evaluate the performance of the investments in order to reduce the risk taken? What measurement can contribute to evaluating a good investment? Therefore, it is interesting to analyze the companies' portfolio performance by studying on the differences in the performance between conventional and Islamic equity mutual fund in Malaysia by using Sharpe ratio.

1.2 Objectives of Research

The general purpose of this study is to analyze the companies' portfolio performance using Sharpe ratio by studying on the differences in the performance between conventional and Islamic equity mutual fund in Malaysia.

A careful review on those questions has led to the development of the following specific research objectives which are:

i. To measure and rank both relative quantitative performances of mutual fund (conventional/Islamic) on the basis of their return, total risk, coefficient of variation and Sharpe ratio. The term performance contains both the return and the risk undertaken by these mutual funds.

ii. To investigate whether both mutual funds (conventional/Islamic) are earning higher returns than the benchmark returns (or market Portfolio/Index returns) in terms of risk.

iii. To determine the relationship between dependent variable and independent variable.

1.3 Scope of Study

The study is on the analysis of the companies' portfolio performance in determining the measure of average daily return, total risk (standard deviation), coefficient of variation and Sharpe ratio. Moreover, to observe the differences in terms of performance between conventional and Islamic mutual fund in the context of Malaysian capital market by comparing them to the stock market index or Kuala Lumpur Composite Index (KLCI) benchmark.

The scopes of the study are stated as follow:

§ The relationship between two variables: the return on equity mutual fund as the dependent variable whereas the return on stock market index (KLCI) as the independent variable for conventional and Islamic fund.

§ The period of study will cover one (1) year starting from January 2007 to December 2007 using the daily basis collected from The Star and New Straits Times newspapers and also from the internet.

§ This research will also focus on the conventional and Islamic Equity Mutual Fund companies available in Malaysia.

1.4 Theoretical Framework

The theoretical framework shows the relationship between the independent variables and dependent variable. The independent variable is the return on KLCI while the dependent variable is the return of Equity Mutual fund companies in Malaysia.

Schematic diagram for the theoretical framework in this study is as follows:

Market Index Equity Fund

Market Index Islamic Equity Fund

Independent Variable Dependent Variable

1.5 Hypotheses

According to Uma Sekaran (2003), a hypothesis can be defined as a logically conjectured relationship between two or more variables expressed in the form of testable statement. Hypothesis can be divided into two categories which are Ho which is a Null Hypothesis and Ha which is an Alternate Hypothesis. The term “null” can be thought of as meaning “no change” or “no difference”. The second hypothesis is called alternative hypothesis. It is summary of the case if the null hypothesis is not true. It is stated that Ha, the alternative hypothesis is a statement of a view that has been prepared to be accepted if Ho is rejected. The hypotheses of this study are:

Hypothesis 1:

Ho: There is no relationship between the return on KLCI and the return on

Conventional equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Conventional equity mutual fund.

Hypothesis 2:

Ho: There is no relationship between the return on KLCI and the return on

Islamic equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Islamic equity mutual fund.

1.6 Limitations of the Study

Data Collection and Cost Limitation

The major source of data gained is from the secondary sources. The data is only available at certain places and it requires cost to obtain the data. Besides, it also requires costs in the process of printing, photocopying, data services and transportations to obtain the information. The information about the topic studied is also difficult to search in the library because of the limited information. As a result, it causes problems to the researcher to gather and collect the information. The information and data related to the study is rather difficult to obtain. Thus, the accuracy of the study depends on the accuracy of the data available and may not perfectly precise. In addition, data is also limited since it relies on the secondary sources alone.

Lack of Experience and Expertise

Since this research is the first research experience for the researcher, undoubtedly there are still lots of things to improve. The lacks of experience especially in data collection and time management have been the limitations to the researcher. Moreover, the researcher has limited knowledge on the topic and needs more understanding on the topic studied.

Time Constraint

Time is very limited for the researcher to complete the research. The researcher has to be very smart in scheduling the time to make sure the research is completed in time. Thus, time constraint has been identified as one of the limitations for the researcher.

1.7 Significance of the Study

This research analyzes on the companies' portfolio performance using Sharpe ratio by studying on the differences in the performance between conventional and Islamic Equity Mutual Fund in Malaysia. Therefore this study will provide some information that can be useful because the data and findings from this research will help other researchers to produce better result in their research. This research is also significant to:

To Researcher

As a finance student, issues in measuring portfolio performance are so much important and crucial. By studying about measurement of portfolio performance in depth, a better understanding and knowledge is gained. This research has given the researcher the opportunity to get the experience in practice as well as in theories.

To other researcher

This study also can be a useful reference to other researchers who are keen to carry on the study regarding the performance of mutual fund in Malaysia. There are several fruitful areas in this study that can be further examined by other researchers. Further study will give an opportunity to other researchers to expand their view and knowledge. To do so, they need to refer to numerous literatures and hopefully, this research can come in handy for them.

To Finance Students

This research will be very useful for finance students in having more knowledge about the company's portfolio performance and the differences in the performance between conventional and Islamic Equity Mutual fund in Malaysia. They can use this research as a guide and as references in their studies in portfolio management and mutual fund in Malaysia.

To Businesses

This research is very important to businesses in realizing the effects of portfolio management on their performance. This is important so that they will have clear direction in deciding their investment.

To Investors

This study plays an important role in decision making since it gives the investors a prior knowledge of which Equity Mutual Fund companies is the best to invest and whether those companies provide high returns on investment. Moreover, revealing the specific volatility patterns in returns might also benefit investors in risk management and portfolio optimization. This research is also important to investors so that they can have a clearer picture of their investment choices. For investors the study can help them to know the risk and return of their investment transaction.

1.8 Definition of Terms

Portfolio

A collection of investments are all owned by the same individual or organization. These investments often include stocks, which are investments in individual businesses; bonds, which are investments in debt that are designed to earn interest; and mutual funds, which are essentially pools of money from many investors that are invested by professionals or according to indices.

Sharpe Ratio
A risk-adjusted measure developed by William F. Sharpe, calculated using standard deviation and excess return to determine reward per unit of risk. The higher the Sharpe ratio, the better the fund's historical risk-adjusted performance.

Mutual Fund (Unit Trust)

A form of collective investment constituted under a trust deed or a pooled investment plan where the capital contributions of investors are combined into a legally formed trust fund.

Equity fund

Equity fund or stock fund is a fund that invests in Equities more commonly known as stocks. Such funds are typically held either in stock or cash, as opposed to bonds, notes or other securities.

Return

Based on Investopedia definition, return can be defined as the gain or lossof an investment over a specified period, expressed asa percentage increase over the initial investment cost. Gains on investments are considered to beany income received from the security, plus realized capital gains.

Risk

The quantifiable likelihood of loss or less-than-expected returns

Risk-adjusted return

A measure of how much an investment returned in relation to the amount of risk it took on. Often used to compare a high-risk, potentially high-return investment with a low-risk, lower-return investment.

Benchmark

A standard, used for comparison. For example, the Nasdaq may be used as a benchmark against which the performance of a technology stock is compared.

Regression Analysis

A statistical technique used to find relationships between variables for the purpose of predicting future values.

Coefficient of Determination

A measure of the correlation between the dependent and independent variables in a regression analysis.

R-squared

A measurement of how closely a portfolio's performance correlates with the performance of a benchmark index, such as the S&P 500, and thus a measurement of what portion of its performance can be explained by the performance of the overall market or index. Values for r-squared range from 0 to 1, where 0 indicates no correlation and 1 indicates perfect correlation.

Kuala Lumpur Composite Index (KLCI)

The Kuala Lumpur Composite Index (KLCI) is a stock market index generally accepted as the local stock market barometer. KLCI was introduced in 1986 to public the need for a stock market index which would serve as an accurate performance indicator of the Malaysian stock market as well as the economy. It is used to be the main index, and is now one of the three primary indices for the Malaysian stock market which the other two are FMB30 and FMBEMAS, Bursa Malaysia. It contains 100 companies from the Main Board with approximately 500 to 650 listed companies in the Main Board which comprise of multi-sectors companies across the year 2000 to 2006 and is a capitalization-weighted index.

2.0 Literature Review

2.1 Chapter Description

Literature review is the documentation of a comprehensive review of the published and unpublished work from the secondary sources of data in the areas of specific interest to the researcher. The reason of the literature review is to ensure that no important variable that has in the past been founded repeatedly to have an impact on the problem is ignored. (Uma Sekaran, 2005 page 63).

2.2 Literature Review of Evaluated Portfolio Performance

Craig W. French (2003) discussed on what is involved in evaluating portfolio performance, including the need to adjust for differential risk and differential time periods, the need for a benchmark, and constraints on portfolio managers. It also considered the difference between the portfolio's performance and the manager's performance. For measurement this paper used well-known risk-adjusted (composite) measures of Sharpe portfolio performance. Investors who had all (or substantially all) of their assets in a portfolio of securities should rely more on the Sharpe measure because total risk is important.

Joel Owen and Ramon Rabinovitch (1998), for the last four decades, numerous authors have been suggesting methods to evaluate portfolio performance. Sharpe (1966) proposed performance measures which had produced a score for every portfolio being evaluated. These scores were used to compare the performance of any two portfolios or rank the performance of all portfolios in a given set. The earlier works by Sharpe (1966), Treynor (1965) and Jensen (1968) represented significant contributions to the evaluation of portfolio performance.

2.3 Literature Review of Sharpe Ratio

Francisco Peñaranda (2007) in her paper commented on developments beyond mean-variance preferences to some alternatives to the Sharpe ratio. The main goal of those measures was to give a similar ranking to Sharpe ratios when returns were symmetrically distributed and showed a preference for skewness when they were not. Moreover, performance measures could be used to guide asset allocation since they can be used as the criterion to maximize with portfolio.

Raphie Hayat (2006), the attractiveness of the Sharpe Ratio came from its intuitiveness and simplicity. The Sharpe Ratio are simple because it could rank funds on the base of a single and intuitive since it only rewarded funds with a higher ratio if their returns were higher with the same level of risk or if the risk was lowered while keeping the same level of return.

Zhidong Bai, Keyan Wang, and Wing-Keung Wong (2006) stated that the asset performance evaluation was one of the most important areas in investment analysis. In order to compare the performance among assets, several statistics had been developed and among them, the Sharpe-ratio statistic was the most prevalent. However, the major limitation of the Sharpe-ratio statistic was that its distribution is only valid asymptotically, but not valid for small samples. Nevertheless, it was important in finance to test the performance among assets for small samples.

Tzu-Wei Kuo and Cesario Mateus (2006), the Sharpe ratio was well known risk-adjusted performance measures and easily understood by an individual investor. Thus, investors could evaluate the exchange-traded funds (ETF's) performance, based on the Sharpe ratio. However, the Sharpe ratio relied on the assumption that returns were normally distributed having these measures difficulty in evaluating the performance with skewed return distributions.

Martin Eling and Frank Schuhmacher said that the classic Sharpe ratio was adequate in evaluating investment funds when the returns of those funds were normally distributed and the investor intended to place all his risky assets into just one investment fund. Because hedge fund returns differed significantly from a normal distribution, other performance measures had been proposed and encouraged in both academic and practice-oriented literature. The Sharpe ratio measured the performance of an investment fund by considering the relationship between the risk premium and the standard deviation of the returns generated by a fund. The Sharpe ratio were an adequate performance measure if the returns of the investment funds were normally distributed and the investor wished to place all his risky assets in just one investment fund.

Andreas G Merikas, Anna A Merikas and Ioannis Sorros (2005) examined the exact relationship between the Sharpe ratio and the information ratio. Sharpe in 1994 asserted that the information ratio was a generalized Sharpe ratio. Sharpe ratios had been estimated for each fund in each category, and an average ratio for each category. The Sharpe ratio would generally be positive since excess returns of funds over the risk free rate would be positive, unlike excess returns of funds over the market, which could be negative, as the return of the risk free bond was smaller but at the same time less volatile than the return of the market.

Cheryl J. Frohlich, Anita Pennathur and Oliver Schnusenberg in their research, Sharpe reward-to-variability ratio was used if total variability was thought to be the appropriate measure of risk, a stock's (portfolio's) risk-adjusted returns could be computed using the Sharpe Index. The Sharpe and Treynor Index eliminated the problem of only considering return as a measure of performance. However, neither ratio was independent of the time period over which it is measured. This means that the ratio can change from one period to another with different results. Moreover, both ratios also ignored the correlation of a fund with other assets, liabilities, or previous realizations of its own return.

Mario Onorato (2004), the Sharpe Ratio of any investment was defined as its excess return, it is return in excess of a benchmark return divided by the standard deviation of excess return. The benchmark represents a risk free investment alternative. Moreover, although the Sharpe ratio has become part of the modern financial analysis, its applications typically did not account for the fact that it was an estimated quantity, subject to estimation errors that would be substantial in some cases. The statistical properties of Sharpe ratios depended intimately on the statistical properties of the return series on which they are based. This suggests that a more sophisticated approach to interpreting Sharpe ratios is called for, one that incorporates information about the investment style that generates the returns and the market environment in which those returns are generated.

Wei Zhen (2004), in his paper said that the Sharpe (1966) and Treynor (1965) performance measures were widely accepted in both academia and industry to assess the Risk-adjusted value of a particular portfolio. It could be shown, after some mathematical treatment, that the Sharpe performance measure was useful when the portfolio of interest represented all of the investor's investment, while Treynor's measure was preferred when the portfolio under discussion was only a portion of the whole investment package.

Robert McCauley and Guorong Jiang (2004), through the Sharpe ratio it compared the returns of portfolios in relation to their risk by dividing their returns in excess of the riskless rate of return by the volatility of their returns. A portfolio with a higher Sharpe ratio was preferred in that it offered a higher return per unit of risk, as measured by return volatility.

William Goetzmann, Jonathan Ingersoll, Matthew Spiegel and Ivo Welch (2002), the Sharpe ratio is one of the most common measures of portfolio performance. It was used as a tool for evaluating and predicting the performance of mutual fund managers. Since then the Sharpe ratio, and its close analogues the Information ratio, the squared Sharpe ratio and M-squared, have become widely used in practice to rank investment managers and to evaluate the attractiveness of investment strategies in general. The appeal of the Sharpe measure was clear. It was an affine transformation of a simple t-test for equality in means of two variables, the first variable being the manager's time series of returns and the second being a benchmark. The Sharpe ratio was also ubiquitous in academic research as a metric for bounding asset prices.

Andrew Worthington and Helen Higgs (2002), the Sharpe ratio (also known as the reward-to-volatility ratio) indicated the excess return per unit of risk and was calculated by dividing the return in excess of the risk-free rate by the standard deviation of returns. In the current context, the Sharpe ratio was the most appropriate performance measure for an investor whose portfolio was composed wholly of a given artist's work.

Verena Kugi (1999), the Sharpe ratio measured the change in the portfolio's return with respect to a one unit change in the portfolio's risk. The higher this "Reward-to-Variability-Ratio" the more attractive was the evaluated portfolio because the investor received more compensation for the same increase in risk. Graphically, the Sharpe ratio was equal to the slope of a straight line connecting the position of the evaluated portfolio, for example a fund, with the risk-free rate. To determine the quality of performance, the Sharpe index of the evaluated portfolio was compared to the Sharpe index of the market or benchmark portfolio. The portfolio's Sharpe index being higher than the market's Sharpe index indicated that the portfolio manager had outperformed the market. Respectively, a lower Sharpe ratio was a sign of underperformance. Any portfolio that was positioned on the capital market line had a Sharpe ratio equal to that of the market and was therefore characterized by neutral performance.

Youguo Liang and Willard McIntosh (1998), the Sharpe's alpha captured the excess return of an investment relative to the composite benchmark. The composite benchmark had the following properties. First, the composite benchmark was a portfolio of benchmarks that best replicated the performance of the investment over the evaluation period. The universe of benchmarks, which was typically large in number and diverse in nature, was selected by managers. The computer, however, would construct the composite benchmark from the universe according to a pre-specified procedure. Second, the composite benchmark was a portfolio without short-selling the individual benchmarks. This property was important and relevant because it was virtually impossible to short many stock and bond indices. Third, the composite benchmark was a “real” portfolio in the sense that the portfolio weights sum to one. Sharpe's alpha could be considered as an extension of the popular Jensen's alpha. It had the benefit of being an alpha - a risk-adjusted return - and retained the flexibility of benchmark selection and construction. It should be a valuable tool in portfolio style and attribution analysis.

2.4 Literature Review of Conventional Mutual Fund

Ong (2000) completely had a contradicting remark to those of Taib et al. (2000). He studied the performance of 53 unit trusts (37 private and 16 government-sponsored) before and during the 1997-98 financial crisis. He found that Malaysian unit trusts were able to out-perform the market before and during financial crisis. In fact, the performance of unit trust during the crisis was ‘‘better'' than before crisis. Also, consistent with previous studies, he found that government-sponsored funds performed better than private funds before the crisis. But the situation reverses during the crisis. He also found that fund size had no influence on the performance.

Do larger mutual funds attract more investments? Shukla and van Inwegen (1995) answered the question in the affirmative manner because larger funds were able to employ more research staff who was then able to provide more information that would lead to better portfolio selection. This relationship was supported by other studies (Chen et al., 1992; Ang et al., 1998; Golec, 1996). Grinblatt and Titman (1998), however, found an inverse relationship between fund size and performance. A possible explanation for this finding was that the degree of performance pressure on the fund manager was so intense that investment styles become aggressive resulting in frequent change in style and hence weaker performance. To what extent fund size influences the choice of mutual fund.

In the Malaysian context, although there have been many empirical studies related to the investment performance of unit trust funds, most studies had analyzed on the overall fund performance. Examples of such studies were provided by Shamsher and Annuar (1995), Tan (1995) and Leong and Aw (1997). Collectively, empirical findings on the overall fund performance indicated that on the average, unit trust funds in Malaysia had performed worse than the market. The few studies that separated the overall fund performance into selectivity and market timing components were provided by Annuar et al. (1997), Low and Noor A. Ghazali (2005) and Low (2005). These studies showed that on average, the timing performance of fund managers was negative. While there were many studies that investigated the investment performance of unit trust funds in Malaysia, up to date no study had examined the price linkages between unit trust funds and the local stock market index.

Milonas (1995) used the Treynor -Mazuy model to evaluate the performance of mutual funds operating in the Greek financial market. The estimation results referred to 10 mutual funds of mixed and equity type for the period between 1993 -1994 and 12 mutual funds of mixed and equity type for the period 1995-1996 using as approach for the market portfolio the General Index of the Athens Stock Exchange (ASE). According to these findings it could not be argued that mutual fund managers exhibited significant timing ability.

Tan (1995) analyzed the performance of 12 unit trusts over a 10-year period, 1984-1993. He concluded that unit trusts in general performed worse than the market portfolio. Consistent with Chua's findings, Tan also concludes that government sponsored funds performed better than private funds.

In later studies, however, found results that were inconsistent with Chua's findings. These studies included Ewe (1994), Shamsher and Annuar (1995), and Tan (1995). Shamsher and Annuar (1995), focused their study on the performance of 54 unit trusts covering the period of late 80s to early 90s. They found that the returns on investment in unit trust were well below the risk free and market returns. Furthermore, the results indicated that not only the degree of portfolios diversification was below expectation but the actual returns and risk characteristics of funds were also inconsistent with their stated objectives.

Ewe (1994) utilized a sample of 37 funds and a period between 1988-1992, with test of performance by Jensen's Alpha Measure and Sharpe Index Measure, reported that while risk adjusted returns overall were less than those of stock market implying that the managers had low forecasting ability. Shamsher and Annuar (1995) found a similar result with Ewe (1994), where the returns on investment in 54 funds for the period 1988 - 1992 were below risk-free and market returns. Besides the performance was inconsistent over time, the degree of diversification of the portfolios was below expectation.

No factor had received as much attention in previous literature as past performance because it was seen to be the simplest and most direct method to gauge the performance of a mutual fund. Still, there seemed to be some doubts as to whether previous performance was a good indicator of future performance. Some literatures seemed to find there was only a slight positive relationship or no relationship at all between previous performance and current returns (Blake et al., 1993; Bogle, 1992; Brown and Goetzman, 1995; Brown et al., 1992).

This paper aims to examine the degree to which mutual funds are related to the market index regardless of whether these funds are actively or passively managed. There are several studies that examined the relationship between mutual funds and local stock market indices. Bailey and Lim (1992) found significant correlations between the returns of country funds and the returns of the market index. However, they found that the pricing of country funds reflected more of the domestic US stocks than of the foreign equities in which these funds are invested.

Empirical studies on the performance of unit trusts in Singapore have been scanty. Notable exceptions were the works by Koh and Cheng (1990) and Koh, Phoon and Tan (1990). These studies generally found poor ex-post performance of the unit trusts in terms of returns, risk-variance efficiency and the degree of diversification. The poor performance of the unit trusts may itself account for the slow growth in the fund management industry in the 70s and 80s.

In 1990, the literature was extended by Cumby and Glen to include international mutual funds. The performance of 15 U. S.-based internationally diversified funds was compared to the Morgan Stanley Index for the U. S., the Morgan Stanley World Index, and to a benchmark combining the world index and Eurocurrency deposits. The time period analyzed was 1982-1988.

Ippolito (1989) examined the relationship between mutual fund investment performance and other variables such as asset size, expenses, turnover, and load status. Domestic mutual fund risk-adjusted returns, net of fees and expenses, were comparable to returns of index funds. However, portfolio turnover was unrelated to fund performance.

Grinblatt and Titman (1988) and Cumby and Glen (1990) found that a large proportion of mutual funds had negative coefficients on quadratic term. Lehman and Modest (1987) and Lee and Rahman (1990) also examined the Treynor -Mazuy regression, however, neither of these articles report whether the significance of the coefficients was due to their being positive or negative.

Evidence from Malaysia with regard to unit trust performance is very limited. Most of the studies use small sample sizes and the results are inconclusive. The earliest study by Chua (1985) with only 12 unit trusts as a sample reports that the performance was well above market return and quite consistent over the period 1974-1984. Later studies by Ewe (1994); Shamsher and Annuar (1995) suggested otherwise, that was unit trusts produced lower returns than the market portfolio.

Chua (1985) found that unit trust funds in Malaysia performed better than the market during his study period, 1974-1984. He concluded that the performance of unit trusts was fairly consistent and fund managers had diversified and performed risk control reasonably well. In addition, he found that government sponsored funds performed better than private funds. This may be due to certain investment ‘‘privileges'' accessible to only government-sponsored funds. Chua (1985) with exclusive samples of 12 Malaysian mutual funds between 1974 to 1984, concluded that funds outperformed the market proxy and performance was fairly consistent over time. High performance funds tend to relate to those with low expense ratio, low asset size and low portfolio turnover.

Kon and Jen (1979) evaluated mutual funds taking into account that systematic risk was often non-stationary. To do this they divided the sample in three different risk regimes and ran a standard regression for each period. They found evidence of different levels of beta (systematic risk) suggesting that a large number of funds engage in timing activities.

An issue of concern to mutual fund investors is the information content of various descriptors, which they may use in an attempt to select funds meeting their investment goals. Researchers have for some time [(McDonald,1974; Kuhle, 1988; Madura and Cheney, 1989), for example] noted a clear relationship between a fund's objective and its raw return but a much weaker relationship between fund objective and risk-adjusted return (McDonald, 1974; Kuhle, 1988).

The literature on the performance of mutual funds has long standing issues. The issues addressed by previous studies included the risk-return performance, selection and market timing abilities of fund managers and the level of diversification of mutual funds. McDonald (1974) estimated the Sharpe, Treynor and Jensen measured for 123 mutual funds using monthly data for the period between 1960 and 1969. The findings showed that majority of the funds did not perform as well as the New York Stock Exchange (NYSE) index.

The literature on the investment performance of mutual funds is extensive and spans several decades. Many of these empirical studies make a comparison between the fund's return with that of the market. Such comparison allows investors to gauge the differences in the performance between actively managed funds and a passively managed portfolio, i.e. the market index. In general, most previous studies with few exceptions have found either negative performance or no performance at the aggregate level for the average mutual funds. Some of the more important studies showing a lack of superior performance by fund managers are Sharpe (1966), Jensen (1968), McDonald (1974), Chang and Lewellen (1984), Cumby and Glen (1990) and Droms and Walker (1994) among others.

In general, the models used to measure performance in mutual funds are variants of the CAPM, including the multi-factor models and models involving higher moments of the return generating process. Research documents that fund performance for the same period can vary significantly depending on the model used to measure performance. Earlier research focused on the Sharpe, Treynor, and Jensen's alpha measures. Treynor's (1965) reward-to-volatility ratio (RVOL) distinguished between total risk and systematic risk, implicitly assuming that portfolios were well diversified and thus ignores any diversifiable risk. Therefore, if beta was the appropriate type of risk, a stock's (portfolio's) risk-adjusted returns can be determined by the Treynor Index. Alternatively, Sharpe's (1966) reward-to-variability ratio was used if total variability was thought to be the appropriate measure of risk; a stock's (portfolio's) risk-adjusted returns can be computed using the Sharpe Index. The Sharpe and Treynor Index eliminate the problem of only considering return as a measure of performance. However, neither ratio is independent of the time period over which it is measured. This means that the ratio can change from one period to another with different results. Moreover, both ratios also ignored the correlation of a fund with other assets, liabilities, or previous realizations of its own return (Hodges, Taylor, and Yoder, 1997).

2.4 Literature Review of Islamic Mutual Fund

Islamic finance is a relatively new phenomenon in the arena of economics and banking (Visser 2004). Although it is common knowledge that Islamic finance is based on the prohibition of interest, its other important features are usually unknown. To clarify, the following provides a concise overview of the rationale behind Islamic finance and its main features.

Abdullah, Mohammed and Hassan (2002) provided a more thorough analysis of the Islamic Equity Fund industry albeit only for Malaysian funds. They analyzed 67 Malaysian unit trust funds including 14 Islamic and 53 Conventional Funds using multiple performance measures like the Sharp Ratio, the Modigliani Measure and the Information Ratio. Abdullah, Mohammed and Hassan also found that the return of the Islamic and Conventional Funds was quite the same. Their overall conclusion is that IEF's in Malaysia follow the benchmark as well as conventional funds and that they both did this reasonably well. However when taking into account risk, Abdullah, Mohammed and Hassan found that the IEFs performed better than conventional funds during bear markets and that conventional funds performed better than IEFs during bull markets. This implied that investors had the option to switch between these funds depending on the market conditions and their personal preferences.

Ahmad (2001) provided a very rough guide to IEF performance by evaluating around 13 IEF's individually. Although Ahmed stated that some Islamic Funds outperformed benchmarks like the MSCI and stated that the IEF industry had outperformed the banking industry, he did not back this with statistical analysis. Thus no clear conclusions could be drawn from his research regarding the performance of IEF's.

One of the earliest studies on Islamic Funds was not until 1997 when Annuar, Shamsher and Ngu (1997) used the model developed by Treynor and Mazuy (1966) to examine the performance of 31 Malaysian mutual funds for the period 1990-1995. Many of these funds were Islamic and thus provide a proxy for Islamic Fund performance. The results were of course biased because conventional funds were also included in the study. Annuar, Shamsher and Ngu found evidence that these Malaysian funds did outperform their benchmark, but were poor at timing the market. Unlike Chen et al., Annuar, Shamsher and Ngu found a positive correlation between market timing ability and security selection ability.

2.6 Literature Review of KLCI benchmark

Abdullah, Mohammed and Hassan (2002) provided a more thorough analysis of the Islamic Equity Fund industry albeit only for Malaysian funds. They analyzed 67 Malaysian unit trust funds including 14 Islamic and 53 Conventional Funds using multiple performance measures like the Sharp Ratio, the Modigliani Measure and the Information Ratio. Abdullah, Mohammed and Hassan concluded that both type of funds slightly underperformed the Kuala Lumpur Composite Index (KLCI) benchmark. However this under performance was statistically insignificant and thus holds no economic meaning.

Shamsher et al. (2000) conducted a study on the performance of 41 actively and passively managed funds in Malaysia covering the period from 1995 through 1999. The performance measures used were the Sharpe's index, the Treynor's index and the Jensen's index. The findings revealed no significant differences in the performance of actively and passively managed funds. Moreover, the returns of these funds were lower than the returns of the market portfolio. The diversification levels of these two funds were less than 50 per cent of the diversification level of the market index as proxy by the Kuala Lumpur Composite Index (KLCI). The selection skills of active fund managers were no better than that of the passive fund managers. The market timing abilities of managers were found to be poor for both the actively and passively managed funds.

In addition, the studies conducted with respect to the performance measurement of Malaysian unit trust funds had utilized market benchmarks such as Kuala Lumpur Composite Index (KLCI) and EMAS Index [Leong and Aw (1997), Ch'ng and Kok (1998)]. These researchers have advocated for more than one kind of market benchmarks for performance measurement. All the prior studies before 1997 had concentrated on using the broad market index i.e. KLCI as the single yardstick.

The findings based on factor models suggested that mutual funds on average underperformed the benchmark indices. For example, Gruber (1996) found out that using a four-factor model that funds underperformed by 65 basis points per year. Since the average expense ratio in his sample was about 113 basis points per year, this implied that mutual funds earned positive risk-adjusted returns, but charges the investors more than the value added. Many studies found evidence of persistence in mutual fund performance. In particular, Brown and Goetzmann (1995) found persistence in risk-adjusted returns based on one- and three-factor models, which was mostly due to funds with bad performance. However, Carhart (1997) demonstrated that most of performance persistence found in the previous studies could be attributed to the one-year momentum effect. The only significant persistence not explained by this factor has consistent underperformance by the worst-performing funds.

In the Malaysian context, the performance of mutual funds or more popularly known as unit trust funds as reported by Shamsher and Annuar (1995), Tan (1995), Leong and Aw (1997), Annuar et al. (1997) and Low and Noor A. Ghazali (2005) concluded that on average, funds were unable to beat the market. Similar findings were reported by Chua et al. (1985) and Koh et al. (1987) for the performance of unit trust funds in Singapore. In evaluating fund performance, many studies in Malaysia have used the Kuala Lumpur composite index (KLCI) as the market benchmark. The one and only study that examined the sensitivity of fund performance to different benchmark portfolios was provided by Leong and Aw (1997). The two benchmarks used were the KLCI and the (EMAS) Exchange Main Board All-Share Index. Their findings showed that when the EMAS Index was used, more funds exhibit better performance than the market based on risk adjusted performance measures. In addition, EMAS Index was also shown to produce higher R-squared than the KLCI. Their findings suggested that the choice of market benchmark was important in measuring the investment performance of Malaysian mutual funds. However, their work focused on the overall fund performance and they did not evaluate the separate contribution of market timing and selectivity to the fund's overall return.

Albeit the existence of non-symmetric relationship between the NAV and the Kuala Lumpur Stock Exchange market capitalization, majority of the studies conducted with respect to the performance measurement of Malaysian unit trust funds had utilized market benchmarks either Kuala Lumpur Composite Index (KLCI) or EMAS Indices. [Chua (1985), Ewe (1994), Shamsher and Annuar (1995), Leong and Aw (1997), Ch'ng and Kok (1998)]. If unit trust funds were to invest in various investment vehicles, it would be more appropriate to examine the performance of mutual funds against a multi-indexed benchmark.

3.0 Research Methodology

3.1 Chapter Description

This chapter discusses about the methodology used in this study. It concerns the framework of specified type of information to be collected, source of data, data collection and procedures of data analysis. Research methodology examines the systematic procedures in gathering the needed information. It is an important part in doing a research because collecting valid and correct data depends on the methodology.

3.2 Research Design

This research analyzes the companies' portfolio performance by studying on the differences in the performance between conventional and Islamic Equity Mutual Fund in Malaysia by using Sharpe ratio. Thus, to measure the performance of mutual fund, all the data is being considered for this study including Net Asset Value (NAV) for equity mutual funds companies in financial market and the prices on market portfolio or Kuala Lumpur Composite Index (KLCI).

3.3 Data Collection

There are two types of data available namely, primary data and secondary data. Nevertheless, this research depends totally on secondary data. Secondary data refers to the information gathered from existing sources. For this study, the data used are the daily Net Asset Value (NAV) and daily prices for market portfolio that cover the period of January 2007 to December 2007. The data is obtained from The Star and New Straits Times newspapers and through the internet.

3.4 Sampling Design

The sampling frame for this study covers both conventional and Islamic mutual fund companies which are available in Malaysia. The sample size for this study is on the equity mutual fund between January to December 2007.

3.4 Measurement the Performance of Mutual Fund

3.5.1 Measuring the Return of Mutual Fund

3.5.1.1 Return

The net return that an investor achieves in investing in a mutual fund depends on dividends or interest payments and capital gains or losses that come from the changes in the net asset value. The return refers to the average monthly return achieved by the mutual funds under consideration. The monthly returns for the evaluation period were calculated using the following equation (Milonas 1995):

(1)

Where,

= Daily return of a mutual fund in the period

= Daily net asset value per unit of a mutual fund in the period

= Dividend of the mutual fund in the period

= Daily net asset value per unit of mutual fund in the period

3.5.1.2 Average Return

The most basic and simple method to evaluate fund returns is by calculating the average total return and comparing it to the average return of the benchmark. Mathematically, average return is defined as:

(2)

Here Rpt is the return on fund p at time t and n represents the number of fund returns in the sample.

3.5.2 Risk Measurement

There are various ways through which the risk can be measured. (a) Standard deviation: measures the dispersion around the mean. (b) Coefficient of variation: measures the risk per unit of return and is the result of the division between standard deviation and the mean return.

3.5.2.1 Standard Deviation

The total risk of the mutual funds under consideration is measured by the standard deviation of the monthly returns which was calculated using the following formula:

(3)

Where,

= Standard deviation (total risk) of the mutual fund

= Number of daily returns

= Daily returns of the mutual fund

= Mean return of the mutual fund

3.5.2.2 Coefficient of Variation

The coefficient of variation expresses the total risk undertaken by the mutual funds under consideration per unit of return achieved. More specifically, the coefficient of variation is given as such:

(4)

Where,

= Standard deviation (total risk) of the mutual fund

= Average (Mean) return of the mutual fund

3.5.3 Techniques of Evaluation

3.5.3.1 Sharpe's Ratio

Shortly after Treynor's technique was published, William Sharpe (1966) introduced an alternative technique for performance evaluation and illustrated the technique in evaluating the performance of a large number of mutual funds. Sharpe's technique is similar to that of Treynor with the difference being that Sharpe used the total risk of a portfolio and not just the systematic risk. Specifically, Sharpe's measure is the ratio of the risk premium of the portfolio, divided by the standard deviation of the portfolio's return:

(5)

Where,

= Average return on the portfolio (mutual fund) over the evaluation period

= Average risk-free return over the evaluation period (Treasury bills return)

= Standard deviation (total risk) of the mutual fund

3.6 Method of Data Analysis

3.6.1 Regression Analysis

The variables are divided into independent and dependent variable. After all data of variables is collected, analysis is carried out to get the findings of the research. All the data is treated and interpreted by using relevant form of analysis method. In this study, methods used include Statistical Package Social Science (SPSS). This SPSS is used in data processing procedure.

3.6.1.1 Simple Linear Regression

The simple linear regression includes one dependent and one independent. The dependent is the one being explained and independent variable is the one used to explain the variation in the dependent variable. The world linear regression means that the regression model gives a straight-line relationship between the variables.

3.6.1.2 Simple Linear Regression Model

A simple linear regression model is an equation, which relates dependent variable to an independent variable and random error, simple regression model is expressed as:

Y = a + b X + Є

Where;

Y : Dependent variable

X : Independent variable

a : Constant variables

b : Coefficient describe how changes in X affect the value of Y

Є : Random error

3.6.2 Coefficient Analysis

Correlation analysis is used to measure the closeness of the association or correlation that exists between two variables. It is also conducted to provide additional empirical evidence to derive the regression result. Besides that, the correlation analysis will also show how well a simple regression model change in the value of the dependent variable.

3.6.2.1 Coefficient of Determination (R²)

Coefficient of determination or R² value is to determine how well the regression model fits the data. A higher value of R² provides a better regression result as compared to a lower value of R². Due to this, strong relationship can be seen between the dependent and independent variables when R² value is higher.

The formula for coefficient of determinant R-Squared is:

R-Squared = Total explained variation

Total variation

3.6.2.2 Correlation of Coefficients (R)

Correlation of coefficients (R) is used to measure how closely the data collected are scattered around the least regression. The value of coefficient of correlation is represented by R and is ranged from -1 to 1. If the value of R is +1.0, there will be a perfect positive linear correlation relationship. However, if the value of R is -1.0, there will be a perfect negative correlation relationship indicated. No correlation is indicated if R is equal to 0. The value of coefficient of correlation can be categorized into several categories of linear correlation coefficient relationship. The tabulated interpretation mentioned is as follows:

Correlation Coefficients Interpretation

Correlation Coefficients

Interpretation

R = +1

Perfect positive linear correlation

0.5 < R < 1

Strong positive linear correlation

0 < R < 0.5

Weak positive linear correlation

R = 0

No linear correlation

-0.5 < R < 0

Weak negative linear correlation

-1 < R < -0.5

Strong negative linear correlation

R = -1

Perfect negative linear correlation

3.6.3 Durbin - Watson

Durbin Watson is the measurement of the correctness of the model used in the study. Besides that, it also indicates the strength of the model used. It is measured through the range from 1.5 and 2.5, which is normally considered as the best value, which is measured the correct model used in the study.

We can refer to the table to determine Durbin Watson correlation.

3.6.4 Test of Significance

T-Statistic is used in order to look at the strength of the relationship between dependent and independent variables. From this method, the significant level will be produce and all the significant level will be produced and all the significant level is identified in acceptance and rejection of hypothesis.

Calculated table t-value:

Where:

n = no. of observation

k = no. of independent variable

Therefore, T - statistic is used to examine whether there is a significant relationship or not between the Kuala Lumpur Composite Index (KLCI) and Conventional Equity Mutual Fund and Kuala Lumpur Composite Index (KLCI) and Islamic Equity Mutual Fund in Malaysia in 2007.

3.6.5 F-statistic

In the determination of rejection region of F - statistic, the one tail is used in order to determine the significance of the combination between the variables. Through the one tail test, it will explain the direction of the relationship between both the independent and the dependent variables. It is computed as the ratio of two samples variance. If the F - statistic is bigger than the critical value of (f), the regression equation is significant to explain the changes in the dependent variable. The formula of F - statistic is shown below:

Whereby:

F : F - statistic

SSR : sum of squares of regression

SSE : sum of squares of residual

n : number of observation

k : number of independent variable

Otherwise, the critical value of F is defined as follows:

Whereby:

 : significant level at 0.05

k : number of independent variable

n : number of observation

4.0 FINDING AND ANALYSIS

4.1 Chapter Description

This chapter discusses and explains on the statistical data, analysis and interpretation from the results of the hypothesis. The main focus of the study is to study on the differences in the performance between Malaysian conventional and Islamic equity fund in 2007 by using Sharpe ratio. For this study, ten companies of equity mutual fund have been chosen randomly. The researcher will select one equity fund that represents a conventional and an Islamic equity fund for each company.

In completing this research paper, the Excel Software is used in calculating and analyzing the data obtained. The result is also being calculated by using the SPSS version 15.0, the relationship between the independent variable and dependent variables will be determined and analyzed whether the returns of ten equity mutual fund are significantly correlated with market benchmark or not.

The researcher has used the Simple Linear Regression Model to test and analyze the data to see the relationship between the independent variable, which is return on KLCI with the dependent variables which are return on equity mutual fund for both conventional and Islamic fund.

The hypotheses of this study are:

Hypothesis 1:

Ho: There is no relationship between the return on KLCI and the return on

Conventional equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Conventional equity mutual fund.

Hypothesis 2:

Ho: There is no relationship between the return on KLCI and the return on

Islamic equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Islamic equity mutual fund.

4.2 Finding of Results

4.2.1 Average Daily Return

4.2.1.1 Average Daily Return for Conventional Equity Mutual Fund

Table 4.2.1.1

Conventional Equity Mutual Fund based on Average Daily Return

No

Company Management

Conventional

Average

Fund Name

Return

1

CIMB Wealth Advisors Berhad

CIMB Principal Equity

1.2853

2

ING Funds Berhad

ING Tactical

0.1512

3

AmInvestment Services Berhad

AmTotal Return

0.1294

4

Kuala Lumpur Composite Index

0.1210

5

MAAKL Mutual Berhad

Equity80

0.1120

6

RHB Investment Management Sdn Bhd

RHB Dynamic

0.1088

7

Prudential Fund Management Bhd

PRUdynamic

0.1024

8

Public Mutual

Public Equity

0.0754

9

Avenue Invest Berhad

EquityExtra

0.0212

10

HLG Unit Trust Berhad

Strategic Fund

-0.0023

11

Apex Investment Services Berhad

Apex Dynamic

-0.0100

Table 4.2.1.1 shows the summary of average daily return of conventional equity mutual fund obtained from estimating equation (2) in 2007 that is ranked in decreasing order. From the table, only three companies average return are higher than the return on the KLCI, they are the CIMB Principal Equity Fund (CIMB Wealth Advisors Berhad), the ING Tactical fund (ING Funds Berhad) and AmTotal Return fund (AmInvestment Services Berhad). The balance average return such as Equity80 fund (MAAKL Mutual Berhad), RHB Dynamic fund (RHB Investment Management Sdn Bhd), PRUdynamic fund (Prudential Fund Management Bhd), Public Equity fund (Public Mutual), EquityExtra fund (Avenue Invest Berhad) are positive except Strategic Fund (HLG Unit Trust Berhad) and Apex Dynamic fund (Apex Investment Services Berhad).

4.2.1.2 Average Daily Return for Islamic Equity Mutual Fund

Table 4.2.1.2

Islamic Equity Mutual Fund based on Average Daily Return

No

Company Management

Islamic

Average

Fund Name

Return

1

RHB Investment Management Sdn Bhd

RHB Mudharabah

0.5273

2

CIMB Wealth Advisors Berhad

CIMB Islamic Equity

0.1997

3

MAAKL Mutual Berhad

Al-Faid

0.1327

4

AmInvestment Services Berhad

AmIttikal

0.1249

5

Kuala Lumpur Composite Index

0.1210

6

Prudential Fund Management Bhd

PRUdana Dynamik

0.1091

7

Avenue Invest Berhad

SyariahExtra

0.0723

8

Apex Investment Services Berhad

Apex Dana Al-Sofi-I

0.0633

9

Public Mutual

Public Islamic Equity

0.0524

10

HLG Unit Trust Berhad

Dana Makmur

0.0262

11

ING Funds Berhad

ING Ekuiti Islam

0.0227

Table 4.2.1.2 shows the summary of average daily return of Islamic equity mutual fund obtained from estimating equation (2) in 2007 that is ranked in decreasing order. From the table, four companies average return are higher than the return on the KLCI, they are RHB Mudharabah fund (RHB Investment Management Sdn Bhd), CIMB Islamic Equity fund (CIMB Wealth Advisors Berhad), Al-Faid fund (MAAKL Mutual Berhad) and AmIttikal fund (AmInvestment Services Berhad). The balance Islamic fund such as PRUdana Dynamik fund (Prudential Fund Management Bhd), SyariahExtra fund (Avenue Invest Berhad), Apex Dana Al-Sofi-I fund (Apex Investment Services Berhad), Public Islamic Equity fund (Public Mutual), Dana Makmur fund (HLG Unit Trust Berhad) and ING Ekuiti Islam fund (ING Funds Berhad) have positive average return.

4.2.2 Total Risk

4.2.2.1 Total Risk for Conventional Equity Mutual Fund

Table 4.2.2.1

Conventional Equity Mutual Fund based on Total Risk

No

Company Management

Conventional

Total

Fund Name

Risk

1

CIMB Wealth Advisors Berhad

CIMB Principal Equity

21.3923

2

HLG Unit Trust Berhad

Strategic Fund

2.1693

3

Apex Investment Services Berhad

Apex Dynamic

1.7779

4

Public Mutual

Public Equity

1.4834

5

Avenue Invest Berhad

EquityExtra

1.4620

6

RHB Investment Management Sdn Bhd

RHB Dynamic

1.3602

7

MAAKL Mutual Berhad

Equity80

1.1950

8

Kuala Lumpur Composite Index

1.1585

9

ING Funds Berhad

ING Tactical

1.1252

10

AmInvestment Services Berhad

AmTotal Return

1.0925

11

Prudential Fund Management Bhd

PRUdynamic

0.9463

Table 4.2.2.1 shows the summary of total risk of conventional equity fund under consideration is measured by standard deviation of daily returns obtained from the estimating equation (3) in 2007 that is ranked in a decreasing order. From the table, seven companies total risk are higher than return on KLCI, they are CIMB Principal Equity Fund (CIMB Wealth Advisors Berhad), Strategic Fund (HLG Unit Trust Berhad), Apex Dynamic (Apex Investment Services Berhad), Public Equity fund (Public Mutual), EquityExtra fund (Avenue Invest Berhad), RHB Dynamic fund (RHB Investment Management Sdn Bhd) and Equity80 fund (MAAKL Mutual Berhad). The balance Conventional funds which are ING Tactical fund (ING Funds Berhad), AmTotal Return fund (AmInvestment Services Berhad) and PRUdynamic fund (Prudential Fund Management Bhd) has a lower total risk than market benchmark.

As can be seen, the CIMB Principal Equity fund (CIMB Wealth Advisors Berhad) provides investors with the highest average return (refer Table 4.2.1.1) as well as highest risk (total risk). It is a positive relationship between risk (total risk) and average return. Moreover, ING Tactical fund (ING Funds Berhad) and AmTotal Return fund (AmInvestment Services Berhad) provides investors with higher return (refer Table 4.2.1.1) as well as relatively low risk as compared to other funds. It is a negative relationship between risk (Total Risk) and average return. It is statistically significant which higher risk is associated with higher return for the whole period but in negative relationship it provides lower risk, higher return.

. 4.2.2.2 Total Risk for Islamic Equity Mutual Fund

Table 4.2.2.2

Islamic Equity Mutual Fund based on Total Risk

No

Company Management

Islamic

Total

Fund Name

Risk

1

RHB Investment Management Sdn Bhd

RHB Mudharabah

11.1016

2

ING Funds Berhad

ING Ekuiti Islam

2.8475

3

CIMB Wealth Advisors Berhad

CIMB Islamic Equity

2.7433

4

Apex Investment Services Berhad

Apex Dana Al-Sofi-I

1.9675

5

Public Mutual

Public Islamic Equity

1.4586

6

HLG Unit Trust Berhad

Dana Makmur

1.3825

7

MAAKL Mutual Berhad

Al-Faid

1.2454

8

Kuala Lumpur Composite Index

1.1585

9

AmInvestment Services Berhad

AmIttikal

1.0947

10

Prudential Fund Management Bhd

PRUdana Dynamik

0.8730

11

Avenue Invest Berhad

SyariahExtra

0.7304

Table 4.2.2.2 shows the summary of total risk of Islamic equity fund under consideration is measured by standard deviation of daily returns obtained from the estimating equation (3) in 2007 that is ranked in a decreasing order. From the table, seven companies total risk are higher than return on KLCI, they are RHB Mudharabah fund (RHB Investment Management Sdn Bhd), ING Ekuiti Islam fund (ING Funds Berhad), CIMB Islamic Equity fund (CIMB Wealth Advisors Berhad), Apex Dana Al-Sofi-I (Apex Investment Services Berhad), Public Islamic Equity fund (Public Mutual), Dana Makmur fund (HLG Unit Trust Berhad) and Al-Faid fund (MAAKL Mutual Berhad). The balance Islamic funds which are AmIttikal fund (AmInvestment Services Berhad), PRUdana Dynamik fund (Prudential Fund Management Bhd) and SyariahExtra fund (Avenue Invest Berhad) has a lower total risk than market benchmark.

As can be seen, RHB Mudharabah fund (RHB Investment Management Sdn Bhd) provides investors with the highest average return (refer Table 4.2.1.2) as well as highest risk (total risk). It is a positive relationship between risk (total risk) and average return. Moreover, AmIttikal fund (AmInvestment Services Berhad) provides investors with higher return (refer Table 4.2.1.2) as well as relatively low risk as compared to other funds. It is a negative relationship between risk (Total Risk) and average return. It is statistically significant which higher risk is associated with higher return for the whole period but in negative relationship it provides lower risk, higher return.

4.2.3 Coefficient of Variance

4.2.3.1 Coefficient of Variance for Conventional Equity Mutual Fund

Table 4.2.3.1

Conventional Equity Mutual Fund based on Coefficient of Variation

No

Company Management

Conventional

Coefficient of

Fund Name

Variation

1

Avenue Invest Berhad

EquityExtra

68.9371

2

Public Mutual

Public Equity

19.6756

3

CIMB Wealth Advisors Berhad

CIMB Principal Equity

16.6438

4

RHB Investment Management Sdn Bhd

RHB Dynamic

12.5075

5

MAAKL Mutual Berhad

Equity80

10.6659

6

Kuala Lumpur Composite Index

9.5715

7

Prudential Fund Management Bhd

PRUdynamic

9.2363

8

AmInvestment Services Berhad

AmTotal Return

8.4432

9

ING Funds Berhad

ING Tactical

7.4422

10

Apex Investment Services Berhad

Apex Dynamic

-177.0657

11

HLG Unit Trust Berhad

Strategic Fund

-943.1617

Table 4.2.3.1 presents the coefficient of variation for the KLCI and ten conventional equity mutual fund companies in 2007. The result obtained from the estimating equation (4) that is ranked in a decreasing order. Coefficient of variation is a measure of relative variability that indicates risk per unit return. From table, five companies coefficient of variation are higher than the return on KLCI, they are EquityExtra fund (Avenue Invest Berhad), Public Equity fund (Public Mutual), CIMB-Principal Equity fund (CIMB Wealth Advisors Berhad), RHB Dynamic fund (RHB Investment Management Sdn Bhd) and Equity80 fund (MAAKL Mutual Berhad). Moreover, the rest of the other five companies are lower than the return on KLCI which are PRUdynamic fund (Prudential Fund Management Bhd), AmTotal Return fund (AmInvestment Services Berhad), ING Tactical fund (ING Funds Berhad), Apex Dynamic (Apex Investment Services Berhad), and Strategic Fund (HLG Unit Trust Berhad).

4.2.3.2 Coefficient of Variance for Islamic Equity Mutual Fund

Table 4.2.3.2

Islamic Equity Mutual Fund based on Coefficient of Variation

No

Company Management

Islamic

Coefficient of

Fund Name

Variation

1

ING Funds Berhad

ING Ekuiti Islam

125.3983

2

HLG Unit Trust Berhad

Dana Makmur

52.7670

3

Apex Investment Services Berhad

Apex Dana Al-Sofi-I

31.0600

4

Public Mutual

Public Islamic Equity

27.8094

5

RHB Investment Management Sdn Bhd

RHB Mudharabah

21.0553

6

CIMB Wealth Advisors Berhad

CIMB Islamic Equity

13.7371

7

Avenue Invest Berhad

SyariahExtra

10.1028

8

Kuala Lumpur Composite Index

9.5715

9

MAAKL Mutual Berhad

Al-Faid

9.3867

10

AmInvestment Services Berhad

AmIttikal

8.7649

11

Prudential Fund Management Bhd

PRUdana Dynamik

8.0018

Table 4.2.3.2 presents the coefficient of variation for the KLCI and ten Islamic equity mutual fund companies in 2007. The result obtained from the estimating equation (4) that is ranked in a decreasing order. Coefficient of variation is a measure of relative variability that indicates risk per unit return. From the table, seven companies coefficient of variation are higher than the return on KLCI, they are ING Ekuiti Islam fund (ING Funds Berhad), Dana Makmur fund (HLG Unit Trust Berhad), Apex Dana Al-Sofi-I fund (Apex Investment Services Berhad), Public Islamic Equity fund (Public Mutual), RHB Mudharabah fund (RHB Investment Management Sdn Bhd), CIMB Islamic Equity fund (CIMB Wealth Advisors Berhad) and SyariahExtra fund (Avenue Invest Berhad). Moreover, the rest of the other three companies are lower than the return on KLCI which are Al-Faid fund from MAAKL Mutual Berhad, AmIttikal fund from AmInvestment Services Berhad and PRUdana Dynamik fund from Prudential Fund Management Bhd.

4.2.4 Sharpe Ratio

4.2.4.1 Sharpe Ratio for Conventional Equity Mutual Fund

Table 4.2.4.1

Conventional Equity Mutual Funds Based on Sharpe's Ratio

No

Company

Conventional

Sharpe's

Management

Fund Name

Ratio

1

ING Unit Trust Berhad

ING Tactical

0.1254

2

AmInvestment Services Berhad

AmTotal Return

0.1092

3

Prudential Fund Management Bhd

PRUdynamic

0.0976

4

MAAKL Mutual Berhad

Equity80

0.0852

5

RHB Investment Management Sdn Bhd

RHB Dynamic

0.0725

6

CIMB Wealth Advisors Berhad

CIMB Principal Equity

0.0596

7

Public Mutual

Public Equity

0.0440

8

Avenue Invest Berhad

EquityExtra

0.0076

9

HLG Unit Trust Berhad

Strategic Fund

-0.0057

10

Apex Investment Services Berhad

Apex Dynamic

-0.0113

Table 4.2.4.1 shows the summary of evaluation performance of conventional equity fund which is measured by Sharpe ratio for the year 2007. The result is obtained from the estimating equation (5) that is ranked in a decreasing order. The higher of Sharpe ratio measured means the better of the company portfolio, it means that when one company has got higher Sharpe ratio, it performs better than the aggregate market and is expected to get high return and high risk in the future. Since the value of KLCI is 0.0958%, only three companies are outperformed which are ING Unit Trust Berhad, AmInvestment Services Berhad and Prudential Fund Management Berhad. The rest of the companies are underperformed in the market and the lowest Conventional Equity fund company is Apex Investment Services Berhad; it means that this company has a weak portfolio performance.

4.2.4.2 Sharpe Ratio for Islamic Equity Mutual Fund

Table 4.2.4.2

Islamic Equity Mutual Funds Based on Sharpe's Ratio

No

Company

Islamic

Sharpe's

Management

Fund Name

Ratio

1

Prudential Fund Management Berhad

PRUdana Dynamic

0.1132

2

AmInvestment Services Berhad

AmIttikal

0.1047

3

MAAKL Mutual Berhad

Al-Faid

0.0983

5

Avenue Invest Berhad

SyariahExtra

0.0849

6

CIMB Wealth Advisors Berhad

CIMB Islamic Equity

0.0691

6

RHB Investment Management Sdn Bhd

RHB Mudharabah

0.0466

7

Public Mutual

Public Islamic Equity

0.0289

8

Apex Investment Services Berhad

Apex Dana Al-Sofi-I

0.0270

9

HLG Unit Trust Berhad

Dana Makmur

0.0115

10

ING Funds berhad

ING Ekuiti Islam

0.0044

Table 4.2.4.2 shows the summary of evaluation performance of Islamic Equity Fund which is measured by Sharpe ratio for the year 2007. The result is obtained from the estimating equation (5) that is ranked in a decreasing order. The higher of Sharpe ratio measured means the better of the company portfolio, it means that when one company has got higher Sharpe ratio, it performs better than the aggregate market and is expected to get high return and high risk in the future. Since the value of KLCI is 0.0958%, only three companies are outperformed which are Prudential Fund Management Berhad, AmInvestment Services Berhad and MAAKL Mutual Berhad. The rest of the companies are underperformed in the market and the lowest Islamic Equity fund company is ING Funds Berhad; it means that the company has a weak portfolio performance.

4.2.5 Simple Linear Regression

4.2.5.1 Simple Linear Regression Model for Conventional Equity Fund

Y = a + bi Xi+ Є

Where;

Y : Conventional Equity Fund

Xi : KLCI

a : Constant variables

bi : Coefficient describe how changes in X affect the

value of Y

Є : Random error

The SPSS output for the simple linear regression model is as follows:

Coefficients(a)

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

.151

.142

1.065

.287

Kuala Lumpur Composite Index

.895

.122

.150

7.365

.000

CON_001

-.253

.021

-.253

-12.091

.000

CON_002

-.057

.021

-.057

-2.708

.007

a Dependent Variable: Conventional Equity Fund

Model Summary(b)

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

Durbin-Watson

1

.282(a)

.080

.078

6.6402836

2.021

a Predictors: (Constant), CON_002, Kuala Lumpur Composite Index, CON_001

b Dependent Variable: Conventional Equity Fund

Y = 0.151 + 0.895 X

From the equation above, the KLCI is correlated to the conventional equity fund performance. From the output, the result can be illustrated by:-

bi= 0.895

Means, one unit change in KLCI leads to a change by 0.895 units in the conventional equity fund. Since the value is positive it indicates a direct relationship. Any increase in KLCI will lead to an increase in conventional equity fund by 0.895 and vice versa.

4.2.5.1.1 Correlation Coefficient, r = 0.282

The r value of 0.282 indicates 28.2% from the independent variable (KLCI) can interpret the dependent variable (conventional equity fund). The relationship between KLCI and the conventional equity fund is positive and can be considered as a weakly linear correlation.

4.2.5.1.2 Coefficient of Determination, r² = 0.080

This value means that about 8% of the variation in the dependent variable (conventional equity fund) performance is explained by the variability in the KLCI. The other 98% variation in the independent variable is explained by other factors.

4.2.5.1.3 Test Hypothesis

Hypothesis 1:

Ho: There is no relationship between the return on KLCI and the return on

Conventional equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Conventional equity mutual fund.

a) Selection of Distribution to use:

The sample size in this is 2228 and σ is not known. Therefore, the t-distribution is selected. The number of degrees of freedom is equal to the sample size minus the number of independent variable minus one that is:

b) Determination of the Rejection Region:

The significance level is determined at 0.05. In this study, two-tail test is used. Since the total area of both rejection regions is 0.05, the area if the rejection in each tail is 0.025, that is:

Area in each tail = ớ /2 = 0.05/2 = 0.025

Since the df is n-k-1 = 2228 - 1 - 1 = 2226 and area in each tail is 0.025, the critical points will be -1.960 and 1.960.

c) Durbin Watson

Durbin Watson is the measurement of the correctness of the model used in the study. Besides that, it also indicates the strength of the model used. It is measured through the range from 1.5 and 2.5, which is normally considered as the best value which is measures the correct model used in the study. Since conventional equity fund is 2.021, it is the best value. This result is based on two lags in order to get the best results for this data. Lag is used to make sure that the data is free from error by increasing the value of R-Square. Before lag is used in this model, the dependent variable that is conventional equity fund cannot be explained by the independent variable (KLCI) because it is very low than current result in Durbin Watson. However the current result in this study shows an improvement level in explaining the relationship between both variables and Durbin Watson statistic.

d) Calculating of Test Statistic:

At 95% confidence level, reject Ho if ta < -1.960 or ta > 1.960

Independent Variable

t- statistic value

Decision

KLCI

ta = 7.365

Fail to accept Ho

e) Decision

Since the calculated t (ta) for KLCI falls within the rejection region, we fail to accept Ho. Therefore, there is a significant positive relationship between conventional equity fund and KLCI.

4.2.5.1.4 F-statistic test

ANOVA(b)

Model

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

8490.412

3

2830.137

64.185

.000(a)

Residual

98063.647

2224

44.093

Total

106554.058

2227

a Predictors: (Constant), CON_002, Kuala Lumpur Composite Index, CON_001

b Dependent Variable: Conventional Equity Fund

F-statistic is used to test whether a significant proportion of total variation in dependent variables is explained by the estimated regression equation. At 95% confidence interval, critical value is one. Since the calculated F- test which is 64.185 is greater than the F-Value of 3.920. Therefore, regression equation is significant to explain the changes in the dependent variable.

4.2.5.2 Simple Linear Regression Model for Islamic Equity Fund

Y = a + bi Xi+ Є

Where;

Y : KLCI

Xi : Islamic Equity Fund

a : Constant variables

bi : Coefficient describe how changes in X affect the value of Y

Є : Random error

The SPSS output for the simple linear regression model is as follows:

Coefficients(a)

Model

Unstandardized Coefficients

Standardized Coefficients

t

Sig.

B

Std. Error

Beta

1

(Constant)

.082

.077

1.070

.285

Kuala Lumpur Composite Index

.854

.066

.255

12.922

.000

ISLAM_001

-.300

.020

-.300

-14.682

.000

ISLAM_002

-.097

.020

-.097

-4.752

.000

a Dependent Variable: Islamic Equity Fund

Model Summary(b)

Model

R

R Square

Adjusted R Square

Std. Error of the Estimate

Durbin-Watson

1

.374(a)

.140

.139

3.6090776

2.073

a Predictors: (Constant), ISLAM_002, Kuala Lumpur Composite Index, ISLAM_001

b Dependent Variable: Islamic Equity Fund

Y = 0.082 + 0.854 X

From the equation above, the Islamic equity fund is correlated to the KLCI performance. From the output, the result can be illustrated by:-

bi= 0.854

Means, one unit change in Islamic equity fund leads to a change by 0.854 units in the KLCI. Since the value is positive it indicates a direct relationship. Any increase in Islamic equity fund will lead to an increase in KLCI by 0.854 and vice versa.

4.2.5.2.1 Correlation Coefficient, r = 0.374

The r value of 0.374 indicates 37.4% from the independent variable (KLCI) can interpret the dependent variable (Islamic equity fund). The relationship between KLCI and the Islamic equity fund is positive and can be considered as a weakly linear correlation.

4.2.5.2.2 Coefficient of Determination, r² = 0.140

This value means that about 14% of the variation in the dependent variable (KLCI) performance is explained by the variability in the Islamic equity fund. The other 86% variation in the independent variable is explained by other factors.

4.2.5.2.3 Test Hypothesis

Hypothesis 2:

Ho: There is no relationship between the return on KLCI and the return on

Islamic equity mutual fund.

Ha: There is a relationship between the return on KLCI and the return on

Islamic equity mutual fund.

a) Selection of Distribution to use:

The sample size in this is 2228 and σ is not known. Therefore, the t-distribution is selected. The number of degrees of freedom is equal to the sample size minus the number of independent variable minus one that is:

b) Determination of the Rejection Region:

The significance level is determined at 0.05. In this study, two-tail test is used. Since the total area of both rejection regions is 0.05, the area if the rejection in each tail is 0.025, that is:

Area in each tail = ớ /2 = 0.05/2 = 0.025

Since the df is n-k-1 = 2228 - 1 - 1 = 2226 and area in each tail is 0.025, the critical points will be -1.960 and 1.960.

c) Durbin Watson

Durbin Watson is the measurement of the correctness of the model used in the study. Besides that, it also indicates the strength of the model used. It is measured through the range from 1.5 and 2.5, which is normally considered as the best value which measures the correct model used in the study. Since Islamic equity fund is 2.073, it is the best value. This result is based on two lags in order to get the best results for this data. Lag is used to make sure that the data is free from error by increasing the value of R-Square. Before lag is used in this model, the dependent variable that is Islamic equity fund cannot be explained by the independent variable (KLCI) because it is very low than current result in Durbin Watson. However the current result in this study shows an improvement level in explaining the relationship between both variables and Durbin Watson statistic.

d) Calculating of Test Statistic

At 95% confidence level, reject Ho if ta < -1.960 or ta > 1.960

Independent Variable

t- statistic value

Decision

KLCI

ta = 12.922

Fail to accept Ho

e) Decision

Since the calculated t (ta) for KLCI falls within the rejection region, we fail to accept Ho. Therefore, there is a significant positive relationship between Islamic equity fund and KLCI.

4.2.5.2.4 F-statistic test

ANOVA(b)

Model

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

4707.862

3

1569.287

120.479

.000(a)

Residual

28968.581

2224

13.025

Total

33676.443

2227

a Predictors: (Constant), ISLAM_002, Kuala Lumpur Composite Index, ISLAM_001

b Dependent Variable: Islamic Equity Fund

F-statistic is used to test whether a significant proportion of total variation in dependent variables is explained by the estimated regression equation. At 95% confidence interval, critical value is one. Since the calculated F- test which is 120.479 is greater than the F-Value of 3.920. Therefore, regression equation is significant to explain the changes in the dependent variable.

5.0 CONCLUSIONS AND RECOMMENDATIONS

5.1 Chapter Description

This chapter concludes the overall findings and provides recommendations to several parties.

5.2 Conclusions

This paper examines the presence of the performance of Conventional and Islamic Equity Fund companies in Malaysia whether they perform better or worse in the market portfolio in year 2007 by using daily Net Asset Value (NAV) collected from the internet and also The Star and New Straits Times newspaper. The conventional and Islamic equity mutual funds under consideration are ranked on the basis of their average returns, standard deviation, and coefficient of variation, systematic risk and Sharpe ratio.

5.2.1 Conclusion for Average Return

With the same market benchmark, three companies for conventional fund and four companies for Islamic fund have higher average return than KLCI. Meaning that, Islamic fund shows better performance compared to the conventional fund generally.

However, when comparing the value of average daily return between conventional and Islamic fund, conventional pay higher return than Islamic. The highest average return for conventional fund is 1.2853% from the CIMB Wealth Advisors Berhad (CIMB Principal Equity fund). While the highest average return for Islamic fund is only 0.5273% from the RHB Investment Management Sdn Bhd (RHB Mudharabah fund).

Overall, it can be said that, from the total of ten companies of conventional and Islamic equity fund, only three to four companies' performances are well above market return. Chua (1985), with only 12 unit trusts as a sample reports that the performance is well above market return and quite consistent over the period 1974-1984. Ewe (1994); Shamsher and Annuar (1995) suggested otherwise, that is unit trusts produce lower returns than the market portfolio. Evidence from Malaysia with regard to unit trust performance is very limited. Most of the studies use small sample sizes and the results are inconclusive.

5.2.2 Conclusion for Total Risk and Coefficient of Variation

Majority of equity fund companies for conventional and Islamic showed their total risk (standard deviation of return) and coefficient of variation (risk-return) higher than the KLCI benchmark.

For conventional fund, the positive relationship between risk (Standard Deviation) and average return shown by CIMB Principal Equity fund (CIMB Wealth Advisors Berhad) that provides investor with higher return as well as relatively high risk as compared to market benchmark. And the negative relationship between risk (Standard Deviation) and average return are ING Tactical fund (ING Funds Berhad) and AmTotal Return fund (AmInvestment Services Berhad) that provide investors with higher return as well as relatively low risk as compared to other funds.

While for Islamic fund, the positive relationship between risk (Standard Deviation) and average return shown by RHB Mudharabah fund (RHB Investment Management Sdn Bhd) that provides investors with higher return as well as relatively high risk as compared to market benchmark. And the negative relationship between risk (Standard Deviation) and average return is AmIttikal fund (AmInvestment Services Berhad) that provides investors with higher return as well as relatively low risk as compared to other funds.

Negative relationship between risk (Standard Deviation) and average return is good compared to the positive relationship; it is because those funds will provide investors with higher return as well as relatively low risk as compared to other funds. Overall, conventional fund has shown higher total risk (standard deviation of return) than Islamic fund. However, Islamic fund has shown higher coefficient of variation (risk-return) than conventional fund. Thus, it can be said that conventional fund have riskier company's portfolio performance in terms of total risk (standard deviation of return) and Islamic fund have higher coefficient of variation (risk-return).

Chua (1985) found out that unit trust funds in Malaysia performed better than the market during his study period, 1974-1984. He concluded that the performance of unit trusts was fairly consistent and fund managers have diversified and performed risk control reasonably well.

5.2.3 Conclusion for Sharpe's Ratio

The higher Sharpe ratio measured means a better company portfolio, it means that when one company has got higher Sharpe ratio, it performs better than the aggregate market and is expected to get high return and high risk in the future.

With the same market benchmark, three companies from conventional and Islamic fund have higher Sharpe ratio than KLCI market. Meaning that, only three companies for conventional and Islamic are performing better in market, while the rest is underperformed. The highest value of Sharpe ratio for conventional fund is 0.1254% from ING Funds Berhad (ING Tactical) and for Islamic fund is 0.1132% from Prudential Fund Management Berhad (PRUdana Dynamic). In conclusion, conventional fund shows better performance compared to Islamic fund in 2007.

Tan (1995) analyzed performance of 12 unit trusts over a 10-year period, 1984-1993. He concluded that unit trusts in general perform worse than the market portfolio. McDonald (1974) estimated the Sharpe, Treynor and Jensen measures for 123 mutual funds using monthly data for the period between 1960 and 1969. The findings showed that majority of the funds did not perform as well as the New York Stock Exchange (NYSE) index. It is evident that mutual fund is not significantly follow their benchmark.

5.2.4 Conclusion for Hypothesis

The relationship between return on KLCI (independent) and return on conventional and Islamic Equity fund (dependent) are analyzed. From the result, it shows that both the return on conventional and Islamic fund show higher calculated T-Statistic than T-Value of 1.960. It is found that the model used is significant and good variables because the F-Statistic and T-Statistic have a greater value than the F and T value from the tables. Thus, both conventional and Islamic equity mutual funds fail to accept Ho.

5.3 Recommendations

There are a few recommendations that can be made by the researcher after conducting the study; the researcher has divided these recommendations into three views:

5.3.1 Recommendation to the Investor

This study provides a result on average daily return and risk as well as Sharpe ratio involving under conventional and Islamic equity fund in Malaysia. The researcher advises the investors to analyze and make comparisons between the return and risk on both conventional and Islamic fund. It is because some other investors are more interested in Islamic investment and some are not. Based on average return, total risk and Sharpe ratio, both companies show similar results. As a whole, investors should invest in conventional than Islamic fund because of higher return and better performance in the market. However, it depends on the investors to choose either conventional or Islamic fund. The researcher also does not encourage investors to invest in funds that have a positive relationship between risk and return, or in other words, high risk and high return.

Otherwise, they should choose to invest in funds that have negative relationship between risk and return or low risk and high return. Thus, based on Sharpe ratio; if investors choose conventional fund, they should choose ING Tactical fund (ING Funds Berhad) or AmTotal Return fund (AmInvestment Services Berhad). If investors choose Islamic fund, they should choose AmIttikal fund (AmInvestment Services Berhad) and also PRUdana Dynamic (Prudential Fund Management Berhad). It is because this fund has a medium average return, but very low total risk and coefficient of variation.

5.3.2 Recommendation to the companies

The higher Sharpe ratios will show how strong company's portfolio performance and the company or investors can make decision on the best investment they can make. It is recommended that the underperformed companies to improve their company position in the market. For conventional fund, the funds included are Strategic Fund from HLG Unit Trust Berhad and Apex Dynamic from Apex Investment Services Berhad because the ratio is negative and located at the lowest ranking in the Sharpe ratio. For Islamic fund, the fund includes ING Ekuiti Islam from ING Funds Berhad.

5.3.3 Recommendation to Future Researcher

There are many other variables that can be used to investigate the performance of mutual fund such as Jensen and Treynor. Future researchers are advised to extend to a longer period and should use a large sample of equity such as domestic and global fund. It must cover return and risk performance compared to market portfolio that can contribute to better conclusive result.