Ethical Dow Jones Indexes and Investment Performance: International Evidence

This study examines the relative importance of the Shariah-Compliant Dow Jones market indexes to capture the dynamic behavior of stock returns at economy and industry levels. The analysis indicates that ethical investment has only an insignificant influence on the performance of stock market returns for both the economy and industry levels. Further, alternative measures of investment performance including the Carhart and Habit Formation models have been used to examine the behavior of the Shariah-Compliant Dow Jones market indexes. The findings suggest a negative market timing ability with both Islamic and conventional indexes. While Islamic indexes are growth focused, conventional indexes are value focused. Further, when investigating the performance of Islamic and conventional Dow Jones indexes during the recent financial crisis, there is evidence supportive of Islamic indexes against conventional ones. For sector groupings, the results indicate that parameter estimates are not consistent, suggesting that Islamic indexes are sector oriented. These results are explained to be a consequence of less diversification in Islamic indexes, leading to higher risk in some sector groupings such as technology and consumption services.


Introduction ©
The topic of Islamic Finance has received significant attention in the financial press, in particular during the recent global financial crisis. According to Sherif and Shaairi (2013, p.27), "it is no longer a niche product serving a specialised market but is now offered in more than 60 countries, with total assets in Islamic banking reportedly exceeding $1.2 billion, Islamic mutual funds estimated to be valued over $58 billion, and issuance of Islamic sovereign and corporate bonds, or Sukuk, to be about $84 billion in 2011." Further, an increasing international tendency towards Islamic finance has gained greater attention and recent dimensions have emphasized its prevalence in the global markets. Its move from a merely banking-based industry into broader aspects of market-based instruments have made Islamic capital markets the most rapid growing sector in the Islamic finance industry, and they have witnessed unprecedented expansion over the last decades (Dewandarua et al., 2015). This expansion may be due to the large growth of the capital value of Muslim investors and their strong demand for Shariah-compliant investment avenues, which prohibits interest (riba), excessive risk-taking (gharar), or gambling (maysir), and concomitantly promotes risk-sharing, profit-sharing, and assetbacked financial transactions (Zaherand Hassan, 2001;El-Gamal, 2006;Iqbal and Mirakhor, 2011;Sherif and Shaairi, 2013).
Within the significant developments of Islamic financial system, the faith-based ethical Islamic investment industry, which is regarded as a subset of  (1952). It has been claimed that ethical investments tend to underperform in the long run because they are subsets of the market portfolio and lack sufficient diversification (Bauer et al., 2006). According to Hickman et al. (1999, p. 73), diversification is a "consequence of the imperfect correlations of returns between securities". Consequently, commonly investors tend to diversify their portfolios in order to minimize their risk and maximize their returns. However, several previous studies do not show a general consensus that ethically screened firms outperform their non-ethical screened counterparts (see, for example, Diltz, 1995;Guerard, 1997;Sauer, 1997 Electronic copy available at: https://ssrn.com/abstract=2953441 Furthermore, although the Islamic finance industry is growing rapidly, and much is known about the performance of conventional indexes in the developed countries, there is a paucity of literature that has investigated the faith-based ethical investments issue in general, and in the developing countries in particular (Annuar et al., 1997;Abdullah et al., 2002;Hussein, 2004;Elfakhani and Hassan, 2005; Girard and Hassan, 2008;Dharani and Natarajan, 2011). Inspired by the above arguments and the stages of evolution of the Islamic financial services seen in Table 1, this study examines and provides a new evidence on the impact of Shariah filtering criteria on the performance of Dow Jones Islamic market indexes relative to their conventional counterparts. To do this it considers seven Islamic regional and five sectorial indexes and conventional counterparts (Global, Asia Pacific, European, USA, UK, Developed markets, Emerging markets; Oil and Gas, Technology, Health Care, Consumer Goods, Consumer Services indexes). To this end, different standard performance ratios, the CAPM, Carhart (1997) model and the Habit Formation model of Campbell and Cochrane (2000), which take into account the financial risk time-variation, were estimated in order to provide precise investment performance evaluations. Thus, this study contributes to the existing literature in several ways: (1) while most empirical and theoretical research focuses heavily on investigating the performance issue using the standard performance ratios, there is only limited academic research conducted on a range of performance measures including recent promising asset pricing models. As this study does this, it enables the risk to be time-varying and have an affect on the scrutiny and quality of the empirical results; (2) this study focuses on different international regions including transition economies, Western countries and the whole World; (3) although the performance of the external habit formation model of Campbell and Cochrane (2000) has been widely examined in the literature using conventional indexes (Hyde and Sherif, 2005), there are only a few studies that have investigated the same relationship using faith-based ethical indexes; and (4) the impact of the global financial crisis on the performance of investment indexes, in particular the ethical investments, is an important and ultimately new empirical question; (5) to this author's knowledge, Treynor and Mazuy's (1966) model has not been adopted to assess the timing ability of Islamic indexes in the previous studies. The remainder of this paper is structured as follows. Section 2 is a brief literature review of studies that have considered the faith-based ethical investment/ indexes. Section 3 provides details of the methodology, standard performance measures and models. Section 4 presents the data and empirical results, and section 5 concludes the paper, stating the Electronic copy available at: https://ssrn.com/abstract=2953441 significance of the main findings and suggesting avenues for future research.

Literature review and hypotheses
There has been a long running debate in the academic literature regarding the performance of ethical investments over conventional investments (Hamilton et al., 1993;Luther and Matatko, 1994;Diltz, 1995;Mallin et al., 1995;Sauer, 1997 Hakim and Rashidian (2002) investigated the performance of Dow Jones Islamic market indexes and again found only insignificant differences between the performance of conventional and non-conventional indexes. In another study, Rahmayanti (2003) investigated the performance of a Shariah-compliant portfolio in the Jakarta Stock Exchange using the return, risk, Sharpe, Jensen, and Treynor ratios along with Shariah-based indexes and the Jakarta Composite Index (JCI). They found mixed results across the different periods of the study. In the same vein, Elfakhani and Hassan (2005) analysed the performance of 46 international Islamic mutual funds during the period 1997-2002 using standard ratios (Sharpe, Treynor, and Jensen Alpha). They examined the performance in different economic conditions using S&P 500 to present conventional benchmark and Dow Jones Islamic index as an Islamic benchmark. They found that the performance of the funds was consistent using different models and benchmarks. Overall, they found no statistically significant risk adjusted abnormal reward and the performance of Islamic mutual funds showed no significant difference to conventional funds. In another study, using US Dow Jones Islamic Indexes, Khathatay and Nisar (2007) examined the Shariah screening rules. Overall, (1) they found that the Malaysian SEC is the most liberal and DJIM was the most conservative; (2) argued that total assets is a superior input to market capitalization when using financial ratio; and (3) that the Islamic finance industry can be promoted by establishing a Shariah rating agency.
Using different faith-based ethical indexes, Hussein (2004) investigated the performance of ethical investments using FTSE Islamic indexes. Their study compared the performance of FTSE Global Islamic index and FTSE4Good to FTSE all-world index during the period 1996-2003. Their findings indicated that the Islamic index performed the same as the FTSE all-world index during the entire period. However, in contrast to the bear market period, in the bull market period the Islamic index outperformed its conventional benchmark. Overall, the study found no clear supportive evidence of the ethical investment compared to its unscreened benchmark.
In addition, Albaity and Ahmad (2008) investigated the performance of the Kuala Lumpur Shariah index (KLSI) and the Kuala Lumpur Composite index (KLCI) in Malaysia using risk adjusted performance measurement, causality and the Johansen cointegration test. Their findings showed only an insignificant return difference between both indexes. Similarly, Girard and Hassan (2008) developed their previous study by analysing the cost of faith-based Electronic copy available at: https://ssrn.com/abstract=2953441 investing using FTSE Islamic and non-Islamic indexes during the period from 1999-2006 and by using Sharpe ratio, Treynor ratio, Jensen's Alpha, Carhartmodel and Johansen cointegration tests. They concluded that investors will have similar rewards to risk and diversification benefits exist for Islamic and non-Islamic indexes with only an insignificant difference between the performances of the two indexes. While Krussl and Hayat (2008) argued that there is insignificant performance difference between IEFs and Islamic and conventional benchmarks during normal market conditions, the studies of Mansor  One further argument that has recently been given much attention is related to the asset pricing model as a measure of performance of Islamic investment structure. Here, Hakim and Rashidian (2002) examined the performance of CAPM and found that the DJIMI performed well when compared to the Dow Jones World (DJW) index, but under-performed the Dow Jones Sustainability (DJS) World Index. Other studies (for example, Hussein, 2004Hussein, , 2005Girard and Hassan, 2005) have investigated the issue further and have suggested that Islamic indexes under-perform during the bear period, but outperform during the bull period, with a clear justifications for investing in growth and small-cap firms. In the same vain, Hussein and Omran (2005) investigated the performance of the Islamic index using the Dow Jones and the Dow Jones World Index, through using CAPM, Sharpe ratio, and Treynor ratio. They found evidence supportive of the performance of Islamic indexes for both the entire period and the bull market period. However, other periods failed to yield economically plausible parameter values. Furthermore, Hassan et al. (2005) investigated the performance of Dow Jones Islamic indexes and MSCI indexes using standard financial performance measures (Sharpe, Treynor, Jensen's Alpha ratios) and Fama and French's model. The indexes include seven geographical areas: Canada, the United Kingdom, the United States, Europe, Emerging Markets, Asia and the World over the period 1996-2005. In general, they found only insignificant differences between Islamic and non-Islamic indexes along with a comparable level of rewards to risk and diversification benefits.
More recently, Mansor and Bhatti (2011) used the CAPM model to examine the monthly performance of 128 Islamic mutual funds in Malaysia IMFs over the period 1990-2009. They found that Malaysian Islamic mutual funds outperformed their traditional benchmarks. Their results also indicated that Islamic and conventional investment decision makers are equipped with substantial positive stock selectivity skills and market timing competence. When comparing IMF with their conventional counterparts, they found IMF fund managers offered significant performance in stock selectivity, but only found insignificant findings for their market timing ability relative to their counterparts. Similarly, Hayat and Kraussl (2011) investigated the characteristics and behavior of the return and risk for 145 open-ended Islamic mutual funds over the period 2000-2009 using a range of approaches to compute Jensen α, systematic risk β, downside risk and market timing ability. They found that the Islamic equity funds provided significantly lower performance compared to the non-Islamic counterpart, in particular during the times of economic turmoil in 2008. They also identified IEF managers as being weak marketing timers whose decisions resulted in the loss of returns. In addition, Abbes (2012) examined the risk and return behavior of a range of Islamic and conventional benchmark indexes across 35 international developed, emerging and GCC markets over the period 2002-2012. Using the common statistics t-test, Abbes (2012) found only insignificant differences between Shariah and conventional counterparts except in Australia and Italy. Their study also examined the relationship between returns and volatility and found only insignificant differences between Islamic and conventional indexes and that the level of debts affects risk across all corresponding market. In the same context, Becketal.(2010) investigated the efficiency hypothesis for Islamic and conventional banking systems, and Cihak and Hesse (2010) investigated the financial stability hypothesis for two Islamic banks and found inconclusive and mixed findings. Elsewhere, Milly and Sultan (2012) conclude that the money invested in Islamic stocks is safer for periods of economic financial distress, and Similarly, Hayat and Kraussl (2011) examined the performance of Islamic (IEFs) and conventional equity funds over the period 2000-2009 and found that the conventional indexes under-performs compared to Islamic indexes. In conclusion, there is no conclusive evidence that ethically screened investments under-perform conventional investments. Thus, the argument about a financial penalty for being an ethical investor is debatable. Importantly, previous studies found evidence that the relative performance of faith-based ethical investing varied across geographical areas due to their characteristics of merit. Given the findings from the literature provided above, this study argues that there is no impact for faith-based ethical investment on the investment performance. Hence, the hypothesis can be identified as follows: Electronic copy available at: https://ssrn.com/abstract=2953441 H0. There is no significant difference of stock returns between Shariah and Conventional investments.

Performance measures.
There is a range of financial measures and asset pricing models that have been recently used to assess the performance of ethical and conventional investments. This section highlights the standard performance measures that have been used in the financial literature.
Absolute risk-adjusted performance measures

Sharpe Ratio (1966)
This ratio was advocated by William Sharpe (1966) and measures the average return on a portfolio in excess of the risk-free rate of return, or the risk premium of a particular portfolio contrasted to the total risk of a portfolio measured by its average deviation. For example, if the return on stock investments is ≤ the risk-free rate, then it indicates no need to invest in risky assets. Consequently, the Sharpe ratio (SR) is a performance measure for portfolio compared to the risk taken. In other words, if SR is significantly higher, then the performance will be much better and the profits for taking on additional risk greater. Importantly, SR is sometimes erroneously identified as a risk-adjusted return and portfolios can be ranked in order of preference, however, it is not easy to gauge the size of relative performance (Leland, 1999). In addition, SR is inappropriate when returns are unexpectedly nonnormal. The SR for portfolio p, which is initially called the reward-to-variability ratio, is then identified as: where (R p ) is the expected return on investments or index; E(R f ) is the risk-free return; δ p is the standard deviation(SD) of the return on investments and defined as: where n is the number of return observations in the sample.

The Treynor Black Appraisal Ratio (AR)
The AR is a further developed version of Jensen's α and the relevant risk-adjusted performance statistic when evaluating new investments. It measures the systematic risk adjusted reward per unit of specific risk taken. AR, which was first advocated by Treynor & Black (1973), is comparable in concept to SR. According to Sharpe (1994), the appraisal ratio is set with the assumption that the risk-free asset is substituted by a benchmark portfolio and identified as: where α is the Jensen's α p of the portfolio and σ is the non-systematic risk.
Jensen's α is the excess return adjusted for systematic risk in the numerator divided by the portfolio's nonmarket risk (i.e., unsystematic risk) in the denominator.

Excess Standard Deviation Adjusted Return (eSDAR)
eSDAR was suggested by Statman (2000), and measures the excess standard deviation adjusted return, which modifies SR and is the short-term excess return of a portfolio over the market return. The higher the value of eSDAR, the higher the returns on the portfolio. This measure is identified as: where R i is the return on index i; R m is the market return; R f is the risk free rate of return; SD i and SD m are standard deviations of the index and its benchmark market index respectively.

Treynor Ratio (1965)
This ratio measures the association between the excess return on investments and its systematic risk. It is drawn undeviatingly from the standard CAPMs.
To measure this ratio, a benchmark index is needed to help estimating the β of the portfolio/investments. It is worth noting that the Treynor ratio is appropriate for a well-diversified portfolio, as it simply seizes the systematic risk (undiversified risk) of the portfolio (Srivastava and Essayyad, 1994) when measuring the performance. Hence, this ratio is frequently used when the portfolio is part of a fully diversified index and is identified as: where R i is the average return on index i; R f is the average risk free rate of return; β i remarks the beta of index i.

Relative risk-adjusted performance measures
These groups of measures evaluate the funds' riskadjusted returns in reference to a benchmark.

Jensen's Alpha (1968)
Jensen's α is based on the excess returns and riskadjusted returns estimated by the standard capital asset pricing model CAPM. While the positive (negative) αimplies that the index is outperforming (under-performing) stocks, zero alpha implies that Electronic copy available at: https://ssrn.com/abstract=2953441 the index performance is normal as expected in CAPM, and has no excess returns over the systematic risk. One significant difference between CAPM and Jensen's α is that the former is based on expected returns while the latter is based on realised returns. This is beneficial for both researchers and investors who have access to historical (realised) data. In addition, it is the type of return that does not bear them the cost associated with systematic risk, as it is positive even when β is zero. However, Jensen's α unreasonable assumes that β is stationary, as investors accustomed to moving between sectors and assets classes with a significant change in β. Furthermore, α is often criticized as being a proxy for other factors that determine returns except market exposure. This indicator is defined as: where i α is the intercept and quantifies the exceeded returns over a given systematic risk. It identifies the percentage of additional return that is due to the investor's choices; i β is the systematic risk of index i; R i is the return on index i; R m is the return on the market index; and R f is the risk free rate of return.

Treynor and Mazuy Model (1966)
This measure, which is known as the Market Timing Ability (γ) model, was advocated by Treynorand Mazuy(1966), who introduced a measure which allows for the ability of investment decision makers to partially shift their investments between safe financial assets (debts) and risky financial assets (securities) depending on whether the market is expected to go up or down 1 . Unlike Jensen's i α or CAPM, the TM model adds a quadratic term or relationship between excess returns on investments and excess returns on markets when timing the market is successfully managed. This implies that investors will increase their investments when the market is up, and γ i is positive and statistically significant. When investors anticipate a rise in the market, they increase their portfolio's β,which enables them to make higher profits. The model is identified as: where R i is the return on index i; R m is the market return; (R m ) 2 is the squared market return, R f is the risk free rate of return, i α is an intercept and quantifies returns over a given systematic risk, i β is 1 Theoretically validated by Jensen (1972) and Bhattacharya and Pfleiderer (1983).
systematic risk i, i ε is an error term, and i γ is a market timing measure.

M 2 Measure
M 2 is an extension measure of the Sharpe ratio (SR), which quantifies the risk-adjusted performance (RAP) of a portfolio relative to the benchmark market index. It is worth noting that M 2 measures the performance of an index relative to the market rather, in contrast to its being an absolute measure like the SR. The better the index, the higher the RAP value 2 . This can be identified as: where SR i is the Sharpe ratio SR of index i;SR m is the Sharpe ratio SR of the market index; and m δ is the standard deviation of the market index.

Asset pricing models. 3.2.1. Fama and
French's Three-Factor Model. The Fama and French (1993) three-factor model controls the event clustering and cross-correlation in returns on investments and is identified as: where R it is the expected return on asset i; R ft is the risk-free rate of return; R mt is the market return; SMB (small minus big) is the difference between returns on a small capitalisation portfolio and a largecapitalisation portfolio; and HML (high minus low) is the difference between returns on a portfolio with a high book-to-market ratio and a portfolio with a low book-to-market ratio. According to Fama and French (1996), the SMB and HML portfolios are organized by market capitalisation and book-to-market value. Thus, the SMB t (Small minus Big) factor is calculated as follows: Carhart's 4-factor model is an extension of the 3factor model of Fama-French (1993). The additional factor is a momentum factor (MOM t ), which is constructed following Carhart(1997) as the average return on the past winner stocks minus the average return on the past loser stocks.
The monthly size breakpoint was the median market value. The return breakpoints were the 30 th and 70 th percentiles. The MOM t variable was therefore defined as: 3. The Habit Formation Model. As evidence against the reliability of the CAPM has accumulated, recent studies and developments in asset pricing have placed an emphasis on the success of the habit formation model of Campbell and Cochrane (2000).
Since differences in investment performance may be explained by differences in systematic risk, size, and value factors (in addition to habit formation specification), the study detailed in this current paper provides new evidence on the performance of faith-based ethical investments by estimating the risk-adjusted returns/performance using the universally accepted habit formation approach of Campbell and Cochrane (2000). The fundamental idea is that when the consumption of investors is low (high) relative to their habit, relative riskaversion is high (low), leading to high (low) expected returns/performance. Campbell and Cochrane (2000) identified the utility function as 3 : 1 ( ) Campbell and Cochrane (2000) define S t (surplus consumption ratio), or the difference between consumption and the habit level as: Using the following Euler equation, investment performance can be examined as: ( ) ( ) where the stochastic discount factor, M t+1 is ( ) Secondly, another strand of research (Sadaf and Andleeb, 2014) has paid significant attention to Zakat as a proxy for risk-free rate of return, meaning that the minimum return required by Muslim investors should cover the amount deducted for Zakat (2.56%), or alms giving, required by each individual Muslim (Sadaf and Andleeb, 2014). Here, Islamic CAPM is specified as: where Zakat% is a proxy or risk-free rate of return and risk premium is identified as:(R m -Zakat%).
Thirdly, much attention has been given to nominal gross domestic product growth (NGDPG) as a proxy for risk free rate of return (Sheikh, 2010; Sadaf and Andleeb, 2014). Consequently, the investor's required rate of return will be based on nominal GDP growth rate and risk premium is measured as beta associated with the difference between return on market and GDP growth rate. Here, the Islamic CAPM is identified as: Finally, inflation was used as a proxy for risk-free rate of return (Hanif, 2011;Sadaf and Andleeb, 2014). By including the inflation factor as a proxy for the risk-free rate investors in the Islamic-majority Electronic copy available at: https://ssrn.com/abstract=2953441 countries, they will gain the same purchasing power over time. Consequently, CAPM is: For the investigation conducted in this paper, the 3month Treasury bill is adopted as the proxy for the risk-free rate. This is justifiable, as it is considered compensation for the excessive inflation that is common in most Muslim countries. Indeed, inflation leads to a significant level of reduction in investor wealth, and hence investors should be compensated by a rate equivalent to that of inflation.

Methodology.
This section details the methodology that has been adopted in this study to achieve its objectives. The parametric t-statistic is utilized to test the null hypothesis associated with the standard financial ratios 4 . 1 To test the behavior of both Dow Jones faith-based ethical (Islamic) and conventional indexes, this study estimated the return on a monthly basis by using the log difference of the price index. This is identified as: where R i,t is the raw returns for index i at time t and price i,t is the price of index i at time t.
To test the null hypothesis H0, the t ratios of the α and β are obtained by dividing the means of α and β by their standard errors. The t ratios for α and β are identified as: The t-statistics are then compared to the critical t statistic. The null hypothesis H0 is rejected if: | t | < t h e critical t.
Since the raw returns are not adjusted for risk, and the ethical (Islamic) indexes and their counterparts are not from the same category of risk, the Capital Asset Pricing Model (CAPM) is utilized in order to estimate the risk-adjusted returns: where α i,t is an intercept or Jensen's measure of performance and β i,t is the risk factor for index i at time t relative to the benchmark m. If α i,t is positive and statistically significant, then the index i outperforms the market index m.
To test the performance of the Treynor & Mazuy model (market timing ability) and the capital asset pricing models, the OLS and Generalized Method of Moments (GMM) were adopted for both ethical and matched conventional indexes. The GMM estimator uses internal instruments; specifically, instruments that are based on lagged values of the explanatory variables that may present problems of endogeneity.
To be exact, all the endogenous right-hand-side variables in the model lagged from t -1 to t -2are used. To check the validity of the model specification when using GMM, the Hansen statistic of overidentifying restrictions to test for any absence of correlation between the instruments and the error term is adopted 5 . 2

Data.
The data adopted in this study includes monthly prices of Dow Jones market indexes over the period January 1999 to July 2013. Seven regional and five sector grouping indexes as seen in Table 3 were adopted to investigate the difference between the performance of faith-based ethical (Islamic) indexes, and conventional indexes. The data were obtained from and provided by various sources. For the portfolio indexes, the Islamic and conventional indexes were obtained from the Dow Jones market index family of Datastream 6

Descriptive analysis.
The analysis begins with the descriptive analysis. Table 4 presents the riskreturn characteristics and summary statistics (mean, standard deviation, minimum, median, maximum, Kurtosis and Skewness) pertaining to Islamic indexes and their mainstream conventional counterparts. Table 4 panels A and B show that the Islamic market indexes display higher mean returns than the conventional market indexes but that they are also slightly more volatile. Most of the returns are non-normal, with evidence of negative excess skewness. This shows that the mean return of the Dow Jones Islamic indexes ranges from 0.15% to 0.35% where the DJIAP presents the lowest mean return and DJIC is the one with the maximum mean return performance. The returns of Islamic Indexes is highly volatile for DJIC and DJIT with standard deviations of 3.83% and 3.78% respectively, while the least volatile indexes are DJIHC and DJICG, with standard deviations of 1.71% and 1.77% respectively. The skewness demonstrates that most indexes are negatively skewed, indicating a distance value far from the normal one. For the conventional indexes, the mean return ranges between 0.01% (DJAP) and 0.33% (DJC).The standard deviation, which is the measure of spread of log returns illustrates that the DJT and DJEM have the highest deviation with 3.72% and 3.55% respectively, while DJHC and DJCG have the lowest deviation with 1.69% and 1.73% respectively. Similarly, the skewness which measures the asymmetry of the probability distributions shows that all indexes are negatively skewed, indicating the higher probability of decrease in returns. These statistics suggest that on average the ethical funds (0.21%) out-perform their conventional peers (0.16%) and relevant indexes. Also, Islamic indexes are on average more risky (SD = 2.61%) than conventional indexes (SD = 2.47%).

Financial performance measures.
In order to improve analysis, this study applied different performance measures and shows the main results for all samples of Dow Jones indexes under consideration in Table 5. While panel A presents results associates with performance measures SR, TR, MM and eSDAR of each region, panel B reports the results of the same performance measures, but for sector groupings.  Starting first with the Sharpe (1966) ratio SR, Table 5 shows that Sharpe Ratios are considerably negative and the average result is -1.21% when compared with non-Islamic indexes that are -1.30%. DJIHC (-1.75%) and DJICG (-1.71 %) are the lowest Sharpe ratios, while DJIC (-0.75%) and DJIT (-0.79%) show the highest ratios indicating best performance among Islamic index series. Overall, and in agreement with Girard and Hassan (2008), the Sharpe ratio indicates that faith-based ethical (Islamic) indexes slightly outperform their conventional peers.
The same pattern of results emerges with other ratios included in Table 5. The comparison of average Treynor Ratio shows that Dow Jones Islamic indexes out-perform their non-Islamic counterparts by -0.071% and -0.074% respectively. The MM ratio, on average, shows that Islamic indexes outperformed -12.78% their non-Islamic counterparts -14.40%, which is consistent with previous studies on SR. For eSDAR, the findings confirm the results associated with the Sharpe and MM Ratios, where IDJIs (Dow Jones Islamic indexes) outperform (-0.027%) their non-Islamic peers (-0.03%). These findings are in line with those of Statman (2000). The results show that in most cases the Islamic indexes seem to exhibit higher risk-adjusted performance than their conventional counterpart, irrespective of the alternative measures. This finding confirms those of Ashraf and Mohammad (2014) and Jawadiet al. (2014). Notably, the performance measures associated with sector groupings have a similar pattern of results, but with a clear indication that Islamic indexes and markets are sector oriented.
Next, the performance of Dow Jones indexes using Jensen's α as identified in equation 6. Table 6 shows that despite the results of non-Islamic indexes being only insignificantly different from the Islamic peers, it is worth noting the following few differences: (i) 2 R shows that non faith-based ethical indexes have on average higher 2 R than faith-based ethical indexes due to the greater diversification opportunities of non-Islamic indexes, (ii) Jensen's αassociated with faith-based ethical indexes show better results over their non-ethical counterparts, which are in line with the Girard and Hassan (2008).
For the systematic risk β, the Islamic indexes exhibit average systematic risks of 0.53%, indicating a lower level of risk associated with Islamic indexes compared to their counterparts. This is justified, as Islamic investors focus on stocks that are complaint with Shariah law (growth stocks rather than dividend income).
Overall, two conclusions can be drawn. First, there is no clear significant difference in performance between faith-based ethical and conventional indexes. Second, ethical indexes are less market sensitive than conventional indexes. To conclude, the previous analysis does not provide any clear conclusions regarding financial performance for conventional and Islamic indexes, which may be due to the appropriateness of standard performance, notably when working with Islamic investments.  In terms of the timing ability (Treynor and Mazuy, 1966) of Dow Jones indexes, Table 7 shows that Islamic indexes outperform conventional indexes (alpha = 0.0009, 5% significance level). For the TM model, αis positive and different from the findings reported in Table 6. For β, the findings show that indicates a negative relationship with the market portfolios (-0.9912) and is statistically significant at the 5% level of significance. While the positive and significant γ implies the ability of indexes to predict the performance when the market is up, the negative γindicates that the timing ability alters the performance in a different way. For the 2 R , it is about 98.4 % for the Islamic index, which is slightly higher than the conventional peers (98.3 %). The higher values of 2 R reflect the strong explanatory power of the TM model compared to the CAPM model. Overall, the findings of TM model imply that the stocks in Dow Jones indexes are unable to time the market, as those types of stocks provide a negative market timing ability for both Dow Jones Islamic and their peers. The same pattern of results applies to the conventional peers and sector groupings.  In an attempt to ascertain the adequacy of the single-factor asset pricing model to explain investment performance, the performance of Islamic and conventional Dow Jones indexes was tested using Carhart (1997) model and the Habit Formation model of Campbell and Cochrane (2000). While Table 8 summarizes the estimates of the multi-factor model of Carhart (1997), Table 9 presents the GMM estimates of the habit formation model. In general, these findings suggest that: (1) the predictability power associated with the multifactor models 2 R is higher than those associated with the single-factor model. This confirms this paper's expectation that multi-factor models are superior in explaining investment returns; (2) ethical investments/indexes are less exposed to the market portfolio than their conventional peers; (3) for ethical indexes, the world, EU, ASP, DEV and emerging markets are heavily exposed to small caps while US and UK indexes are relatively more invested in large caps. Similarly, Technology and Consumption Service sector ethical indexes are heavily exposed to small caps, while Oil and Gas, Health care and Consumption Goods investments are comparatively more invested in large caps. Interestingly, and in line with Guerard (1997), the negative relationship with the HML factor indicates that Islamic indexes are more growth-oriented, while non-ethical indexes, which have a positive relationship with the HML factor are more valueoriented; (4) the inclusion of market risk, size, bookto-market and momentum (Carhart factors) has no clear impact on the difference in performance between faith-based ethical and conventional indexes. Overall, the previous analysis does not provide clear conclusions regarding financial performance for either conventional or Islamic indexes. For the habit formation model, Table 9 presents the GMM estimates for the Fourteen Islamic and conventional Dow Jones indexes, (World, EU, ASP, DEV, EMRG, US, UK) and Ten sector groupings (OG, TEC, TEL, CG, CS) using stock returns and consumption for each index. The findings show that the model is not rejected on the basis of Hansen's test of over-identifying restrictions and can be estimated to be both of theoretically plausible value and also significant.
These results are broadly comparable to the findings of Hyde and Sherif (2005) who found supportive evidence for the Habit Formation specifications with U.K. stock returns. For the performance of Islamic and conventional indexes, the study detailed in this current paper also found that in most cases the Islamic indexes either outperformed the non-Islamic indexes or that there is only an insignificant difference in performance between both indexes. Note: INST = GMM estimation performed with instrument, (1, Ct/C t-1 , R m>t-1 , C t-1 /C t-2 , R m , t-2. χ 2 is Hansen's test of overidentifying restrictions. Asymptotic standard errors are in parentheses and asymptotic p-values are in brackets. In order to improve the analysis and as a robustness test, this study went further by testing the same previous investment performance measures, but did so during the financial crises. Table 10 presents a comparison between the performance of Islamic Dow Jones indexes and conventional peers during the financial crisis periodusing the same set of performance measures as before (SR, TR, MM and eSDAR). It is clear from Table 10    With regard to Jensen's α,  In terms of the investigation of the faith-based ethical DJ indexes using TM model during the crisis period. Table 12 shows a similar pattern to the previous results: faith-based ethical indexes slightly outperformed conventional indexes during the financial crisis period. On average there were performances of α = 0.26%, β = 5.23%, γ = -74.05%, 2 R = 97.65% for conventional investments, while the faith-based ethical indexes on average showed performance of α = 0.36%, β = 7.47%, γ = -71.35%, 2 R = 97.82%. For sector groupings, the performance is dominated by Technology and Consumption Services sector groupings, as they still performed well in comparison to the rest of the sector groupings. This again implies that the faithbased ethical indexes are sector oriented. With regard to investigating whether faith-based ethical investments have superior performance over their unscreened benchmarks, and whether sector-specific investment performance differs from the economy level using the multi-factor tests during the finical crises, the same pattern of results emerges. Again, SMB (small minus big) is the difference between returns on a small capitalisation portfolio and a large-capitalisation portfolio; HML (high minus low) is the difference between returns on a portfolio with a high book-to-market ratio and a portfolio with a low book-to-market ratio. MoM is the average return on the past winner stocks minus the average return on the past loser stocks. The findings show some improvements in 2 R for both the fourfactor model (0.80) and the single-factor CAPM model (0.50) when using conventional indexes. For the Islamic indexes, the same pattern holds, as the adjusted 2 R stands at 0.78, and 0.57 for the 4-factor and the CAPM models respectively. In general, this implies that the four-factor model (Carhart, 1997) is more effective in explaining index performance, suggesting strong time-variation in betas 8 . Overall, the above results show cautious support for the null hypothesis H 0 of this study: that, there is no clear evidence that ethically screened investments underperform conventional investments. Whilst the financial performance of faith-based ethical (Islamic) and conventional investments is relatively close in the first sample (pre-crisis), faith-based ethical (Islamic) indexes appear to outperform their peers during the financial crisis. However, the results are strongly dependent on the performance measure and the period and region under consideration.

Conclusion
Using seven Dow Jones faith-based ethical (Islamic) indexes to their seven conventional counterparts, a range of investment performance measures and most recent asset pricing models, this paper aimed to provide new evidence regarding whether faith-based ethical investments have superior performance over their unscreened benchmarks, and whether sectorspecific investment performance differs from the aggregate market level. This study covers the period 1999-2013, which made it possible to capture the impact of the recent global financial crisis on the performance of investments.
From the analysis of this paper, a number of interesting results can be drawn. Firstly, there is no convincing evidence supporting the performance differences between faith-based ethical (Islamic) and conventional indexes, in particular after the inclusion of the common factors such as size, book-to-market, momentum, and the habit formation specifications of Muslim and non Muslim investors. Secondly, ethical investments have distinct investment styles compared to conventional investments. For example, ethical investments are typically less exposed to market return variability compared to conventional investments. In addition, the world, EU, ASP, DEV and emerging markets ethical indexes are heavily exposed to small caps while the US and UK indexes, on the other hand, are comparatively more invested in large caps. Similarly, Technology and Consumption Service sector groupings for the ethical indexes are heavily exposed to small caps, whereas Oil and Gas, Health Care and Consumption Goods indexes are relatively more invested in large caps. Thirdly, overall, similar rewards for risk and diversification benefits exist for both indexes. When controlling performance for style and time variability, the findings notably show that Islamic indexes are growth-based, whereas conventional indexes are value-based. Finally, when investigating the performance of ethical indexes relative to conventional indexes overtime time, in particular during the recent financial crisis, the Islamic indexes outperform the conventional ones. Interestingly, the findings shown here indicate that ethical indexes perform better than standard conventional indexes and this explains ethical investments returns. The conclusions with respect to indexes performance when using sectorial indexes were, however, unaffected. This clearly indicates that Islamic and markets are sector oriented. For the performance of habit formation model with faithbased ethical Islamic and conventional indexes, this study found that the 'Habit Formation' specification of Campbell and Cochrane (2000) appears to offer a more accurate and feasible explanation and works well as a measure of investment performance for both ethical and conventional indexes.
The findings of this study offer insights for international investors, institutions and policy makers interested in the performance of faith-based ethical Islamic investments in both the emerging and developed market setting. While investors can reconcile faith with finance, the financial market authorities could revise their regulations and legislation to enable banks and markets to include these types of products, and to propose new products with similar characteristics. Policy makers may consider the feasibility of establishing an integrated Islamic financial market by forming an economic union among Muslim countries with similar economic characteristics. While this study helps rectify some of the gaps in the existing literature into faith-based ethical investments in general, and in emerging markets in particular, it highlights a number of others for further research. The most possible immediate expansion would be to include certain ethical indexes (Christian and Jewish) omitted in this study due to the unavailability of the data. Such indexes may also be influential in explaining the performance of faithbased ethical investments. In addition, future research should also look into Islamic sub-indexes and include other institutions that are Islamic subsidiaries of conventional banks. Finally, to more effectively explain the reason why Islamic indexes and markets are sector oriented will be a key subject for future research.