“Performance evaluation of Saudi equity mutual funds: Fama decomposition model”

This paper is in pursuit of analyzing and elongating prior research on the performance evaluation of mutual funds by a comparative analysis with three categories of 82 Saudi equity funds during 2011 to 2016 using Fama’s decomposition model. The paper also made an attempt to explore the relationship with the risk reward ratio to the relative performance measure in predicting the future performance of the Saudi equity fund returns. The empirical results show that Saudi local equity funds perform better followed by Arabian and international/global equity funds in terms of expected signs and diagnostic tests.


INTRODUCTION
Asset and wealth management industry has long played a prominent role in the sphere of Saudi Arabian financial market. Saudi Arabia is a regional leader in mutual fund activity and has led to the recent GCC expansion of the segment, both in terms of fund classes and volumes, in areas such as private equity and venture capital. In 1979, Saudi Arabia became the first country in the Gulf Cooperation Council (GCC) to feature a mutual fund, a shortterm US dollar-denominated instrument established by National Commercial Bank, sparking an appetite for similar investment vehicles in markets around the Gulf (The Report: Saudi Arabia, 2016). As a regional leader and center of mutual fund activity, Saudi Arabia is the largest fund domicile in the GCC and the second largest in the Middle East and North Africa (MENA) region, accounting for USD 26.2 billion of the USD 79.6 billion MENA fund market as of mid-2015, according to Zawya. The saturated market of developed economies and the recent opening of Saudi Arabian Stock Exchange (Tadawul) to qualified foreign investors in June 2015, Saudi Arabian mutual fund industry and its performance of the funds grabs the attention of international investors. According to EY GCC wealth and asset management report 2015, as of July 2015, GCC mutual fund market accounted for USD 36 billion in assets, across 375 funds, the major contributor Saudi Arabia accounts for 80% of the total. Saudi Arabia hosts a total of 276 funds with 44 fund managers as of November 2016. Table 1 summarizes the category wise funds domiciled in Saudi Arabia. It is evident from Table 1 that Saudi mutual funds are highly concentrated on equities, which account for nearly 62% of the total funds. The present research study thus focused on analyzing the performance of equity funds -local, equity funds -international and equity funds -Arabian for being the reason of high concentration and the type of funds that primarily invest in equities.

PERFORMANCE EVALUATION OF MUTUAL FUNDS
While evaluating the mutual funds, two factors are considered important, namely investment performance and risk, involved among a vast number of factors that affect the mutual fund records (Kothari & Warner, 2001). As highlighted by Strong (2008) the essence of performance evaluation of any investment is to associate the returns realized with a measure of risk. It is also evident from the previous literatures like Treynor (1965) that traditional measures, though evaluating the returns with beta as a measure of systematic risk, ignore the diversifiable risk or residual risk. In contrast to Treynor (1965), Sharpe (1964) measures and assesses the total risk to realized returns. On the other hand, Jensen (1968) ranked the mutual funds based on Jensen's Alpha estimated by applying Capital Asset Pricing Model.
Even though these measures evaluate the risk and returns of investment funds, it is instruc-tive to investigate the reasons behind a superior or substandard performance of an investment fund. One method for doing so is Fama's return decomposition developed by Eugene Fama in 1972. As compared with any other performance evaluation measure, Fama decomposition measure manifests a superior performance in evaluating mutual funds competence to earn excess returns and also decomposes the returns at different risk levels. Fama decomposes the excess return from the risk free rate into two components: risk premium for bearing risk and risk premium due to selectivity. On the other hand, the total risk is decomposed into manager's risk and investor's risk.
The present research thus attempts to segregate the investment performance of Saudi equity funds into three components: the returns investor chose to take, the added return the manager chose to seek and the return from the manager's good selection of securities. A further attempt to identify the discrete contribution of each performance measure that greatly influences the fund returns using regression model is also made.  Number of funds   1  Equity funds -local  102  2  Equity funds -international/global  28  3  Equity funds -US  4  4 Equity funds -European 3 5

LITERATURE REVIEW
Literatures on performance evaluation of mutual funds date back in 1960's, initiated by Treynor (1965) who introduced the concept of fund performance and a measure for rating fund management performance. As further extension to Treynor's work, many performance evaluation measures were proposed by Sharpe (1964), Jensen (1968), and Fama (1972)  At the same time, Barakat, Nazmy, and Al-Jabli (2011) made an attempt to identify the constraints affecting the efficiency of mutual funds in Saudi Arabian financial market. The study identified weakness in the organizational structure including the management style and lack of expertise as few constraints and proposed a set of recommendations to overcome the constraints. The study concluded that even though the markets were informationally inefficient, lots of opportunities to realize abnormal returns were present in both the markets. But the markets were unable to realize this opportunity to transform this to investors in the form of affirmative returns.
Over the same period, Al Hamdan Anas (2012) examined the investor's attitude towards Zakah, its subjective norms and the intention to pay Zakah on Saudi mutual funds using the theory of reasoned action. The study concluded that the inves-tors' attitude and the subjective norms greatly influence the Zakah compliance among owners of mutual funds.
A study by BinMahfouz and Hassan (2012) compared the characteristics of conventional and Islamic equity mutual funds in Saudi Arabia. As compared to their market benchmarks and conventional fund, the risk adjusted performance measures of Islamic mutual funds in Saudi Arabia remain unstirred with Sharia screening process. The systematic risk analysis results of the study highlighted that the Islamic funds in Saudi Arabia are less exposed to market risk compared with conventional and benchmark funds.

DATA SOURCES
To analyze the performance of the Saudi equity funds, a data set is created with a total of 82 equity funds comprising 45 mutual funds listed under the category of equity funds -local, 20 funds listed under the category of equity funds -

PERFORMANCE EVALUATION MEASURES OF SAUDI EQUITY FUNDS
Proper performance evaluation always involves a recognition of both the return and the riskiness of the investment. For measuring the returns, the monthly NAV of the funds has been used in this paper. Standard deviation is used to capture the total risk and beta to measure the systematic risk of the investment.

Fund returns
The NAV of a mutual fund as a bottom line in evaluating its performance is the market value of the securities owned by the fund in addition to due receivables and cash after deducting any fund commitments. Indefinitely, the percentage returns of the fund is thus measured by comparing the prices of mutual fund's NAV at the beginning and the end of the investment period.

Risk measures
The essence of performance evaluation of mutual funds greatly lies in associating a measure of risk with the fund returns. The degree of risk taken in the fund by the manager must be completely evaluated regardless of the fund returns.

Standard deviation: proxy for total risk
Standard deviation, the measure of dispersion, is used to capture the total risk of the investment, besides, beta measures the systematic risk of the same. To measure the fund's total risk, we use the equation: where : Rp monthly return of the mutual fund , p : Rp average monthly return of the mutual fund , p : n number of observations.

Beta: proxy for systematic risk
To estimate the systematic risk of the funds, Beta measure is calculated initially. Beta measure investigates the average sensitivity of an individual mutual fund in response to the market returns: .

Covariance between fund return
and benchmark return Variance of benchmark return β = In calculating the benchmark returns, the present study used the month end closing balances of Tadawul All Share Index, calculated as

Fama's components of investment performance
Fama's decomposition measure shows how to fragment the funds realized returns into three components: 1. Excess returns: the return the investor should earn for the risk he/she chose.
2. Compensation for systematic risk: additional return for the risk chosen by the manager.
3. Return from selectivity: returns from the manager's ability in selecting the good securities.
Return from selectivity is further decomposed into compensation for diversification and net selectivity. Diversification measures additional return that compensates the portfolio manager for bear- ing diversifiable risk. After accounting for diversification, the residual performance on selectivity is attributed to net selectivity (Saritha, 2012).

EMPIRICAL RESULTS AND DISCUSSION: OUTCOMES OF FAMA'S COMPONENTS OF PERFORMANCE
The portfolio performance evaluation on the selected 82 Saudi equity funds carried out using the Fama's components of performance measure are shown in Table 2. a great number of funds, the total excess returns earned by most of the mutual funds are comparatively negative. The total excess returns resulted negative as the mutual funds were able to earn the returns less than the risk free rate.

Compensation for systematic risk of Saudi equity funds
In mutual fund investments, the incremental returns that the investors earn for taking some risk than the target level are labelled as investors' risk and the added returns the managers choose to seek for their systematic risk constitute risk premium. The additional returns earned by the market above risk free times the mutual fund's beta represent compensation for systematic risk.
The study identifies that the Saudi equity funds strive hard to compensate their investors for the systematic risk inherited as the market returns during the sample period is comparatively low. Even though the compensation for systematic risk earned by the funds is negative, the fund managers are active on superior stock selection and market timing to beat the benchmark, which is reflected in the funds returns on selectivity.

Return for selectivity of Saudi equity funds
Return for selectivity is the returns earned by the fund over the returns on the Security Market Line.
As it is evident from  are able to select superior stocks that are available within the Saudi market.

Compensation for improper diversification of Saudi equity
fund Compensation for improper diversification is the difference the return corresponding to the beta implied by the total risk of the portfolio and the return corresponding to its actual beta, i.e., systematic risk (Strong, 2008). With the increasing size of the portfolio, the diversifiable risk increases. Apparently the diversification return should be close to zero for well diversified portfolios, as the beta of actual and implied will be almost equal.  Table 2 shows that 41 percent of the equity funds -Arabian, 10 percent of equity funds -international and 8.8 percent of the equity funds -local, are perfectly diversified, whereas majority of the funds suffer of diversification. It can be further interpreted that though most of the sample funds deprive proper diversification, the long-term track records of the funds have to be monitored. Apparently, the fund managers also have to strive hard in the area of security analysis to create well diversified and better performing funds.

Net selectivity of Saudi equity funds
Net selectivity estimates the portion of the return for security selection in excess of the returns imparted by the diversification component (Strong, 2008 Table 2 unveils that 86.67% of Saudi equity fundslocal, 80% of equity funds -international and 82.35% of equity funds -Arabian have earned positive net selectivity depicting the superior performance of the fund managers ability in stock selection. Surprisingly the study identifies that majority the sample funds though earned positive net selectivity, their compensation for improper diversification is negative. Apparently, the research infers that the fund managers' stock selection ability is though superior they lack proper diversification of their portfolios.
Positive net selectivity of the vast majority of the funds indicate that though the fund managers did a fairly good job in generating positive returns, the numbers are not striking owing to reduced percentages.
In consideration of the bright future of the Saudi mutual funds market, an attempt is made by the researchers in predicting the future performance of the Saudi equity funds using econometric techniques.

Relationship of risk reward ratio
to future returns of Saudi equity funds

Regression model for equity funds -local
As an extension of further analysis on the outcomes of Fama decomposition, a regression model is developed with risk reward ratio as dependent variable for each of the fund performance categories during the sample period. Risk reward ratio in general estimates the prospective future returns for each dollar we risk and is calculated as a ratio of current fund returns to the total risk. This model is carried out to identify the discrete contribution of each performance measure that greatly influences the fund returns using the statistical software E-views. 12 3 . -statistics 0.00 Prob F As the above results show, each of the estimated regression coefficients is individually and statistically highly significant. So, we reject null hypothesis of that each individual independent variable influence on the relative return equals zero. Collectively, all three variables are also highly statically significant, because the p-value of the computed F-value (for 3 and 41 d.f.) of 178.3425 is extremely low.
The most significant findings of the above model is that all signs of the model parameters as expected, are positive indicating that incorporation of related risk to the model will bring better results in the fund returns, which reflect the theory and the reality of the analysis. Partially, the crucial variable among the individual variables of the model is the Compensation for systematic risk (Sis), where it has the highest influence on the relative return of the equity funds. A one percent point change in this variable will enhance the relative change of the dependent variable by 12.702%. The results of the model shows that the Compensation for systematic risk will play an important influence on the future return of Saudi equity funds -local. As stated earlier, as the sample funds as a whole lack compensation for systematic risk, a change in the risk inheritance might bring attractive fund returns in the future. Compensation for improper diversification (dr) also has a vital effect on the return on equity funds -local, as the coefficient size is 8.491.
Despite that the Net selectivity (Ns) has the lowest impact on equity return, but incorporating net selectivity will enhance the return of the Saudi equity funds -local as the coefficient of determination 2 0.9288, r = which is very high. This means that approximately 93% of variation of the relative return on equity funds -local are explained jointly by the variation of all the three independents variables of this model, so we can reject the null hypothesis that all partial slopes simultaneously equal to zero H0: 223 0 γγγ = = = or 2 0. r =

Model test
With an attempt to preserve the validity of the data and variables included in models and to maintain the robustness of the results, the Classical Linear Regression Model (CLRM) will be tested using autocorrelation and heteroskedasticity tests. Towards that end, to detect the presence of autocorrelation in the residuals, Durbin-Watson test and White's heteroskedasticity test is used to examine the existence of heteroskedasticity on the model. The (D-W) value of the model at 1.826, which is closer to 2 shows more evidence of no autocorrelation. For heteroskedasticity, the results showed that * n ( 2 R of the auxiliary regression model) 2 1 k χ − ≈ equals 24.99, which is greater than critical value of 12.6, and is also significant at 5% level, thus rejecting the null hypothesis.

Regression model for equity fundsinternational/global
A regression model estimating the risk reward ratio relationship with performance measures of equity funds -international ( ) RrI takes the following form: 12 3 . The model indicates that the risk reward ratio is less influential on the fund performance measures dr and Sis. Inclusion of more risk in international equity funds will have a positive influence on the net selectivity returns of the funds, while holding the other variables constant, since the p-value is highly significant for this variable. Collectively, the model is highly significant in term of F-test, where F-value is quite large (169.75), and the probability of F is significant at 1%. Diagnostic tests were conducted to test the validity of the estimated model with the test for heteroskedasticity by deploying White's heteroskedasticity test again. The calculated 2 χ is 11.62, which is greater than critical values, and this value is insignificant at any level, so we accept the null hypothesis of homoskedasticity. Secondly, the autocorrelation test has been estimated using Durbin-Watson test. The value of D-W test approximately equals 2.0 (1.997), which means that we accept the null hypothesis of no autocorrelation (absence of autocorrelation).

Regression model for equity funds -Arabian
Classic Linear Regression Model similar to the above two models on Arabian equity funds ( ) RrA takes the following form: 12 3 . The signs of all parameters are as expected and highly significant at 1%. It can also be further interpreted that incorporating increased risk in Arabian equity funds are expected to generate more of fund returns. The proposed model is high-ly significant at 1% with 2 R of 97% and F-value at 126.795. The result of White's test also reveal sthat we accept the null hypothesis of homoskedasticity (no hetero), at 10% level of significance. Concerning the autocorrelation test, the value of Durbin-Watson is around 2 (1.95), which indicates the rejection of any kind of autocorrelation.
A close examination of the above three models reveal that among the three equity funds categories, local funds perform superior compared to the other two followed by equity funds -Arabian. As the findings of the study reveal that local equity funds performance is superior, a further attempt is made to implement Vector Auto Regression (VAR) to capture the relative importance of various shock and their impacts on the performance variables.

Variance decomposition
The investigation of the dynamic relations among the variables in the particular sample period is carried out through variance decompositions. The unrestricted Vector Auto Regression (VAR) model is implemented to capture the relative importance of various shock and their impacts on the variables of our concern. Tables 3 to 6 show the variance decomposition of all variables of the model; the relative significance of each structure shock to other variables. Table 1 indicates that over 10 period ahead, 100% of relative return can be accounted for its own innovation in the 1 st period. It was noted as time goes on; its contribution will be high till last period. Therefore, we can say that over the 10 years risk reward ratio of local equity funds is highly explained by its own shocks. The succeeding contribution of shock of relative return comes from the shock of the net selectivity; where the highest was approximately over 6% starting from the 6 th period. This contribution continue to be the same until last period; the 10 th . The other essential following contributions to the relative return come from net selectivity; where the highest 72.78% is as indicated in Table 3, followed by diversification; where the highest was 64.18% and systematic risk was around 45%. All of the above results are consistent with the estimated regression model.

CONCLUSION AND SCOPE FOR FURTHER RESEARCH
The primary focus of this study is to evaluate the performance of the Saudi equity funds listed on Tadawul Stock Exchange using Fama decomposition measures. The empirical results shows that Saudi equity funds -local perform better followed by Arabian and international/global equity funds. The re-   gression model developed on local funds is highly significant in terms of expected signs and diagnostic tests indicating that a change in the risk inheritance level might bring attractive fund returns in the future. The results of the variance decompositions also reveal that more relative important changes come from the changes in net selectivity, diversification, and systematic risk.
These results suggests that Saudi Arabia is prepared for a new stage of development including mutual funds, which will see the Tadawul integrate further with global capital flows in the new era. The development of capital markets in the region is promising and the asset management industry is set for a significant growth in the future, which merits further investigation of different investment vehicles under different market situations and provides an avenue for future research.