Islamic investing

Islamic investing

Review of Financial Economics 21 (2012) 53–62 Contents lists available at SciVerse ScienceDirect Review of Financial Economics journal homepage: www...

437KB Sizes 0 Downloads 33 Views

Review of Financial Economics 21 (2012) 53–62

Contents lists available at SciVerse ScienceDirect

Review of Financial Economics journal homepage: www.elsevier.com/locate/rfe

Islamic investing Christian Walkshäusl ⁎, Sebastian Lobe 1 University of Regensburg, Center of Finance, Universitätsstraße 31, 93053 Regensburg, Germany

a r t i c l e

i n f o

Article history: Received 19 April 2011 Received in revised form 25 January 2012 Accepted 27 February 2012 Available online 8 March 2012 JEL classification: G11 G12 G15 Keywords: Islamic finance Sharia-compliant investments Asset pricing Performance evaluation International markets

a b s t r a c t Using a large international sample of 35 developed and emerging markets, we analyze whether Islamic indices exhibit a different performance to conventional benchmarks. While there is no compelling evidence of performance differences in robust Sharpe ratio tests and after controlling for market risk, we find a significantly positive four-factor alpha for the aggregate developed markets region. This outperformance stems, however, mainly from the U.S. and is largely attributable to the exclusion of financial stocks in Shariascreened portfolios. As the extensive downturn of financials is related to the recent financial crisis, we do not argue that this outperformance will continue over time. The style analysis reveals that Islamic indices invest mainly in growth stocks and positive momentum stocks. This, for a passive portfolio intriguing result can, however, be explained by the strong sector allocation towards energy firms and their strong momentum characteristic during the sample period. © 2012 Elsevier Inc. All rights reserved.

1. Introduction One of the major innovations in the financial community is the rapid growth of Islamic financial services around the world. Today, Sharia-compliant assets amount to $939 billion worldwide. While there are more than 600 Islamic funds available, investors have begun shifting their assets from actively managed mutual funds to passive index-based investments. 2 For investors wanting to allocate capital in accordance with their religious beliefs, it is therefore of interest to know whether a passive portfolio of equities selected by Islamic screening procedures exhibits a different performance than a conventional market portfolio. As there are concerns that a screened portfolio is likely to underperform a conventional (unscreened) portfolio (e.g., Rudd (1981), Grossman and Sharpe (1986)), we pose the questions: Does an Islamic investor have to sacrifice financial performance by investing religiously? Should that be the case, how much is the performance difference? And how does a Sharia-compliant screening process affect the return behavior with regard to common risk factors?

⁎ Corresponding author. Tel.: + 49 941 943 2729. E-mail addresses: [email protected] (C. Walkshäusl), [email protected] (S. Lobe). 1 Tel.: + 49 941 943 2727. 2 See, Shaping a New Tomorrow: Global Wealth 2011, a report by The Boston Consulting Group, May 2011. 1058-3300/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.rfe.2012.03.002

To answer these questions, we investigate in this paper the financial performance and investment behavior of Islamic indices in comparison to conventional market benchmarks in a large international sample encompassing 35 developed and emerging markets around the world using a contemporary measurement framework based on bootstrap simulations and multi-factor models. Islamic investments must act in accordance with the principles of the Sharia, the Islamic law, governing all aspects of a Muslim's life. Islamic investments are insofar often referred to as Shariacompliant investments to underscore this circumstance. One of the most distinctive features of Islamic banking and finance is the fact that the payment and receipt of interest is not permitted (Riba). The Sharia encourages instead the use of profit-sharing and partnership schemes. Adherence to these principles is generally conducted by a Sharia supervisory board. A panel of Islamic scholars approves proposed companies and monitors the compliance of their business activities with the guidelines of the Sharia. The following typical screening criteria for Sharia-compliant investments can be summarized: Inherently, investments in preferred stocks and bonds are unacceptable under the Islamic law, since both promise a fixed rate of return and grant no voting rights (Naughton and Naughton (2000)). Further, companies whose core business involves alcohol, conventional financial services, entertainment, pork-related products, tobacco, or weapons are excluded. In addition, restrictions are applied based on certain financial ratios, meaning companies with unacceptable levels of debt or impure interest income are excluded from the

54

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

universe of investable stocks. Typically, companies with debt of more than one third of market capitalization are excluded (Hussein and Omran (2005)). Despite the growing interest in Islamic finance, there are few empirical studies exploring the performance of Islamic investment vehicles. Elfakhani and Hassan (2005) analyze the performance of 46 international Islamic mutual funds from 1997 through 2002 in comparison to a corresponding Islamic market index and a conventional market index. Based on the traditional measures of Sharpe, Treynor, and Jensen, their results suggest that the performance of Shariacompliant funds and the chosen benchmarks is very similar. Using likewise a traditional performance evaluation framework, Hussein and Omran (2005) examine the performance of the Dow Jones Islamic Market Index (DJIMI) and its 13 sub-indexes (based on size and industry) against their conventional counterparts over the period 1996 to 2003. The authors find that Islamic indexes outperform in bull markets, but underperform in bear markets. Employing Sharpe's (1992) Style Analysis, Forte and Miglietta (2007) examine the asset allocation of the Sharia-compliant FTSE Islamic Europe index as a faith-based investment vehicle in comparison to conventional socially responsible investing (SRI) equity indexes. Finding distinctive differences in style, sector, and country exposure, they conclude that faith-based investments and SRI investments should be addressed as separate investment approaches. Albaity and Ahmad (2008) investigate the performance of the Kuala Lumpur Syariah Index (KLSI) and the Kuala Lumpur Composite Index (KLCI) from 1999 through 2005. Their results provide no evidence of significant statistical differences in risk-adjusted returns between the Sharia-compliant index and the conventional stock market index. Hoepner, Hussain, and Rezec (forthcoming) analyze the financial performance and investment style of 262 Islamic equity funds from twenty countries. The analyzed mutual funds underperform in eight countries, but outperform in three countries. Islamic funds tend to mimic the return behavior of small firms. Our findings are summarized as follows. While there is no compelling evidence of performance differences between Islamic indices and conventional benchmarks in robust Sharpe ratio tests and after controlling for market risk, we find a significantly positive four-factor alpha for the aggregate developed markets region. This outperformance stems, however, mainly from the U.S. and is largely attributable to the exclusion of financial stocks in Sharia-screened portfolios. As the extensive downturn of financials is related to the recent financial crisis, we do not argue that this outperformance will continue over time. The style analysis reveals that Islamic indices invest mainly in growth stocks and positive momentum stocks. This, for a passive portfolio intriguing result can, however, be explained by the strong sector allocation towards energy firms and their strong momentum characteristic during the sample period. The remainder of the paper is organized as follows. In Section 2, we describe the international data and present summary statistics. Section 3 outlines the methods and results for performance testing based on the Sharpe ratio. Section 4 examines whether the financial performance of Islamic indices is significantly different from conventional benchmarks using time-series factor regression tests. Section 5 contains additional analyses for robustness concerns, and Section 6 concludes. 2. Data and summary statistics The data for our study is compiled from four different sources. The return data for the Islamic indices and conventional benchmarks are obtained from Morgan Stanley Capital International (MSCI). In general, we study monthly total returns (so-called gross returns with reinvested dividends), except for the Value at Risk (VaR) measure, where we employ daily total returns for its computation. There are in total 44 Islamic country indices (23 from developed markets and 21 from

emerging markets) available from MSCI. However, we have to exclude nine markets from our analysis because of data issues. Hungary, Ireland, and Portugal due to non-continuous time-series return data from MSCI. Colombia, Czech Republic, Egypt, Morocco, Poland, and Russia on account of low firm coverage in Datastream/Worldscope disallowing us to form adequate explanatory factors (i.e., SMB, HML, and WML) for these markets over the complete sample period. Thus, our final data set leaves use with 35 investigable single markets (21 from developed markets and 14 from emerging markets). In addition, we include in our analysis the aggregate (multi-country) indices of the developed and emerging markets regions as highly diversified portfolios. The explanatory factors for the United States are from Kenneth French's data library. 3 As Griffin (2002) and Fama and French (2011) show that capital markets do not seem to be integrated suggesting that size, value, and momentum factors are country-specific, we form respective factor portfolios for the other markets in our sample. For the construction of non-U.S. explanatory factors, we use total return data on common stocks from Datastream to form respective zero-investment, factor-mimicking portfolios for size, book-tomarket equity, and momentum in each market. All accounting data (e.g., the book value of common equity) for the construction of nonU.S. explanatory factors is obtained from Worldscope. Firms must have a positive book value to be included in the sample and as common in the asset pricing literature, we exclude financial firms with Standard Industrial Classification (SIC) codes between 6000 and 6999 in the factor portfolios. To ensure the quality of our data, we apply the screening procedures proposed by Ince and Porter (2006). All data is denominated in U.S. dollars and the risk-free rate is calculated using the one-month U.S. Treasury bill rate. The sample period is June 2002 to June 2011 (109 months). Table 1 provides summary statistics for our set of Islamic indices and their conventional benchmarks. Panel A and B describe the single markets and Panel C the aggregate markets. The average annualized Islamic index returns in developed market countries (Panel A) range from 4.2% to 22.4%, while the conventional benchmarks produce average returns from 4.5% to 19.8% per year. The average index returns in emerging markets (Panel B) are considerably higher compared to developed markets. They vary for the Islamic indices between 11.3% and 32.5% per year and from 10.5% to 32.3% per year for the conventional benchmarks. Thus, indicating a largely similar return spectrum between the two index variants in the emerging markets countries over the sample period. This is confirmed by the statistics of the aggregate (multi-country) indices in Panel C. While the Islamic index outperforms the conventional benchmark by 1.5% per year in developed markets, the average annualized index return is not distinguishable from its conventional benchmark in emerging markets. Fig. 1 illustrates the growth of the index values of the Islamic indices in comparison to the conventional benchmarks over the sample period along with the year-by-year returns for the aggregate developed (Panel A) and emerging (Panel B) markets regions. While the growth of the two index variants is largely similar in emerging markets, the developed markets Islamic index outpaces the conventional benchmark as of 2007 with the beginning of the recent financial crisis and the extensive downturn of bank and financial services stocks. We will discuss this issue in more detail in the robustness section. The annualized standard deviations of monthly returns in Table 1 tend to be smaller for the Islamic indices of developed markets, while they seem to be higher in emerging markets respectively relative to the conventional benchmarks. However, it is obvious that the higher average returns of emerging markets indices are accompanied by higher levels of volatility. 3 The data library is accessible through: http://mba.tuck.dartmouth.edu/pages/ faculty/ken.french/.

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

55

Table 1 Summary statistics. Islamic Market

Abbr.

Mean

Conventional Std

Skew

Kurt

JB

VaR

Mean

Std

Skew

Kurt

JB

VaR

Panel A: developed markets countries Australia AUS 22.4 Austria AUT 19.0 Belgium BEL 13.4 Canada CAN 18.9 Denmark DEN 20.8 Finland FIN 12.7 France FRA 10.6 Germany GER 16.3 Greece GRE 9.0 Hong Kong HKG 12.1 Italy ITA 13.1 Japan JPN 4.2 Netherlands NED 9.7 New Zealand NZL 16.0 Norway NOR 21.5 Singapore SIN 16.6 Spain SPA 16.3 Sweden SWE 19.7 Switzerland SWI 11.9 United Kingdom UK 10.6 United States USA 7.1

25.3 32.7 20.3 25.5 23.1 31.4 21.3 26.1 29.3 19.0 21.5 17.0 26.9 25.3 30.5 21.8 23.8 28.2 16.0 18.1 14.7

− 0.9 − 0.8 − 0.3 − 0.9 − 1.1 − 0.1 − 0.7 − 0.8 − 0.8 − 0.6 − 0.3 − 0.4 − 0.5 − 0.4 − 0.8 − 1.3 − 0.6 − 0.2 − 0.7 − 0.4 − 0.9

6.0 6.3 4.2 6.0 5.3 4.1 4.0 5.1 5.6 4.8 3.4 3.7 4.1 4.8 4.8 9.4 5.3 5.1 4.1 4.1 4.5

57.3 62.4 8.7 55.0 43.4 5.3 13.8 30.5 41.8 22.0 2.7 5.6 10.0 17.5 26.1 217.0 30.2 21.5 14.1 8.7 24.2

− 5.4 − 5.7 − 4.2 − 5.3 − 4.1 − 8.4 − 4.7 − 4.5 − 6.5 − 3.4 − 4.9 − 3.6 − 4.2 − 4.4 − 6.2 − 4.3 − 5.4 − 4.9 − 3.0 − 4.5 − 2.9

17.7 15.4 7.8 16.2 16.8 10.3 10.2 13.0 5.3 13.1 8.2 4.5 8.5 13.5 19.8 16.2 14.8 18.1 10.3 7.9 5.8

23.0 30.4 26.6 22.0 22.4 30.5 22.9 26.4 33.8 21.4 24.5 17.1 24.1 21.7 31.4 22.7 26.4 27.3 17.1 18.2 15.9

− 0.8 − 1.0 − 1.5 − 0.8 − 1.1 − 0.1 − 0.6 − 0.6 − 0.3 − 0.2 − 0.5 − 0.2 − 0.9 − 0.8 − 0.9 − 0.7 − 0.5 − 0.3 − 0.5 − 0.5 − 0.8

5.4 7.5 8.5 6.2 6.2 3.9 4.2 4.7 5.2 5.0 4.0 3.6 4.8 4.7 5.4 7.9 4.5 5.0 3.6 4.8 4.6

37.4 111.4 178.6 57.8 68.4 3.9 13.4 19.0 23.1 18.4 9.3 2.1 30.1 23.5 41.2 118.8 15.9 19.7 6.1 18.7 23.8

− 5.0 − 6.1 − 4.4 − 4.5 − 4.5 − 6.8 − 4.9 − 4.5 − 6.4 − 3.3 − 4.7 − 3.9 − 4.2 − 4.6 − 5.9 − 3.9 − 5.6 − 4.9 − 3.1 − 4.5 − 3.1

Panel B: emerging markets countries Brazil BRA 32.5 Chile CHI 24.4 China CHN 22.5 India IND 21.9 Indonesia INO 30.2 Korea KOR 19.2 Malaysia MAL 17.1 Mexico MEX 21.1 Peru PER 23.8 Philippines PHI 25.6 South Africa SAF 19.5 Taiwan TAI 11.3 Thailand THA 25.6 Turkey TUR 29.1

41.2 23.1 27.8 31.3 36.9 29.1 19.4 26.5 40.5 34.8 28.8 26.1 30.5 47.9

− 0.5 − 0.8 − 0.4 − 0.2 − 0.5 0.0 − 0.4 − 0.5 − 0.2 0.1 − 0.5 − 0.2 − 0.3 0.5

4.5 7.3 3.7 4.7 5.1 3.6 5.4 5.0 3.2 3.2 4.1 2.8 5.6 6.3

13.8 97.1 4.8 14.4 25.3 1.5 28.9 23.9 1.2 0.5 11.4 1.2 32.5 53.3

− 6.3 − 3.1 − 5.0 − 5.1 − 5.0 − 5.4 − 2.5 − 5.4 − 5.4 − 7.4 − 5.2 − 3.5 − 5.1 − 6.3

31.8 24.0 21.7 26.4 32.3 18.5 15.6 19.3 31.7 17.9 20.9 10.5 23.0 29.7

37.7 21.9 27.9 31.8 34.6 30.3 17.6 24.1 33.6 25.4 27.8 26.0 28.1 46.1

− 0.6 − 0.7 − 0.5 − 0.1 − 0.5 − 0.2 − 0.3 − 1.0 − 0.5 − 0.3 − 0.6 − 0.1 − 0.2 − 0.1

4.4 6.2 3.8 4.9 5.5 3.5 4.5 6.4 4.5 3.9 3.5 2.8 6.2 3.5

14.4 53.9 6.5 16.8 33.5 1.8 12.9 70.5 14.7 4.7 6.7 0.6 47.9 1.4

− 5.8 − 3.0 − 4.6 − 5.3 − 5.1 − 5.8 − 2.9 − 4.7 − 5.4 − 4.0 − 4.5 − 3.9 − 5.0 − 5.6

Panel C: aggregate markets Developed DM Emerging EM

15.9 24.6

− 0.9 − 0.8

4.9 4.9

32.2 27.7

− 2.8 − 4.0

7.4 18.9

16.9 24.1

− 0.9 − 0.8

5.0 5.1

33.7 30.9

− 2.9 − 3.8

8.9 18.9

Notes: This table presents summary statistics including the mean (annualized), standard deviation (annualized), skewness, kurtosis, the Jarque–Bera (JB) statistic, and the median Value at Risk (VaR) measure for the Islamic indices and the conventional benchmarks over the sample period June 2002 to June 2011 (109 months). The JB statistic tests the null hypothesis that the index returns follow a normal distribution. For JB values equal or smaller than 9.2, 6.0, or 4.6 the null hypothesis cannot be rejected at the 10%, 5%, or 1% level of significance. The VaR is computed using rolling window estimation technique. The VaR is defined as the lowest daily index return observed during the past 100 trading days (corresponding to a 1% VaR). The allocation of countries to developed and emerging markets follows the MSCI Market Classification.

Some of the Islamic indices seem to have not only higher average returns than their conventional benchmarks, but they also exhibit lower standard deviations (e.g., Belgium among developed markets and Korea among emerging markets). This implies not only a meanvariance dominance but possibly a first order stochastic dominance. While the mean-variance criterion takes only the first two moments of the return distribution into account, the stochastic dominance considers the entire probability distribution of returns and is therefore less restrictive. Portfolio choice decisions are typical examples for the application of stochastic dominance rules (Ogryczak and Ruszczynski (1999)). In line with Levy (1992), let F(r) and G(r) be the cumulative return distributions of the Islamic index and the conventional benchmark in the market, then the Islamic index is said to dominate the conventional benchmark by first order stochastic dominance, if and only if F(r) ≤ G(r) for all return observations r with at least one strict inequality. Testing for first order stochastic dominance in each market, we find that no index variant exhibits first order stochastic dominance over the other in any market.

The skewness measure is negative in almost all markets and the kurtosis is in general greater than three. Hence, the Jarque–Bera test statistic rejects the hypothesis that the index returns follow a normal distribution for the vast majority of single markets and all aggregate markets. In line with Bali, Demirtas, and Levy (2009), the VaR measure is computed using rolling window estimation technique. The VaR is defined as the lowest daily index return observed during the past 100 trading days (corresponding to a 1% VaR). The median values of the lowest daily returns obtained from the last 100 days range in developed markets from − 8.4% to −2.9% for the Islamic indices and from −6.8% to −3.1% for the conventional benchmarks. For emerging markets, the corresponding median values are in the range of −7.4% to −2.5% and − 5.8% to −2.9%, respectively. The differences in the VaR measures between the two index variants are mainly in the range of 50 basis points in absolute magnitude for each market. The largest absolute differences stem from the Philippines (3.4%) and Finland (1.6%), where the Islamic indices exhibit a significantly higher

56

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

a

Growth of Index Value

Growth of Index Value

b

Year-by-Year Returns

Year-by-Year Returns

Fig. 1. Performance of Islamic indices and conventional benchmarks. Panel A: aggregate developed markets. Panel B: aggregate emerging markets. Notes: This figure shows the growth of the index values of the Islamic indices in comparison to the conventional benchmarks over the sample period along with the year-by-year returns for the aggregate developed (Panel A) and emerging (Panel B) markets regions. The solid green line illustrates the index value of the Islamic index and the dotted black line represents the conventional benchmark. The green bar illustrates the annual return of the Islamic index and the white bar represents the conventional benchmark.

downside risk than their conventional benchmarks. In sum, however, the Islamic indices do not seem to be associated with a higher downside risk relative to the conventional benchmarks. However, as in Bali, Demirtas, and Levy (2008) and Bali et al. (2009) we observe at the country level a positive relation between the absolute downside risk and the index returns. Thus, a higher average index return tends to be associated with a higher absolute (more negative) VaR measure.

3. Sharpe ratio tests We begin our analysis of the risk-adjusted performance of Islamic indices in comparison to conventional benchmarks by conducting differences-in-Sharpe ratio tests. The Sharpe ratio for each index is calculated as the index's excess return over the one-month U.S. Treasury bill rate divided by its standard deviation. The Sharpe ratio difference (ΔSR) in each market is given by SR(ISL)–SR(MKT), referring to the measures of the Islamic index and the conventional market benchmark, and we test the null hypothesis of equal Sharpe ratios of the form H0: ΔSR = 0. To obtain respective test statistics, we use the classic method of Jobson and Korkie (1981) and the bootstrap method of Ledoit and Wolf (2008) to compare the performance measures. The JK test statistic is given by



μ i σ m −μ m σ i pffiffiffi θ

ð1Þ

with

θ¼

  1 1 2 2 1 2 2 μμ 2 2 2 2σ i σ m −2σ i σ m σ im þ μ i σ m þ μ m σ i − i m σ im ; T 2 2 2σ i σ m

ð2Þ

where μ is the excess return and σ is the standard deviation. The subscript i denotes the Islamic index and m denotes the market index (i.e., the conventional benchmark). T is the number of monthly observations and σim is the covariance of the Islamic and market index returns. z follows asymptotically a normal distribution. Eq. (2) uses the Memmel (2003) correction to revise a typographical error in the original proposal. Although the JK test seems to be the standard approach for performance comparison, this test is not robust against tails heavier than the normal distribution and time-series characteristics. Since both effects are quite common with financial return data and are obviously the case in our data set (see Table 1), we apply additionally the bootstrap method of Ledoit and Wolf (2008) for robust performance testing. We construct a studentized time-series bootstrap confidence interval with nominal level 1-c for the Sharpe ratio difference (ΔSR). If this interval does not contain zero, then the null hypothesis of equality is rejected at the nominal level c and we declare the two performance measures different. We employ the circular block bootstrap of Politis and Romano (1992). The bootstrap procedure uses a data-dependent choice of block size based on the calibration function of Loh (1987). The significance levels considered are 1%, 5%, and 10%. All bootstrap p values are computed employing 5000 resamples. Due

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

to the bootstrap procedure, closed-form expressions can obviously not be formulated for this test. 4 Table 2 presents the annualized Sharpe ratios for the Islamic indices and the conventional benchmarks over the sample period along with the Sharpe ratio differences (ΔSR) and the p values for the null hypothesis of equal Sharpe ratios according to the two test methods. Except for Japan, the Sharpe ratio differences are positive in developed markets and on average close to zero in emerging markets. The highest positive Sharpe ratio difference stems from Belgium with an annualized value of 0.35, while the highest negative difference comes from Peru with −0.35 per year. According to the Jobson and Korkie (1981) method, we detect six significantly positive Sharpe ratio differences for the countries of Belgium, Germany, Italy, United Kingdom, United States, and the aggregate developed markets region. Thus, Islamic indices seem to outperform their conventional benchmarks on a risk-adjusted basis in these markets and underperform in the markets of India and Peru which exhibit significantly negative Sharpe ratio differences. 5 However, as Jobson and Korkie (1981) acknowledge in their paper, the power of their test in detecting performance differences is weak. Thus, the chance is high that the statistical inference is flawed, leading for instance to false rejections of the null hypothesis though the performance is similar. This seems to be apparently the case when we move on to the results of the more powerful Ledoit and Wolf (2008) approach. We find no significantly different Sharpe ratios in any market. Italy is the sole country, where the difference is at least significant at the 10% level. In sum, the results lead to no rejection of the null hypothesis of equal Sharpe ratios. In both market segments, Islamic indices do not exhibit a significantly different risk-adjusted performance compared to their respective conventional benchmarks. This is valid at the country as well as the aggregate levels. 4. Time-series factor regression tests To provide further insights into the performance and investment style of Islamic indices, we employ the Capital Asset Pricing Model (CAPM) of Sharpe (1964) and Lintner (1965) and the four-factor framework following Carhart (1997). While the CAPM controls solely for market risk (through beta), the four-factor model provides additional controls for common return patterns associated with size, book-to-market equity (value), and momentum (Banz (1981), Rosenberg, Reid, and Lanstein (1985), and Jegadeesh and Titman (1993)). 4.1. Explanatory factors and regression setup The construction of the size (SMB), value (HML), and momentum (WML) factors for non-U.S. markets follows Fama and French (1993) and Carhart (1997). SMB and HML are created from a two-by-three sort on market equity (as of the portfolio formation date) and bookto-market equity (B/M, for the fiscal year ending in the previous calendar year) forming six value-weighted portfolios. The portfolios are formed in general each June and monthly value-weighted returns are calculated from July of year t to June of year t + 1. However, to comply with national conventions, two exceptions are made to this rule. First, the size and value factors of Japan and India are formed in September due to the fact that the fiscal year end is March for the vast majority of firms in these two markets. Second, the explanatory factors of Australia, New Zealand, and South Africa are formed in December on account of June as the fiscal year end. As in Fama and French (1993), the six-month gap ensures that the accounting data are publicly 4 For readers not familiar with the bootstrap method, we have to refer for further information to the work by Ledoit and Wolf (2008) to save space. 5 At the 10% level, the number of significant Sharpe ratio differences amounts to 13 (with ten positive and three negative).

57

Table 2 Sharpe ratio tests. ΔSR

p (JK)

p (LW)

Panel A: developed markets countries AUS 0.81 0.69 AUT 0.52 0.44 BEL 0.57 0.22 CAN 0.67 0.65 DEN 0.82 0.66 FIN 0.35 0.28 FRA 0.41 0.36 GER 0.55 0.42 GRE 0.24 0.10 HKG 0.54 0.52 ITA 0.52 0.26 JPN 0.14 0.16 NED 0.29 0.28 NOR 0.64 0.57 NZL 0.56 0.54 SIN 0.68 0.63 SPA 0.61 0.49 SWE 0.63 0.60 SWI 0.63 0.49 UK 0.48 0.33 USA 0.36 0.25

0.12 0.08 0.35 0.02 0.16 0.07 0.05 0.14 0.14 0.01 0.26 − 0.02 0.01 0.07 0.02 0.05 0.12 0.03 0.14 0.15 0.11

0.06 0.15 0.02 0.21 0.06 0.17 0.14 0.02 0.12 0.22 0.02 0.21 0.23 0.09 0.22 0.16 0.13 0.18 0.08 0.04 0.04

0.25 0.55 0.15 0.85 0.32 0.75 0.55 0.11 0.58 0.86 0.08 0.85 0.93 0.32 0.90 0.72 0.54 0.75 0.38 0.16 0.16

Panel B: emerging markets countries BRA 0.74 0.79 CHI 0.97 1.00 CHN 0.74 0.71 IND 0.64 0.77 INO 0.77 0.88 KOR 0.60 0.55 MAL 0.79 0.78 MEX 0.72 0.72 PER 0.54 0.89 PHI 0.68 0.63 SAF 0.61 0.69 TAI 0.36 0.33 THA 0.78 0.75 TUR 0.57 0.60

− 0.05 − 0.03 0.03 − 0.13 − 0.11 0.05 0.01 0.00 − 0.35 0.05 − 0.07 0.03 0.03 − 0.03

0.12 0.20 0.15 0.03 0.07 0.15 0.23 0.25 0.03 0.19 0.13 0.18 0.22 0.21

0.53 0.84 0.64 0.18 0.33 0.62 0.95 0.98 0.16 0.77 0.54 0.76 0.84 0.84

Panel C: aggregate markets DM 0.44 EM 0.69

0.11 − 0.02

0.02 0.19

0.12 0.76

Market

SR (ISL)

SR (MKT)

0.33 0.71

Notes: This table reports the annualized Sharpe ratios for the Islamic indices (SR(ISL)) and the conventional market benchmarks (SR(MKT)) over the sample period. The Sharpe ratio for each index is calculated as the index excess return over the one-month U.S. Treasury bill rate divided by its standard deviation. ΔSR reports the difference of the Sharpe ratios between the Islamic index and the conventional benchmark for each market. The null hypothesis of equal Sharpe ratios is tested using the classic method of Jobson and Korkie (1981) and the bootstrap method of Ledoit and Wolf (2008). The p values for the null hypothesis according to the two methods are reported.

known before the factor portfolios are formed. The relevant breakpoint for the sort on market equity is median size (forming two size groups: small and big), while the B/M breakpoints are the 30th and 70th percentiles (forming three B/M groups: low, neutral, and high). SMB is the simple average of the returns on the three smallstock portfolios minus the simple average of the returns on the three big-stock portfolios, while HML is the simple average of the returns on the two high-B/M portfolios minus the simple average of the two low-B/M portfolios. WML is formed each month from a two-by-three sort on market equity (as of the previous month) and the cumulative prior twelvemonth return (skipping the most recent month (Jegadeesh (1990)) forming six value-weighted portfolios. The size sort uses median market equity and the sort on prior return is based on the 30th and 70th percentiles of the ranked values. WML is the simple average of the returns on the two high-prior return portfolios minus the simple average of the returns on the two low-prior return portfolios.

58

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

As in Griffin (2002), we form the monthly time-series of multicountry explanatory factors for the developed and emerging markets regions as weighted averages of the country factors using a country's portion in terms of total market equity (as of the previous month) as the weighting measure. Table 3 presents the average monthly premiums for the size (SMB), value (HML), and momentum (WML) factors in each market with the corresponding t-statistics. Most of the countries exhibit a negative size premium indicating that large-capitalization firms outperformed low-capitalization firms during the sample period. Notable exceptions are Canada and the United States which show high average SMB returns of 0.54% per month (t = 1.68) respectively 0.41% per month (t = 1.77). The average value premium based on HML is

Table 3 Factor premiums. Market

SMB Mean

HML t(Mean)

WML

Mean

t(Mean)

Mean

t(Mean)

Panel A: developed markets countries AUS − 0.04 − 0.07 AUT − 0.38 − 0.90 BEL − 0.55 − 1.98 CAN 0.54 1.68 DEN − 1.20 − 3.63 FIN 0.22 0.58 FRA 0.07 0.24 GER − 0.32 − 0.93 GRE − 0.91 − 2.03 HKG − 0.07 − 0.15 ITA − 0.54 − 2.04 JPN 0.07 0.25 NED 0.26 0.72 NZL − 0.25 − 0.78 NOR − 0.12 − 0.30 SIN − 0.60 − 1.53 SPA − 0.42 − 1.25 SWE − 0.27 − 0.65 SWI 0.04 0.13 UK − 0.65 − 1.85 USA 0.41 1.77

0.78 1.16 1.01 0.23 1.07 0.48 0.54 1.07 0.29 0.79 0.60 0.79 0.70 0.34 0.46 0.93 0.63 0.69 1.02 0.53 0.12

2.98 2.27 3.31 0.72 2.89 1.08 2.08 3.70 0.65 2.00 2.62 3.45 2.07 0.97 1.13 3.24 1.95 1.87 3.30 2.59 0.49

1.08 1.48 1.00 1.09 2.05 0.85 1.08 0.99 0.43 1.04 1.31 0.26 0.49 1.50 1.55 0.30 0.74 1.06 1.03 1.03 − 0.12

2.48 2.15 1.81 1.72 3.70 1.76 1.92 1.67 0.67 1.81 2.72 0.76 0.77 4.07 2.32 0.53 1.39 1.39 1.80 1.84 0.22

Panel B: emerging markets countries BRA 0.39 0.90 CHI − 0.10 − 0.31 CHN 0.44 1.04 IND 0.60 1.29 INO − 0.90 − 1.56 KOR − 0.91 − 2.14 MAL − 1.15 − 3.88 MEX 0.14 0.34 PER − 0.30 − 0.50 PHI − 0.49 − 0.88 TAI 0.17 0.53 THA − 0.37 − 0.85 TUR 0.35 0.70 SAF − 0.25 − 0.63

0.67 1.57 0.60 1.36 1.45 1.87 0.92 0.26 1.20 0.95 0.77 0.23 1.33 1.12

1.18 3.92 1.74 3.00 2.20 5.42 3.79 0.52 1.53 1.43 1.34 0.64 2.71 2.85

0.40 1.13 0.14 1.38 0.37 1.50 0.54 0.49 − 0.90 − 0.40 0.27 0.66 0.19 1.55

0.70 2.93 0.33 2.31 0.63 3.37 1.31 0.88 − 1.11 − 0.55 0.63 1.35 0.45 3.17

Panel C: aggregate markets DM 0.19 0.85 EM − 0.15 − 0.69

0.49 0.97

2.63 3.17

0.62 1.00

1.33 2.01

Notes: This table presents average monthly premiums on the size (SMB), value (HML), and momentum (WML) factors in each market and the t-statistics for the average monthly premiums over the sample period. The explanatory factors for the United States are from Kenneth French's data library. The construction of the non-U.S. zeroinvestment, factor-mimicking portfolios follows Fama and French (1993) and Carhart (1997) using a two-by-three sort on market equity and book-to-market equity (creating SMB and HML), and a two-by-three sort on market equity and the cumulative prior twelve-month return (creating WML). SMB, small minus big, mimics the common pattern in returns related to size. HML, high minus low, mimics the return behavior associated with book-to-market equity. WML, winner minus loser, mimics the return pattern related to short-term past returns.

positive in all considered markets and more than two standard errors from zero in 20 of the 35 countries. As in Fama and French (1998, 2011), we find that the value premium is considerably larger in markets outside the United States. Echoing previous results in the literature (e.g., Rouwenhorst (1998), Chui, Titman, and Wei (2010)), we find that the zero-investment mimicking factor portfolios related to momentum produce on average economically meaningful premiums, however, they tend to be weaker in terms of statistical significance during our sample period 2002 to 2011. In sum, we detect ten WML country returns that are more than two standard errors from zero. Regarding the multi-country factors for the aggregate developed markets region, we find on average a positive SMB return of 0.19% per month (t = 0.85) which is mainly due to the large U.S. size premium during our sample period, a HML return of 0.49% per month (t = 2.63), and a WML return of 0.62% per month (t = 1.33). For the aggregate emerging markets region, the size factor is negative with an average premium of −0.15% per month (t = −0.69), and the value and momentum factors exhibit average returns of 0.97% per month (t = 3.17) respectively 1.00% per month (t = 2.01). We measure the risk-adjusted performance of Islamic indices by running the following time-series factor regressions: Ri −Rf ¼ ai þ bi MKT þ ei ;

ð3Þ

Ri −Rf ¼ ai þ bi MKT þ si SMB þ hi HML þ wi WML þ ei :

ð4Þ

In these regressions, Ri is the return on Islamic index i, Rf is the risk-free rate (the one-month U.S. Treasury bill rate), MKT is the market excess return (the return on the conventional market benchmark in excess of the risk-free rate), SMB, HML, and WML are, respectively, the returns on the explanatory factors related to size, value, and momentum. ai, the alpha estimate, is the measure of out- or underperformance, and ei is the regression residual. Regression (3) refers to the CAPM, while regression (4) describes the four-factor model. 4.2. Regression results Table 4 summarizes the regression estimates for the Islamic indices over the sample period 2002 to 2011 using the CAPM and the four-factor model. The magnitude of the CAPM alphas is on average positive for developed markets and close to zero for emerging markets. However, only the Belgium Islamic index produces an alpha estimate that is significantly different from zero (0.64% per month, t = 2.15). The beta slopes as the systematic risk measures indicate a similar level of risk as the conventional benchmarks with an average value close to one. In sum, the CAPM results are similar to the outcomes of the bootstrap Sharpe ratio tests and thus provide no compelling evidence of performance differences between Islamic indices and conventional benchmarks at the country and aggregate levels. However, moving from the one-factor to the four-factor framework reveals a somewhat different picture. The additional controls for common return patterns related to size, value, and momentum tend to strengthen the alpha estimates in the four-factor framework in particular in developed markets countries. We find now three significantly positive alpha estimates for the Islamic indices of Belgium (0.66% per month, t = 2.06), Switzerland (0.41% per month, t = 2.25), and the United States (0.16% per month, t = 2.11). The inference drawn from the single markets is in general confirmed in the multi-country regions. Due to the fact the United States accounts for the largest portion of total market equity, the four-factor alpha for the aggregate developed markets region is significantly positive as well (0.18% per month, t = 2.25), while the aggregate emerging markets region shows an alpha estimate that is effectively zero. Though Islamic indices seem to outperform their conventional benchmarks in developed markets, this outperformance is largely attributable to the recent financial crisis and is therefore less likely to be maintained

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

59

Table 4 Average estimates from monthly time-series factor regressions. Market

CAPM a

Four-factor model b



a

b

s

h

w



Panel A: developed markets countries AUS 0.33 (1.45) AUT 0.34 (0.86) BEL 0.64 (2.15) CAN 0.10 (0.48) DEN 0.39 (1.52) FIN 0.27 (0.60) FRA 0.09 (0.69) GER 0.31 (1.88) GRE 0.39 (0.83) HKG 0.05 (0.38) ITA 0.52 (1.96) JPN − 0.01 (− 0.12) NED 0.10 (0.30) NZL 0.17 (0.53) NOR 0.21 (1.13) SIN 0.14 (0.69) SPA 0.37 (1.07) SWE 0.15 (0.66) SWI 0.24 (1.35) UK 0.25 (1.60) USA 0.15 (1.53)

1.05 (28.04) 0.97 (15.84) 0.65 (13.72) 1.11 (29.46) 0.96 (17.39) 0.90 (12.14) 0.91 (35.93) 0.96 (35.16) 0.71 (11.39) 0.86 (29.54) 0.79 (18.65) 0.96 (26.64) 1.00 (19.59) 1.04 (18.18) 0.95 (32.73) 0.91 (22.28) 0.77 (13.51) 0.98 (22.43) 0.86 (22.11) 0.94 (20.09) 0.90 (34.09)

0.91

0.38 (1.75) − 0.08 (− 0.24) 0.66 (2.06) 0.01 (0.03) 0.18 (0.74) 0.28 (0.72) 0.00 (− 0.01) 0.34 (1.94) 0.25 (0.53) − 0.02 (− 0.16) 0.49 (1.71) 0.01 (0.05) 0.20 (0.60) 0.40 (1.15) 0.12 (0.66) 0.18 (0.89) 0.24 (0.67) 0.32 (1.31) 0.41 (2.25) 0.21 (1.38) 0.16 (2.11)

1.05 (29.22) 0.85 (19.78) 0.66 (15.06) 1.12 (36.25) 0.96 (26.05) 0.96 (16.68) 0.97 (43.14) 0.98 (36.24) 0.69 (12.59) 0.87 (34.19) 0.82 (19.10) 0.95 (30.02) 0.98 (18.12) 1.03 (20.62) 0.93 (40.22) 0.94 (30.43) 0.82 (16.22) 0.93 (27.17) 0.86 (22.59) 0.99 (31.86) 0.97 (53.13)

− 0.01 (− 0.22) − 0.47 (− 5.78) − 0.10 (− 0.93) 0.03 (0.54) − 0.18 (− 2.57) 0.09 (0.80) 0.05 (1.11) − 0.04 (− 0.72) − 0.28 (− 2.25) 0.00 (0.00) 0.00 (0.03) − 0.10 (− 2.12) − 0.22 (− 2.39) − 0.03 (− 0.29) − 0.13 (− 2.67) − 0.03 (− 0.60) 0.13 (1.29) − 0.01 (− 0.14) − 0.05 (-0.75) 0.02 (0.39) − 0.03 (− 0.84)

− 0.25 (− 3.13) 0.23 (3.64) − 0.11 (− 1.11) − 0.12 (− 2.21) − 0.27 (− 5.00) 0.39 (4.34) − 0.12 (− 2.51) − 0.08 (− 1.45) − 0.11 (− 0.95) 0.03 (0.86) − 0.13 (− 1.16) 0.01 (0.14) − 0.02 (− 0.17) 0.24 (2.84) 0.01 (0.14) − 0.11 (− 1.79) 0.08 (0.80) 0.01 (0.15) − 0.19 (-3.53) − 0.13 (− 1.80) − 0.16 (− 4.40)

0.12 (2.25) 0.07 (1.50) 0.03 (0.56) 0.08 (2.81) 0.14 (3.75) − 0.30 (− 3.84) 0.10 (4.26) 0.03 (1.08) − 0.16 (− 1.64) 0.04 (1.48) 0.07 (1.24) − 0.07 (− 2.17) − 0.05 (− 0.95) − 0.21 (− 2.69) 0.06 (2.16) 0.07 (1.90) 0.11 (1.76) − 0.11 (− 3.12) 0.02 (0.58) 0.09 (3.12) 0.07 (4.32)

0.92

Panel B: emerging markets countries BRA − 0.11 (− 0.47) CHI 0.09 (0.32) CHN 0.10 (0.73) IND − 0.27 (− 1.16) INO − 0.21 (− 0.70) KOR 0.16 (0.73) MAL 0.09 (0.48) MEX 0.13 (0.47) PER − 0.57 (− 0.82) PHI 0.39 (0.81) SAF − 0.08 (− 0.31) TAI 0.09 (0.47) THA 0.27 (0.70) TUR 0.13 (0.23)

1.07 (49.97) 0.97 (18.60) 0.98 (44.69) 0.95 (41.59) 1.02 (30.50) 0.93 (35.91) 1.03 (20.21) 1.01 (25.50) 0.97 (13.14) 1.19 (17.88) 0.98 (26.76) 0.97 (37.91) 0.96 (20.61) 0.92 (13.24)

− 0.26 (− 1.01) 0.18 (0.63) 0.10 (0.66) − 0.22 (− 0.87) − 0.16 (− 0.52) − 0.01 (− 0.03) − 0.10 (− 0.53) 0.12 (0.39) − 0.75 (− 1.04) 0.38 (0.77) 0.02 (0.11) 0.20 (1.31) 0.44 (1.10) 0.01 (0.01)

1.09 (42.26) 0.94 (21.07) 0.98 (52.06) 0.95 (38.94) 1.01 (30.56) 0.93 (36.90) 1.06 (25.94) 1.01 (19.96) 0.98 (12.33) 1.20 (17.25) 0.89 (28.42) 1.02 (46.25) 0.84 (13.20) 0.92 (18.05)

0.08 (1.23) − 0.11 (− 1.27) 0.00 (− 0.04) − 0.07 (− 1.75) − 0.09 (− 1.57) − 0.06 (− 1.02) − 0.17 (− 2.75) − 0.04 (− 0.48) 0.11 (0.81) 0.10 (1.12) − 0.34 (− 5.46) − 0.07 (− 1.53) − 0.30 (-2.67) 0.13 (0.99)

0.11 (2.19) − 0.06 (− 1.04) 0.00 (0.02) 0.04 (1.11) − 0.10 (− 2.31) 0.08 (1.26) − 0.08 (− 1.11) 0.01 (0.25) 0.13 (1.35) 0.08 (1.10) − 0.26 (− 4.28) − 0.17 (− 6.18) − 0.20 (-1.88) 0.08 (0.58)

− 0.01 (− 0.30) 0.04 (0.60) 0.04 (0.99) 0.01 (0.31) 0.10 (2.07) − 0.02 (− 0.44) 0.10 (2.21) 0.04 (0.80) − 0.02 (− 0.20) 0.06 (0.87) 0.15 (2.46) − 0.02 (− 0.62) − 0.02 (-0.28) − 0.09 (− 0.60)

0.81 0.72 0.92 0.86 0.77 0.95 0.95 0.66 0.94 0.81 0.92 0.80 0.79 0.95 0.91 0.74 0.90 0.84 0.91 0.95

0.96 0.85 0.97 0.94 0.91 0.93 0.88 0.85 0.64 0.75 0.89 0.94 0.79 0.79

0.87 0.72 0.93 0.90 0.81 0.96 0.95 0.67 0.94 0.81 0.93 0.80 0.81 0.96 0.91 0.74 0.91 0.86 0.92 0.96

0.96 0.85 0.97 0.94 0.92 0.93 0.89 0.84 0.63 0.75 0.93 0.96 0.80 0.78

(continued on next page)

60

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

Table 4 (continued) Market

CAPM a

Panel C: aggregate markets DM 0.16 (1.83) EM − 0.02 (− 0.16)

Four-factor model b



a

b

s

h

w



0.93 (38.87) 1.01 (66.73)

0.96

0.18 (2.25) 0.01 (0.09)

0.94 (49.36) 0.99 (67.59)

0.04 (1.29) − 0.18 (− 3.93)

− 0.19 (− 4.21) − 0.02 (− 0.46)

0.10 (5.55) − 0.02 (− 1.05)

0.97

0.97

0.98

Notes: This table reports time-series factor regression estimates and the corresponding t-statistics for the Islamic indices over the sample period using the CAPM and the four-factor model. a is the average alpha estimate. b, s, h, and w are the factor loadings (betas) related to the market, size, value, and momentum factors. The t-statistics in parentheses are based on robust standard errors using the White (1980) methodology. The regression R2 values are adjusted for degrees of freedom.

over time. We will provide evidence for this argument in the robustness section with subperiod regressions. The investigation of the factor loadings related to size, value, and momentum reveals the investment behavior of Islamic indices. Across all single markets, we find nine significantly negative SMB slopes, indicating that Islamic indices are more likely to invest in largecapitalization stocks. This is particularly true for the smaller developed markets like Austria, Denmark, and Greece, but also Japan as one of the main equity markets in the sample, and the majority of emerging markets countries. The average slope on size is significantly negative in the aggregate emerging markets region with a value of −0.18 and a t-statistic of −3.93. The coefficient estimates on HML tell us that Islamic indices invest primarily in growth stocks (with low book-to-market ratios). Of the 14 factor loadings that are more than two standard errors from zero, ten are significantly negative ranging from − 0.27 to − 0.10 in magnitude. The negative loadings on the value factor, i.e., the explicit non-exploitation of the positive value premium (the observable return pattern that high book-to-market stocks earn on average higher returns) acts as a burden in the performance measurement and therefore strengthens the alpha estimates in the four-factor framework. Regarding the Islamic index returns’ momentum characteristic reveals that ten out of the 14 significant WML slopes are positive. Thus, Islamic indices show the tendency to allocate capital to stocks that have performed well over the last twelve months relative to other stocks in the considered market. However, the growth- and momentum-oriented investment behavior tends to be more pervasive in developed markets than in emerging markets. The slopes on HML and WML for the aggregate emerging markets region are close to zero, while they are economically substantial and statistically significant for the aggregate developed markets region with values of −0.19 (t = − 4.21) and 0.10 (t = 5.55), respectively. Thus, the assumption that the return behavior of developed markets is distinct from emerging markets (e.g., Harvey (1995), Eun and Lee (2010)) seems to be also valid for Islamic indices. The positive and significant loadings on the momentum factor are rather intriguing with regard to the fact that equity indices are passive investment instruments as opposed to actively managed mutual funds which usually follow a momentum strategy (Hendricks, Patel, and Zeckhauser (1993), Grinblatt, Titman, and Wermers (1995), Carhart (1997)). How can this return feature of Islamic indices be explained? To address this question, we focus on the industry allocation of Islamic indices. To obtain a first (rough) assessment of which industries exhibit momentum characteristics in their returns, we employ the ten U.S. industry portfolios from Kenneth French's data library. 6 In particular, we regress the monthly excess returns of the industry portfolios on the explanatory factors of the four-factor framework over the time

6 Firms are assigned to the following ten industries: non-durables, durables, manufacturing, energy, high tech, telecommunication, shops, health, utilities, and others. The detailed definition according to the Standard Industrial Classification (SIC) codes is accessible through the data library.

period 2002 to 2011. We identify that the average returns of the energy sector load heavily on the momentum factor with an average coefficient estimate of 0.36 and a t-statistic of 3.52. 7 Due to the fact that firms from the energy industry account for more than one fourth of the index wealth in the U.S. Islamic index, this gives a reasonable explanation for the positive and significant WML slope. We are well aware that not all stocks in the energy sector may exhibit momentum characteristics in the same strength or direction as in the U.S., but as the strong sector allocation towards energy firms is likewise present in developed markets (on average about 23% of index wealth) and emerging markets (on average about 31%), it provides at least a rationale for the respective markets exhibiting the return feature of positive momentum. 5. Robustness In this section, we check the robustness of our inference that Islamic indices do not sacrifice portfolio performance in comparison to conventional benchmarks around the world. We address two important issues. First, we examine the impact of the recent financial crisis on the performance of Islamic indices using split-sample timeseries factor regressions. Second, we further test whether our results hold when local currencies and country-specific risk-free rates are applied. 5.1. Impact of the recent financial crisis Table 5 presents average subperiod alpha estimates and the corresponding t-statistics for the Islamic indices from split-sample timeseries factor regressions using the CAPM and the four-factor model. The first subperiod is from June 2002 to December 2007 (67 months) and the second subperiod is from January 2008 to June 2011 (42 months). The split date of the sample period is chosen to reflect to impact of the recent financial crisis and is in line with the business cycle dates of the National Bureau of Economic Research (NBER). Regarding the single markets, the number of significant alpha estimates has decreased from the earlier to the latter subperiod. While the Islamic indices of Belgium and Germany (in the CAPM and four-factor framework), and Australia and South Africa (in the four-factor framework) exhibit economically substantial and significantly positive intercepts in the period before the outbreak of the financial crisis, all alpha estimates are rendered insignificant in the crisis-influenced time period 2008 to 2011. This is likewise true for the significantly negative alpha of the Peruvian Islamic index from the first to the second subperiod. A notable exception is the United States, where we find an increased and significantly positive four-factor alpha of 0.31% per month (t = 2.20) in the time period 2008 to 2011. As the U.S. capital market was heavily affected by the recent financial crisis, it is less surprising that the Islamic index variant, as an equity portfolio 7 The second largest significantly positive WML slope comes from the utilities sector (0.17, t = 3.22).

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62 Table 5 Average subperiod alpha estimates from monthly split-sample time-series factor regressions. Market

CAPM

Four-Factor Model

2002–2007

2008–2011

2002–2007

2008–2011

a

a

a

t(a)

a

t(a)

Panel A: developed markets countries AUS 0.33 1.41 0.28 0.63 AUT 0.61 1.39 − 0.61 − 0.78 BEL 0.76 2.31 0.67 1.19 CAN 0.03 0.16 − 0.06 − 0.16 DEN 0.13 0.42 0.41 1.08 FIN 1.03 1.85 − 0.10 − 0.21 FRA 0.13 0.82 0.11 0.45 GER 0.45 2.60 0.19 0.59 GRE 0.15 0.40 0.53 0.48 HKG 0.04 0.31 0.04 0.12 ITA 0.46 1.49 0.44 0.91 JPN − 0.12 − 0.80 0.23 0.98 NED − 0.18 − 0.37 0.23 0.74 NZL 0.33 0.87 0.21 0.35 NOR 0.31 1.25 0.00 − 0.01 SIN 0.32 1.25 − 0.03 − 0.11 SPA 0.52 1.43 0.46 0.67 SWE − 0.05 − 0.15 0.28 0.69 SWI 0.04 0.18 0.55 1.87 UK 0.22 1.39 0.32 1.00 USA 0.06 0.69 0.22 1.20

t(a)

0.55 0.34 0.79 − 0.19 0.44 1.01 − 0.03 0.49 − 0.08 0.00 0.53 − 0.08 − 0.07 0.11 0.32 0.38 0.40 0.14 0.35 0.17 0.07

2.20 0.86 2.27 − 0.80 1.34 1.99 − 0.18 2.72 − 0.21 0.04 1.50 − 0.47 − 0.13 0.28 1.29 1.37 0.98 0.50 1.47 0.91 0.72

0.46 − 0.98 0.70 0.03 0.04 0.06 0.12 − 0.02 − 0.11 0.03 0.34 0.16 0.16 0.72 − 0.12 0.14 0.55 0.22 0.42 0.32 0.31

1.09 − 1.52 1.06 0.09 0.09 0.14 0.53 − 0.05 − 0.09 0.11 0.63 0.65 0.52 1.23 − 0.38 0.42 0.78 0.51 1.51 1.05 2.20

Panel B: emerging BRA − 0.10 CHI 0.45 CHN 0.17 IND − 0.36 INO 0.27 KOR 0.26 MAL 0.32 MEX − 0.25 PER − 2.16 PHI 0.06 SAF 0.09 TAI 0.01 THA 0.79 TUR 0.34

− 0.87 − 0.93 − 0.18 − 0.84 − 1.80 0.38 − 1.07 1.19 0.61 0.89 − 0.76 0.77 − 0.93 − 0.41

− 0.33 0.63 0.21 − 0.39 0.37 0.09 0.00 − 0.10 − 2.54 − 0.02 0.60 0.10 0.64 0.47

− 0.91 1.75 1.16 − 1.14 1.11 0.25 0.01 − 0.25 − 3.37 − 0.03 2.24 0.53 1.06 0.57

− 0.27 − 0.40 0.13 − 0.24 − 0.90 0.00 − 0.35 0.69 0.69 0.56 − 0.55 0.28 − 0.06 − 1.07

− 0.75 − 0.86 0.48 − 0.76 − 1.51 − 0.02 − 1.10 1.40 0.52 0.61 − 1.83 1.22 − 0.16 − 0.91

1.32 − 0.69

0.12 0.13

1.17 0.76

0.35 − 0.11

2.67 − 0.63

markets countries − 0.29 − 0.30 1.41 − 0.40 0.97 − 0.04 − 1.10 − 0.28 0.74 − 1.00 0.80 0.10 1.29 − 0.32 − 0.68 0.59 − 2.87 0.79 0.11 0.78 0.25 − 0.30 0.03 0.18 1.32 − 0.36 0.41 − 0.41

Panel C: aggregate markets DM 0.08 0.96 EM 0.02 0.10

0.24 − 0.12

t(a)

Notes: This table reports average subperiod alpha estimates and the corresponding t-statistics for the Islamic indices from split-sample time-series factor regressions using the CAPM and the four-factor model. The first subperiod is from June 2002 to December 2007 (67 months) and the second subperiod is from January 2008 to June 2011 (42 months). The t-statistics are based on robust standard errors using the White (1980) methodology.

deliberately excluding bank and financial services stocks, provides an enhanced performance in such a case. 8 However, the outperformance is only significant after additionally controlling for size, value, and momentum. The U.S. CAPM alpha is statistically not distinguishable from zero between 2008 and 2011. As we consider the extensive downturn of financials in the recent financial crisis as an exceptional event on the capital market, we do not argue that this outperformance will continue over time. The same argumentation applies to the aggregate developed markets region in which the United States market is a heavy weight.

8 For instance, the bankruptcy of Lehman Brothers and the nearly collapses of Bear Stearns and Merill Lynch. The S&P 500 Banks index lost almost three fourth (− 73.8%) of its value from January 2008 to the bottom in February 2009.

61

5.2. Influence of local currencies and country-specific risk-free rates To facilitate comparisons between different countries, we denominate all returns in U.S. dollars and calculate excess returns using the U.S. Treasury bill rate. However, some of the (emerging) countries in the sample have experienced high and volatile inflations. Assuming that the risk-free rate is a combination of the real rate of interest and the inflation premium, country-specific risk-free rates may be much more volatile. To examine whether the use of the U.S. Treasury bill rate as the standard risk-free rate is influential in the results observed in this paper, we re-run regression (3) using index returns denominated in local currencies and country-specific risk-free rates. 9 Table 6 shows the average one-month risk-free rate for each nonU.S. country and summarizes the CAPM regression estimates for the Islamic indices over the sample period. The local risk-free rate is proxied by the national one-month interbank rate, e.g., like the EURIBOR which is commonly used for this purpose in Eurozone countries. While the level of the average risk-free rate of 0.22% per month for non-U.S. developed markets countries is largely comparable to the U.S. Treasury bill rate of 0.16% per month, the risk-free rate of emerging markets countries is considerably higher in magnitude and much more volatile across countries with an average of 0.52% per month and a range of 1.64% per month between the low and high end of rates. However, the results in local currencies using country-specific risk-free rates are largely similar to the U.S. dollar denominated evidence. All alpha estimates are within two standard errors from zero suggesting no significant performance differences between Islamic indices and conventional benchmarks.

6. Conclusions The literature on the performance of Islamic indices is scarce. While prior research examines mostly mutual funds, the few papers focusing on Sharia-compliant indices study mainly one single equity market and use a relatively straightforward measurement framework. This paper advances the research in this field by investigating the performance and style of Islamic indices around the world in one comprehensive study employing a large international data set combining 35 developed and emerging markets and a contemporary evaluation framework based on bootstrap simulations and multifactor models. We examine whether Islamic indices perform differently to conventional benchmarks using robust differences-in-Sharpe ratio tests and time-series factor regressions controlling for market risk (CAPM) as well as size, value, and momentum characteristics in the four-factor framework. In sum, we do not find compelling evidence of performance differences between Islamic indices and conventional benchmarks. Though we find a significant four-factor alpha for the U.S. Islamic index over the sample period, this outcome is largely attributable to the recent financial crisis and the related decline of financial stocks which are excluded in Sharia-compliant indices. Given the largely similar performance of the two index variants in the United States in the pre-financial crisis period and the assumption that the extensive downturn of bank and financial services stocks is special to the crisis, we do not argue that this outperformance will continue over time. Regarding the investment behavior of Sharia-compliant indices, we find that in developed markets, Islamic indices allocate capital primarily to growth stocks and firms with positive momentum, while they tend to invest particularly in large-capitalization stocks in emerging markets. Our findings contribute to the ongoing debate on whether screened portfolios (motivated through social responsibility or 9

Hence, the aggregate (multi-country) regions are excluded from the analysis.

62

C. Walkshäusl, S. Lobe / Review of Financial Economics 21 (2012) 53–62

Table 6 Average non-U.S. CAPM regression estimates using local currencies and countryspecific risk-free rates. Rf

a

b

t(a)

t(b)



Panel A: developed market countries AUS Australian Dollar AUT Euro BEL Euro CAN Canadian Dollar DEN Danish Krone FIN Euro FRA Euro GER Euro GRE Euro HKG Hong Kong Dollar ITA Euro JPN Japanese Yen NED Euro NZL New Zealand Dollar NOR Norwegian Krone SIN Singaporean Dollar SPA Euro SWE Swedish Krona SWI Swiss Franc UK United Kingdom Pound

Market

Currency

0.44 0.20 0.20 0.22 0.23 0.20 0.20 0.20 0.20 0.15 0.20 0.02 0.20 0.49 0.29 0.10 0.20 0.21 0.07 0.30

0.36 0.33 0.45 0.10 0.36 0.22 0.05 0.30 0.24 0.05 0.42 − 0.01 0.12 0.19 0.20 0.12 0.30 0.13 0.17 0.24

1.04 0.98 0.52 1.19 0.98 0.87 0.86 0.95 0.65 0.86 0.74 1.00 1.06 0.97 0.94 0.91 0.69 1.01 0.82 0.93

1.62 0.83 1.76 0.53 1.43 0.51 0.41 1.83 0.52 0.38 1.63 − 0.11 0.37 0.59 1.05 0.61 0.90 0.58 1.01 1.55

13.89 13.83 11.59 22.18 14.68 10.23 28.13 32.99 9.56 29.69 12.81 31.51 15.74 11.40 24.69 19.50 11.54 17.94 16.66 17.84

0.77 0.75 0.62 0.87 0.83 0.72 0.92 0.93 0.57 0.94 0.70 0.94 0.78 0.59 0.92 0.88 0.62 0.87 0.80 0.87

Panel B: emerging market countries BRA Brazilian Real CHI Chilean Peso CHN Chinese Yuan Renminbi IND Indian Rupee INO Indonesian Rupiah KOR South Korean Won MAL Malaysian Ringgit MEX Mexican Peso PER Peruvian Nuevo Sol PHI Philippine Peso SAF South African Rand TAI Taiwanese Dollar THA Thai Baht TUR Turkish Lira

1.23 0.02 0.24 0.53 0.76 0.31 0.20 0.56 0.31 0.42 0.76 0.10 0.22 1.66

− 0.03 0.04 0.10 − 0.28 − 0.24 0.16 0.10 0.16 − 0.59 0.81 − 0.12 0.08 0.27 0.04

1.07 1.00 0.98 0.94 1.04 0.94 1.02 0.99 0.97 1.13 1.08 0.97 0.96 0.87

− 0.14 0.14 0.73 − 1.21 − 0.78 0.83 0.52 0.56 − 0.84 1.44 − 0.47 0.46 0.68 0.07

32.93 20.33 44.64 34.39 25.58 26.48 16.91 19.35 12.72 13.44 20.01 32.13 18.32 10.93

0.90 0.76 0.97 0.92 0.87 0.93 0.84 0.77 0.62 0.68 0.82 0.92 0.75 0.67

Notes: This table presents CAPM regression estimates for the non-U.S. Islamic indices denominated in local currencies using country-specific risk-free rates. Rf is the average monthly country-specific risk-free rate proxied by the national one-month interbank rate in each non-U.S. country. a is the average alpha estimate and b is the factor loading (beta) related to the market factor. The t-statistics are based on robust standard errors using the White (1980) methodology. The regression R2 values are adjusted for degrees of freedom.

religious objectives) are determined to underperform conventional (unscreened) portfolios (e.g., Rudd (1981), Grossman and Sharpe (1986), Diltz (1995), Renneboog, Ter Horst, and Zhang (2007) and the references within). Our results provide supportive evidence that religious stock screens in the context of index-based Sharia-compliant investments do not reduce financial performance in comparison to conventional index investments around the world. This insight is particularly helpful for Sharia-compliant investors as they can pursue passive stock investments in accordance with their religious beliefs without sacrificing financial performance. References Albaity, Mohamed, & Ahmad, Rubi (2008). Performance of Syariah and composite indices: Evidence from Bursa Malaysia. Asian Academy of Management Journal of Accounting and Finance, 4, 23–43. Bali, Turan G., Demirtas, K. Ozgur, & Levy, Haim (2008). Nonlinear mean reversion in stock prices. Journal of Banking & Finance, 32, 767–782. Bali, Turan G., Demirtas, K. Ozgur, & Levy, Haim (2009). Is there an intertemporal relation between downside risk and expected returns? Journal of Financial and Quantitative Analysis, 44, 883–909.

Banz, Rolf W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics, 9, 3–18. Carhart, Mark M. (1997). On persistence in mutual fund performance. Journal of Finance, 52, 57–82. Chui, Andy C. W., Titman, Sheridan, & Wei, K. C. John (2010). Individualism and momentum around the world. Journal of Finance, 65, 361–392. Diltz, J. David (1995). The private cost of socially responsible investing. Applied Financial Economics, 5, 69–77. Elfakhani, Said, & Hassan, M. Kabir (2005). Performance of Islamic mutual funds, American University of Beirut, Working Paper. Eun, Cheol S., & Lee, Jinsoo (2010). Mean-variance convergence around the world. Journal of Banking & Finance, 34, 856–870. Fama, Eugene F., & French, Kenneth R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33, 3–56. Fama, Eugene F., & French, Kenneth R. (1998). Value versus growth: The international evidence. Journal of Finance, 53, 1975–1999. Fama, Eugene F., & French, Kenneth R. (2011). Size, value, and momentum in international stock returns, University of Chicago, Working Paper. Forte, Gianfranco, & Miglietta, Federica (2007). Islamic mutual funds as faith-based funds in a socially responsible context, Bocconi University, Working Paper. Griffin, John M. (2002). Are the Fama and French factors global or country specific? Review of Financial Studies, 15, 783–803. Grinblatt, Mark, Titman, Sheridan, & Wermers, Russ (1995). Momentum investment strategies, portfolio performance, and herding: A study of mutual fund behavior. The American Economic Review, 85, 1088–1105. Grossman, Blake R., & Sharpe, William F. (1986). Financial implications of South Africa divestment. Financial Analyst Journal, 42, 15–29. Harvey, Campbell R. (1995). Predictable risk and returns in emerging markets. Review of Financial Studies, 8, 773–816. Hendricks, Darryll, Patel, Jayendu, & Zeckhauser, Richard (1993). Hot hands in mutual funds: Short-run persistence of relative performance, 1974–1988. Journal of Finance, 48, 93–130. Hoepner, Andreas G. F., Rammal, Hussain G., & Rezec, Michael (2011). Islamic mutual funds' financial performance and international investment style: evidence from 20 countries. European Journal of Finance, 17, 829–850. Hussein, Khaled, & Omran, Mohammed (2005). Ethical investment revisited: Evidence from Dow Jones Islamic Indexes. Journal of Investing, 14, 105–126. Ince, Ozgur S., & Porter, R. Burt (2006). Individual equity return data from Thomson datastream: Handle with care! Journal of Financial Research, 29, 463–479. Jegadeesh, Narasimhan (1990). Evidence of predictable behavior of security returns. Journal of Finance, 45, 881–898. Jegadeesh, Narasimhan, & Titman, Sheridan (1993). Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance, 48, 65–91. Jobson, J. D., & Korkie, Bob M. (1981). Performance hypothesis testing with the Sharpe and Treynor measures. Journal of Finance, 36, 889–908. Ledoit, Oliver, & Wolf, Michael (2008). Robust performance hypothesis testing with the Sharpe ratio. Journal of Empirical Finance, 15, 850–859. Levy, Haim (1992). Stochastic dominance and expected utility: Survey and analysis. Management Science, 38, 555–593. Lintner, John (1965). The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. The Review of Economics and Statistics, 47, 13–37. Loh, Wei-Yin (1987). Calibrating confidence coefficients. Journal of the American Statistical Association, 82, 155–162. Memmel, Christoph (2003). Performance hypothesis testing with the Sharpe ratio. Finance Letters, 1, 21–23. Naughton, Shahnaz, & Naughton, Tony (2000). Religion, ethics and stock trading: The case of an Islamic equities market. Journal of Business Ethics, 23, 145–159. Ogryczak, Wlodzimierz, & Ruszczynski, Andrzej (1999). From stochastic dominance to mean-risk models: Semideviations as risk measures. European Journal of Operational Research, 116, 33–50. Politis, Dimitris N., & Romano, Joseph P. (1992). A circular block-resampling procedure for stationary data. In Raoul LePage, & Lynne Billard (Eds.), Exploring the limits of bootstrap. New York: John Wiley & Sons. Renneboog, Luc, Ter Horst, Jenke, & Zhang, Chendi (2007). Socially responsible investments: Methodology, risk exposure and performance, Tilburg University, Working Paper. Rosenberg, Barr, Reid, Kenneth, & Lanstein, Ronald (1985). Persuasive evidence of market inefficiency. Journal of Portfolio Management, 11, 9–17. Rouwenhorst, K. Geert (1998). International momentum strategies. Journal of Finance, 53, 267–284. Rudd, Andrew (1981). Social responsibility and portfolio performance. California Management Review, 23, 55–61. Sharpe, William F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk. Journal of Finance, 19, 425–442. Sharpe, William F. (1992). Asset allocation: Management style and performance measurement. Journal of Portfolio Management, 18, 7–19. White, Halbert (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48, 817–838.